database, in a fully overlapped heterogeneous network setting, changes of access mode are very frequent .... 4.4.5 Call session setup between GPRS & PSTN .
Integration of Heterogeneous Wireless Access Networks with IP Multimedia Subsystem by Peyman TalebiFard B.Sc., Carleton University, 2006 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Electrical & Computer Engineering)
THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) November, 2008 c Peyman TalebiFard 2008
Abstract Next generation heterogeneous wireless networks are expected to interwork with Internet Protocol (IP)-based infrastructures. Conventional network services operate like silos in that a specific set of services are offered over a specific type of access network. As access networks evolve to provide IP-based packet access, it becomes attractive to break these service silos by offering a converged set of IP-based services to users who may access these services using a number of alternative access networks. This trend has started with third generation cellular mobile networks, which have standardized on the use of the IP Multimedia Subsystem (IMS) to manage user access to a wide variety of multimedia services over the mobile Internet, while facilitating interworking of heterogeneous wireless and landline access networks. The future users of communication systems will subscribe to both IP-based and Circuit Switched (CS) based services and in the foreseeable future a single database that handles user profiles across all domains will be required. Home Subscriber Server (HSS) as an evolved version of Home Location Register (HLR) is one of the key components of IMS. In deploying HSS as a central repository database, in a fully overlapped heterogeneous network setting, changes of access mode are very frequent and conveying this information to HSS imposes excessive signaling load and delay. In our proposed scheme we introduce an Interface Agent (IA) for each location area that caches the location and information about the access mode through which a user can be reached. This method results in significant amount of savings in signaling cost and better delay performance. The existing call delivery approaches in cellular networks may not be well suited for future communication systems because they suffer from unnecessary usage of network resources for call attempts that may fail which adds to excessive signaling delays and queuing costs. Reducing the number of queries and retrievals from the database will have a significant impact on the network performance. We present a new scheme based on Reverse Virtual Call setup (RVC) as a solution to the call delivery problem in heterogeneous wireless networks and evaluate the performance of this framework.
ii
Table of Contents Abstract
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ii
Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iii
List of Tables
vi
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix
Dedication
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
x
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1.1
1.0.1
Business Aspects of Integration
. . . . . . . . . . . . . . . . . . . . . . . . .
1
1.0.2
Driving Forces of Convergence . . . . . . . . . . . . . . . . . . . . . . . . . .
2
Conversational Communications over Converged Heterogeneous Networks . . . . . .
4
1.1.1
Voice Call
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
1.1.2
Video Call
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
1.1.3
Multimedia Call . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
1.1.4
Push-to-X Call . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
1.1.5
Location-based Service
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
1.1.6
Message Based Communications . . . . . . . . . . . . . . . . . . . . . . . . .
7
1.1.7
Multimdia Messaging Service . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
1.1.8
Instant Messaging Service
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
1.1.9
Unified Messaging Service
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
1.1.10 Content-on-Demand Communications . . . . . . . . . . . . . . . . . . . . . .
8
1.1.11 Browsing Service
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
1.1.12 Content Download
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
1.1.13 Content Streaming
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
1.1.15 Mobile Broadcast and IPTV . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
1.1.16 Peer-To-Peer Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
1.1.17 Session Persistence
9
1.1.14 Content Push
1.2
Contributions
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
iii
Table of Contents 2 Background 2.1
2.2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
IP Multimedia Subsystem (IMS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.1.1
Architectural Elements
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.2
Home Subscriber Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.3
Service Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Integrating Heterogeneous Networks using IMS . . . . . . . . . . . . . . . . . . . . . 21 2.2.1
Architecture
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.2
QoS Provisioning for Heterogeneous Wireless Access Interworking . . . . . . 21
2.2.3
Mobility Support
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 Novel Database Architecture and Signaling Scheme for IP-Based Heterogeneous Wireless Access Interworking 3.1
3.2
3.3
3.4
Introduction
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.1.1
Location Management in 3G Cellular Systems
3.1.2
User Registration, Location Update and Access Mode changes . . . . . . . . 37
3.1.3
IMS release 7 and 8 solution for multi-mode support
Proposed Scheme
. . . . . . . . . . . . . . . . . 35 . . . . . . . . . . . . . 39
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.1
User Registration under the proposed method
. . . . . . . . . . . . . . . . . 41
3.2.2
Change of Access Mode under the Proposed Method
3.2.3
Implications on Access Mode changes for WLAN Re-Authentication . . . . . 45
. . . . . . . . . . . . . 45
Performance Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.3.1
Analysis of Signaling Cost
3.3.2
Signaling Cost of Registration & Location Update . . . . . . . . . . . . . . . 47
3.3.3
Signaling Cost of Access Mode Update
3.3.4
Analysis of Delay
3.3.5
Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Conclusion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 . . . . . . . . . . . . . . . . . . . . . 48
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4 Call Delivery over IP-Based Heterogeneous Wireless Access Networks . . . . . 57 4.1
Introduction
4.2
Call Delivery and Mobility Management
4.3 4.4
4.5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2.1
Call Delivery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2.2
Reverse Virtual Call Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2.3
Call Delivery Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
System Architecture & Problem Formulation
. . . . . . . . . . . . . . . . . . . . . 65
Proposed Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.4.1
Registration
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.4.2
Initiating a Call
4.4.3
Terminating a call . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.4.4
Call session setup between WLAN & PSTN
4.4.5
Call session setup between GPRS & PSTN . . . . . . . . . . . . . . . . . . . 71
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 . . . . . . . . . . . . . . . . . . 70
Performance Analysis and Evaluations . . . . . . . . . . . . . . . . . . . . . . . . . . 71
iv
Table of Contents 4.5.1
Cost Analysis
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.5.2
Delay Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.6
Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.7
Implementation Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.8
4.7.1
Best Network Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.7.2
QoS Aware Multimedia Call Setup
Conclusion
5 Conclusion 5.1
. . . . . . . . . . . . . . . . . . . . . . . 79
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Design Issues, Technical Challenges, and Future Research Directions . . . . . . . . . 83 5.1.1
IMS Session Setup in a Heterogeneous Mobile Environment . . . . . . . . . . 83
5.1.2
Customer Profile and Identity Management . . . . . . . . . . . . . . . . . . . 83
5.1.3
Resource Management for QoS . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.1.4
Seamless Mobility Management and Vertical Handover Support
5.1.5
Location Management – Tracking of User Locations . . . . . . . . . . . . . . 84
5.1.6
Multicast/Broadcast Services Across Multiple Access Technologies . . . . . . 85
5.1.7
Issues with Call Admission Control
5.1.8
Charging Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.1.9
Migration from HLR to HSS . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
. . . . . . . 84
. . . . . . . . . . . . . . . . . . . . . . . 85
5.1.10 Design Considerations for Mobility Support and QoS
. . . . . . . . . . . . . 86
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Appendices A Statement of Co-Authorship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
v
List of Tables 3.1
Role of core elements of IMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2
Parameters for cost of signaling vs. probability of mobility . . . . . . . . . . . . . . . 52
3.3
Cost savings as a result of change in ratio of Intra-LA to Inter-LA mobility . . . . . 52
3.4
Latency Parameters for transmission and query . . . . . . . . . . . . . . . . . . . . . 53
4.1
Signaling cost values (C L & C Q ) for transmission and query . . . . . . . . . . . . . 75
4.2
Latency Parameters (DL & DQ ) for transmission and query . . . . . . . . . . . . . 76
vi
List of Figures 1.1
Development of telecom networks from past to future
. . . . . . . . . . . . . . . . .
5
2.1
Overview of core elements of the IMS architecture specified by 3GPP . . . . . . . . . 13
2.2
Role of Application Servers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3
Functionality of HSS node at logical level across multiple domains . . . . . . . . . . 28
2.4
Details of Service Layer in SOA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5
Policy-based Access Control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.6
Charging for Converged Services. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.7
Layout of logical interfaces for integration with IP core networks . . . . . . . . . . . 32
2.8
Network architecture for QoS conceptual models . . . . . . . . . . . . . . . . . . . . 33
3.1
Interworking architecture that shows the physical distribution of HSS . . . . . . . . 36
3.2
Signaling diagram for location update and registration of a MT when it moves to a new LA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3
Signaling diagram for change of access mode when a MT is roaming within a LA . . 39
3.4
User movement and registration procedure . . . . . . . . . . . . . . . . . . . . . . . . 42
3.5
Communication of policy decisions for access mode grant permission . . . . . . . . . 44
3.6
Signaling diagram for change of access mode under the proposed method . . . . . . . 45
3.7
Signaling diagram for location update and access mode update in HSPA and WLAN 46
3.8
Sinaling diagram for a typical base scenario . . . . . . . . . . . . . . . . . . . . . . . 49
3.9
Location areas and possible configuration of coverage within a LA . . . . . . . . . . 53
3.10 Average latency with changes in Pr[Inter-LA-Mobility] . . . . . . . . . . . . . . . . . 54 3.11 Signaling cost vs. probability of mobility . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.12 Signaling cost saving vs. probability of LA change . . . . . . . . . . . . . . . . . . . 56 4.1
Network configuration for the base scenario . . . . . . . . . . . . . . . . . . . . . . . 58
4.2
User registration in a typical base scenario . . . . . . . . . . . . . . . . . . . . . . . . 60
4.3
Originating a call/session in a typical base scenario . . . . . . . . . . . . . . . . . . . 61
4.4
Terminating a call/session in a typical base scenario . . . . . . . . . . . . . . . . . . 62
4.5
Signaling diagram for delivering a call in the base scenario for the example of GPRSPSTN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.6
Interworking model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.7
User registration in the proposed model . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.8
Call initiation in the originating network for the proposed model . . . . . . . . . . . 69
4.9
Call termination in the called party terminating network for the proposed model . . 70
vii
List of Figures 4.10 Signaling flow diagram for call delivery procedure for the example of PSTN to WLAN 71 4.11 Cost of signaling for base situation and proposed method versus the number of originating calls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 4.12 Savings in cost of signaling for the proposed scheme as the percentage of failed calls increases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 4.13 Comparison of delay for successful and unsuccessful call for base situation and proposed method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 4.14 Average delay performance for proposed scheme when varying session setup failure rate θ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 4.15 Message flow for multimedia call setup . . . . . . . . . . . . . . . . . . . . . . . . . . 80
viii
Acknowledgements I would like to give my special thanks to my supervisor, Professor Victor C.M. Leung, for his kindly support and guidance. Progress of this research work would have not been possible without his advice and encouragement. Thanks are also to Mr. Terrence Wong from TELUS for his cooperation. I would also like to appreciate TELUS and the Natural Sciences and Engineering Research Council of Canada for providing grants to support this work.
ix
Dedication I dedicate this work to my beloved parents, grandparents and family for their continuous prayers and support. I furthermore dedicate this work to the martyrs of the Baha’i faith in Iran, to the imprisoned Baha’i friends and to the Baha’i youth of Iran who were denied the right of education solely on the basis of their beliefs in the Baha’i faith.
x
Chapter 1
Introduction 1 The Internet is composed of different domains operated by Internet service providers (ISPs) with different capabilities, policies and access networks. Next generation networks (NGNs) will employ standards and architectures that are based on the Internet Protocol (IP) suite. The IP platform can integrate diverse access networks in a common scalable framework, and through IP Multimedia Subsystem (IMS), extend a wide range of multimedia services to subscribers over heterogeneous wireless and wireline access networks. Integration of heterogeneous networks with IMS may be considered under the following two aspects: 1. From the perspective of service providers managing the networks according to different transmission media such as cable, satellite, radio that may also implement different protocols such as Asynchronous Transfer Mode (ATM), IP or Multi-Protocol Label Switching (MPLS). 2. From the point of view of users who can access different services with different features, availability and prices [1].
1.0.1
Business Aspects of Integration
Enterprises are moving from centralized data architectures towards content-aware networking, which is a more flexible approach based on Service Oriented Architectures (SOA). In traditional IP networking, packets are routed based on destination IP addresses. Drawbacks of traditional IP networking are complexity in the application layer and increase in the cost of information technology capital and operating expenses. In content-aware networking, messages are routed based on content and context. It embeds business rules in high speed and low latency networks while decreasing the IT capital cost and operational expenses [2]. In the context business aspects in content aware networking, reusing the existing IP infrastructures for content-aware networks brings the following advantages to the enterprises: • It allows distributed or SOA applications to use the shared infrastructures on a demand basis to filter route and transform information faster than traditional software middleware. Enterprises can invest on this shared infrastructure and reduce their capital and operational costs significantly. • As executions are done in hardware, it virtually eliminates performance degradations and ensures that data latency is low and predictable in cases where content routing and filtering is required. 1 A version of this chapter has been published. P. TalebiFard, T. Wong, and V.C.M Leung, ”Integration of heterogeneous wireless access networks with IP-based core networks”, book chapter in Heterogeneous Wireless Access Networks, Ekram Hossain ed., Springer, 2008.
1
Chapter 1. Introduction • Application servers no longer need to focus on complex business rule executions, event processing and analysis. Content-aware networks allow service providers to filter, route and transform information to enterprise customers on their behalf as well as to their partners. It therefore raises the position of service providers from providers of bandwidth commodity to strategic suppliers of services [2]. IMS provides an IP-based service centric creation and control framework that leverages common resources for services. It therefore creates the ability for rapid creation of multimedia and enhanced services. Furthermore, IMS application servers are capable of interacting with Web 2.0 applications. Web 2.0 applications are social networking applications that emphasize on services and are based on user participation. One of the significant requirements of enhanced services is application billing and charging functions. IMS features can also be leveraged to offer Fixed Mobile Convergence (FMC) solutions in conjunction with wireless partners and cable operators. Cable operators would then be able to offer enhanced services that provide ease of use and convenience. For instance, users can receive message waiting indications for messaging applications on their TV. Enhanced services offered by IMS can further be used as a distribution channel [3]. For example in messaging applications and services, mailboxes can be utilized so that users could receive targeted offers of new services and the option to purchase the services. One of the attractive drivers for next generation IMS enhanced services is that operators are able to offer services to small and medium sized businesses via the software-as-a-service (SaaS) business model.
1.0.2
Driving Forces of Convergence
Transitioning to all-IP NGNs builds the foundation of convergence at three levels, namely application, network and service. Application Convergence: It can provide numerous opportunities for carriers because mobile devices are increasingly capable of supporting multiple functions such as voice, video, email, and web browsing. Network Convergence: Referes to integration of several application-specific networks into a single IP-based multi-service infrastructure. This leads to significant reduction of infrastructure costs, capital and operating expenses. It is therefore important that providers develop new services to substitute for legacy applications over the new infrastructure because running old services on new IP networks causes more problems and higher operating costs. Service Convergence: Technical capabilities of IP and IMS are also enabling service convergence. Service convergence delivers more intelligent application level and subscriber level service control. In other words, service providers will be able to bill, operate and manage their services over a wide range of access networks. Multi-modal mobile devices that provide several alternate means to access the Internet are becoming widely available. However, mobile users are more interested in being able to receive 2
Chapter 1. Introduction services in an access agnostic manner rather than dealing with specific access technologies. They want the technology specifications of the access network to be transparent to them. IP technology is capable of offering service to users in a technology transparent manner. IP-based access networks such as wireless local area networks (WLANs) are becoming important components of public access networks. Furthermore, IP-based wireless access networks can integrate with the Internet easier and with lower cost than traditional circuit switched networks. On the other hand, market and business drivers are also playing a major role in convergence. The telecommunications industry is now in a consolidation phase where the biggest challenge is the integration of heterogeneous networks with IP core networks and IMS. Using an IP backbone to integrate various networks is advantageous because many of the integration issues can be solved with standard network elements and next-generation applications and services can be efficiently and effectively deployed over the resulting NGN. From a non-technical point of view, convergence is a matter of cost, competition and regulatory legacy. One of the driving forces of convergence is cost, which has several influencing factors. The first one is accomplished by a shared infrastructure employing a common technology, which leads to reductions in network management costs, support staff costs, and costs associated with service modifications and intra-office calling charges. The second cost driver is capturing of new revenue streams through new business models. The third cost driver is increased productivity and efficiency through introduction of new features. The fourth cost driver is better control of capital that leads to reduction of operational and financial risks through deployment of a single management interface. The fifth driver is client demands for IP-based products. Another driving force of convergence is competition. An example of a feature that can create competition is number portability, a feature that allows users to switch between service providers without changing their telephone numbers. This mechanism enhances fair competition among telecommunication service providers. There are three types of number portability: 1. Location portability, which allows a subscriber to move to another location without changing the number; 2. Service portability, which enables a subscriber to keep the same telephone number when changing services; 3. Operator portability, which allows a subscriber to change service provider without changing the telephone number [4]. Bundling is a necessary strategy for service providers to acquire new customers, keep their existing customers, and increase revenue streams. In this manner, customers who are more comprehensively engaged with a service provider are less likely to switch to another provider. Regulatory driving forces are different from the aforementioned, which are driven by economics [5]. During different periods in the evolution of contemporary telecom networks, different layers in the network protocol architecture become converged. Figure 1.1 shows development of networks from the past to the future. The past approach based on the so-called silo model is evolved to the future service-based approach. In this approach, resources are being integrated to support a variety of services. 3
1.1. Conversational Communications over Converged Heterogeneous Networks Providers are moving from service specific silo architectures towards a horizontal, modular architecture that promotes functional reuse and simple integration. The industry believes service providers can achieve optimum cost structure and reduced time-to-market by adopting SOA design principles. IMS aligns SOA design principles and provides an IP-based service-centric control framework that leverages common resources for services to enable rapid creation of multimedia and enhanced services. Furthermore, IMS application servers are capable of interacting with Web 2.0 applications, which are social networking applications that emphasize on collaborations. One of the significant requirements of Web 2.0 services is event-based rating and charging functions which can be enabled by the Policy and Charging Control Function (PCRF) defined in IMS standards. This feature can also be leveraged to offer Fixed Mobile Convergence (FMC) solutions in conjunction with wireless partners and cable operators. For instance, users can receive message waiting indications for messaging applications on their television sets. Enhanced services offered by IMS can further be used as a distribution channel [3]. For example, in messaging applications and services, mailboxes can be utilized so that users could receive targeted offers of new services and the option to purchase the services. One of the attractive drivers for next generation IMS enhanced services is that service providers are able to offer services to small and medium sized businesses via the software-as-aservice (SaaS) business model. Integration of heterogeneous access networks via a core IP-network supporting IMS enables flexible creation and provisioning of application services via the SOA, which can be broadly divided into the following layers from the bottom up: • Interworking and media layer • Session control layer • Service layer The interworking and media layer consists of the core-IP network inter-working with the heterogeneous access networks to provide various media transport capabilities. SIP signaling is used for initiating and terminating sessions via CSCFs, media servers and other interfaces such as PSTN gateways, under the management of the session control layer. These two bottom layers of the SOA can be implemented using the IMS. In the next subsection, the functionalities of the Service Layer are described in details.
1.1
Conversational Communications over Converged Heterogeneous Networks
Conversational communications consist of interactive voice, video and other multimedia applications that are sensitive to round trip latency. Users should be able to port conversational services over multiple network access interfaces within the service provider’s converged network without experiencing service degradation. Conversational services are sensitive to delays but have less impact from packet losses. In the legacy Public Switched Telephone Network (PSTN) environment employing time division multiplexing (TDM), the typical round trip latency of a domestic landline call is about 60ms while mobile-to-mobile calls could be as high as 250ms due to air interface delays. 4
1.1. Conversational Communications over Converged Heterogeneous Networks
GSM
PSTN
Service Integration
Internet
Common Application Server Layer
Network Layers
Service 1
Service 2
Common Control Layer
Not Reusable
Not Reusable
Not Reusable
Not Reusable
Not Reusable
Not Reusable
IMS Common Media Layer
Heterogeneous Access Network Technology (Common function)
Device Layer
Figure 1.1: Development of telecom networks from past to future
5
1.1. Conversational Communications over Converged Heterogeneous Networks It is expected that an all-IP-based converged network will provide a similar, if not better, user experience for conversational services. In the following few examples of conversational communications are provided.
1.1.1
Voice Call
It is expected that voice will remain the most important telecom service in the foreseeable future. However, the delivery mechanism for voice service is evolving from TDM to Voice over IP (VoIP) due to value-added services and cost reduction requirements. Although new voice features are easier and more cost effective to implement over an IP network, the TDM-based PSTN network will co-exist with IP for another decade or two. The converged voice network must be able to serve existing PSTN end-points. Service providers should also consider using the IMS network to simulate (e.g., new IP phone equipped with existing PSTN phone features) and emulate (i.e., using IMS and the Session Initiation Protocol (SIP) to replicate PSTN signaling for existing PSTN phones) PSTN functions to support voice service migration. These functions will be described in more detail in Section 2.1.
1.1.2
Video Call
With increasing bandwidth in broadband networks and increasing capability of end-user devices, video calls are expected to become a valued-added service to pure voice communication. It is expected that many of the devices connected to smart homes and mobile networks will be equipped with a camera and be able to send and receive video calls. This video call service will be built on the all-IP architecture. The calling party can use an E.164 number (i.e., regular phone number) or a SIP Universal Resource Identifiers to reach the called party. In addition to a network-based address book, the user can also find a friends presence information on their devices. Due to multiple enduser devices and access networks requirements, the convergence team should specify the bandwidth and codec requirements for video calls to minimize format conversions. If format conversion is needed, it should be performed in an application server to reduce handset processing load. A typical video call may require up to 128kbps for a 4x4 inch mobile or desktop screen.
1.1.3
Multimedia Call
Multimedia call allows a user to communicate to one or more users with different devices over various access networks. A multimedia call can carry voice, video, power-point presentation, etc. A typical application could be in enterprise or vertical markets such as healthcare. Since the multimedia call may involve both TDM and IP services, the underlying networks must be able to bridge the functions and capabilities of these networks in order to deliver the user experience. Conversion functions are necessary for signaling, SIP & SS7, and for bearer, TDM & IP.
1.1.4
Push-to-X Call
Push-to-X refers to a half-duplex communications service between two or more parties and is usually used for short and quick conversations. Push-to-X described in this chapter is an IP-based
6
1.1. Conversational Communications over Converged Heterogeneous Networks system that allows a user to send voice and multimedia contents to other parties. With the recent developments in the Open Mobility Alliance (OMA), it is expected that Push-to-X can be deployed in mobile, desk-top (soft client) and gaming devices. Similar to video calls, Push-to-X call utilizes SIP signaling over the IP infrastructure for user registration, and to access presence, user capability and address book information.
1.1.5
Location-based Service
Location-based technology enables the linkage between physical objects and associated data, effectively turning the world into a geospatial information bulletin board. Location-based data accessed as part of a use case might include: • Environmental details • Cultural information • Historical information • Mythology • Social information about people nearby • Geo-demographic information about the local community • Micro-local commercial information • Specialized enterprise and industry data • Safety information based on actuarial data about health, accidents, and crime • Political data • Facilities information such as telecommunications availability • Local public services from government agencies • Tying to-do lists to location information to facilitate running errands such as grocery shopping or reminding the user to pick up some spare light-bulbs when passing by a hardware store
1.1.6
Message Based Communications
Messaging is becoming an important part of overall communication needs for consumer and enterprise users. Although North American adoption of messaging is not as high as in Europe (due to a flat rate voice plan), it is expected that advanced messaging services proven on other continents will benefit local users by simplifying the services and allowing them to better manage their time. A messaging service is delay insensitive but highly sensitive to packet loss. The bandwidth requirement of each service varies depending on the media and applications. Messaging services can be sent over multiple access interfaces to different end-user devices in the converged network because of its capabilities. Service quality of various access interfaces plays an important role in determining 7
1.1. Conversational Communications over Converged Heterogeneous Networks how end-users perceive these services; e.g., 2.5G wireless access supports a lower throughput than DSL, 3G or Wi-Fi. Future home and office desktop phones will be equipped with a 4x4 inch color screen that can display multimedia content, provide touch screen functions and select contact lists. These phones will be IP-based and be connected to home or office gateways.
1.1.7
Multimdia Messaging Service
MMS technology originated in the mobile environment to deliver multimedia content such as pictures, and small files. The existing MMS delivery mechanism is person-to-person with a store-andforward mechanism. Future MMS will retain this service characteristic with access to user/device capabilities, presence and location information.
1.1.8
Instant Messaging Service
IM is a growing communication tool in the Internet community and is expected to replace part of the voice service. There are a number of IM services, such as MSN, Yahoo and AOL, which dominate the market place. A service providers IM service should be able to inter-work with these solutions to gain market traction. Presence information is the key enabler of IM service and will be shared with other multimedia enablers to create compelling applications. There are two open standard bodies (i.e., OMA and Third Generation Partnership Project 2, 3GPP2) developing presence capability in IMS and mobile environment and they should be considered as the base for presence architecture for this document.
1.1.9
Unified Messaging Service
Messaging functions defined in this chapter include SMS, voice mail, Email, Multimedia Messaging Service (MMS), and Instant Messaging (IM). We explore the possible integration of these functions into a Unified Messaging (UM) service to enable compelling services and cost reduction in the convergence environment. The recent development of voice Extended Markup Language (XML)[6] technology is a good example of how messaging services can be converged in text, voice and multimedia.
1.1.10
Content-on-Demand Communications
As increased bandwidth and advanced device capability become available in consumer and enterprise segments, content-on-demand services become feasible. These services allow consumers to access content in real-time over multiple access networks and allow enterprises to push commercial advertisements to user devices based on their locations in order to target specific customers. Telecom operators around the world are implementing an integrated entertainment strategy to defend their positioning with the cable companies. Entertainment is believed to be the 3rd highest area of spending in household budgets, behind education and healthcare, and is one of the major drivers for an all-IP network architecture in the telecom industry.
8
1.1. Conversational Communications over Converged Heterogeneous Networks
1.1.11
Browsing Service
Browsing service refers to Internet surfing with various terminals and devices. Internet surfing clients on desktops and laptops are standardized with several available web browsers, while wireless devices use Wireless Application Protocol (WAP) clients to customize the content to be presented on the limited screen. In order to converge the Internet surfing experience, not only does the system have to know all end-user device capabilities but also be able to port services over multiple access interfaces.
1.1.12
Content Download
Content download is popular especially in the music and gaming segments. To protect intellectual property rights, the industry is leading toward Digital Rights Management (DRM) mechanisms to restrict users in how they can port or share content between various devices.
1.1.13
Content Streaming
Content streaming refers to a media file that is being played on the server side and streaming through the telecom network. Different from download-and-play, streaming requires QoS support on the network to ensure user experience. Depending on the content type, streaming content can be real-time like news or static like movie trailers.
1.1.14
Content Push
Content push refers to content delivery over various access interfaces based on user demand or network trigger. Typical applications can be location-based advertisements or map-quest services.
1.1.15
Mobile Broadcast and IPTV
Broadcast refers to uni-directional content delivery over a wireless or fixed broadband to enduser device or terminal. Different content broadcast technologies exist in landline and mobile environments.
1.1.16
Peer-To-Peer Application
Peer-to-peer service refers to direct communications between two end-user devices. Peer-to-peer communications are different from client-server communications in that the service provider likely does not have control of the user traffic in a peer-to-peer communication session. With the Internet experience and young generation adoption of this service, peer-to-peer service will inevitably be requested by end-users.
1.1.17
Session Persistence
During the course of a communication session between two peering points the characteristics of the session may change, for example an audio call can transform into a multimedia call. In addition,
9
1.2. Contributions the communication session can transform between devices and/or networks (for example, a fixed line call can become a mobile call while maintaining the session). Communication devices are evolving to be personal devices and need to change characteristics dynamically depending on the individual, environment, and/or the event.
1.2
Contributions
In this thesis, database architecture and signaling scheme is proposed for IP-based heterogeneous wireless access interworking. The aim of this work is to reduce the delay and signaling cost as a result of frequent changes of access mode which leads to excessive queries of database. The emphasis of this work is on the IMS view of deploying a unified logical database. Furthermore, the problem of call delivery in a heterogeneous wireless access network setting is addressed. This scheme is based on Reverse Virtual Call setup (RVC) and utilization of application servers in Service Delivery platform (SDP). One of the advantages of applications in SDP control the task of call delivery is that it facilitates the charging / billing tasks. Another advantage is that it facilitates control of access to network resources via a centralized policy enforcement function.
10
Chapter 2
Background2 2.1
IP Multimedia Subsystem (IMS)
The requirements for enabling ubiquitous and unlimited access for users to networks and to enable and enhance the competition of service providers, have led to the definition of NGNs in which service related functions are made independent from underlying transport related technologies. Furthermore, NGN will enable service providers of any type to deliver their services to end users in a network and terminal agnostic manner. IMS [7] is the enabler of converged communication technologies over NGNs [8]. The IMS is an overlay network initially defined by the 3rd Generation Partnership Project (3GPP) in the past years and is currently in the development phase. Session Initiation Protocol (SIP) is mainly used in IMS for signaling. Extensions of SIP define several service enablers such as voice over IP (VoIP), multimedia streaming, presence, instant messaging, push to talk, etc. IMS provides an architecture that enables ubiquitous cellular access, convergence of fixed and mobile networks, user mobility, access-agnostic application development and a service centric framework that makes the development of new revenue generating services possible. 3GPP is in charge of developing specifications for the Global System for Mobile communications (GSM). GSM has two modes of operation: circuit switched and packet switched. GSM circuit switched (CS) mode uses technologies that are commonly used in the PSTN. CS networks have two planes: signaling plane and media plane. The signaling plane is in charge of service invocation that includes protocols to establish a connection between calling and called terminals. The media plane is in charge of user data transmissions between end terminals. General Packet Radio Service (GPRS) is a standard for GSM packet switched network that forms the base for 3GPP release 4 packet switched domain. This domain allows users to connect to the Internet via native IP-based packet switching. IMS provides an architecture that enables ubiquitous cellular access, convergence of fixed and mobile networks, user mobility, access-agnostic application development and a service centric framework that makes the development of new revenue generating services as described in the previous section possible [9]. The aims and motivations of IMS are: 1. To combine the latest trends in technology (i.e., fixed, wireless and mobile networks convergence); 2
Versions of this chapter have been published. (1) P. TalebiFard, T. Wong, and V.C.M Leung, ”Integration of heterogeneous wireless access networks with IP-based core networks”, book chapter in Heterogeneous Wireless Access Networks, Ekram Hossain ed., Springer, 2008. (2) V.C.M. Leung, T. Wong and P. TalebiFard, ”Breaking the Silos Access and Service Convergence over the Mobile Internet”, Proc. ACM MSWiM, Vancouver, BC, Oct. 2008 (invited)
11
2.1. IP Multimedia Subsystem (IMS) 2. To enable user handover, roaming and mobile Internet; 3. To build a common platform for creating drivers of multimedia applications and possibility of using any terminal type such as personal computer, PDA, mobile telephone, etc.; 4. Convergence of various data communication types. In order for IMS to deliver services to end users, the following requirements should be met: 1. Enabling establishment of IP multimedia sessions – Audio and video communication services over packet switched networks are of the most important services that benefit from this. 2. Support of QoS negotiation – IMS shall allow operators to differentiate between groups of customers by controlling the QoS provisioned to each session. 3. Interworking with the Internet and CS networks – Interworking with the Internet expands the potential sources and destinations of multimedia sessions. Interworking with CS networks allows co-existence of IMS-based NGN with the legacy PSTN and enables connections between IMS terminals and ordinary telephones in PSTN domains. 4. Support of subscriber roaming between service domains - Roaming capability enables subscribers to access subscribed service while visiting a foreign service domain. Roaming has always been supported by cellular networks in the second generation and beyond, but limited to service domains employing a common access technology. IMS shall retain this requirement and further extend this capability across service domains employing heterogeneous access technologies. 5. Policy-based service delivery - This allows operators to control services delivered to end users by imposing general policies that apply to all the users in a network and individual policies that apply to a particular user. 6. Rapid service creation – IMS shall reduce the time it takes to introduce new services by standardizing service capabilities so that new services can be readily composed from these capabilities. This obviates the need to pre-establish a large selection of standardized services except for those that are widely subscribed to. 7. Multiple access technology support – IMS is IP-based and independent of the network access technology so long as it supports IP [9]. IMS shall support access to networks other than GPRS, such as WLAN and broadband wireless access.
2.1.1
Architectural Elements
3GPP is also responsible for standardizing a collection of IMS functions linked by standardized interfaces. Figure 2.1 gives an overview of the IMS architecture as specified by 3GPP. One of the most important components in the signaling plane is the signaling protocol for session control. SIP [10] is the protocol chosen to perform session control signaling in IMS. The main purpose of SIP is to deliver a session description to the user equipment (UE) at the current 12
2.1. IP Multimedia Subsystem (IMS)
BGCF
IM-MGW
S-CSCF
I-CSCF MGCF
IM-SSF
MRFC
P-CSCF
PSTN UE
Figure 2.1: Overview of core elements of the IMS architecture specified by 3GPP
13
2.1. IP Multimedia Subsystem (IMS) location of the end user. SIP is completely independent of the format of the object it transports. It means that SIP can deliver a description of a session written in Session Description Protocol (SDP) or any other format. For instance, it can be used to deliver IM. SIP can track the location of a user by means of identifiers such as SIP Universal Resource Identifiers (URI) in the Internet. Core network subsystems of IMS are briefly described as follows [9]: • Home Subscriber Server (HSS) and Subscriber Location Functions (SLF) databases. HSS is the central repository for network provisioning and user-related information that includes user profile data (e.g. identification, contact information, preferences, security information, subscription information, etc.). HSS is technically an evolution of the Home Location Register (HLR) in GSM and GPRS. It incorporates all of the interfaces and functions associated with a mobile networks’ HLR. The HSS can serve as the master database and distribute provisioning and user profile data to individual network elements, applications, service enablers, etc. as needed. This not only will enable service provider to converge its wireless and landline provisioning systems but also establish a single interface between an operator’s converged network and its IT organization over which provisioning and billing information is communicated. SLF is a simple database that maps users’ addresses to entries in the HSS. • SIP servers that are collectively known as Call Session Control Functions (CSCFs). There are three types of CSCFs: Proxy-CSCF (P-CSCF), Interrogating-CSCF (I-CSCF) and ServingCSCF (S-CSCF). P-CSCF is the first point of contact between the IMS-capable UE and the IMS network through which all the requests initiated by the IMS UE or destined to the IMS UE traverse. P-CSCF can be located either in the visited network or in the home network. I-CSCF acts as a SIP proxy located at the edge of an administrative domain. It also has an interface to the SLF and HSS to enable it to retrieve user location information from the databases and route the SIP request to the appropriate S-CSCF. A network may include a number of I-CSCFs for scalability and redundancy. It is often located in the home network and in some cases in the visited network. S-CSCF is the central node of the signaling plane, which is located in the home network and acts as a SIP server. It also performs session control, provides SIP routing services, enforces the policy of the network operator and acts as a registrar. The entire SIP signaling to/from the UE traverses the allocated S-CSCF. A network usually includes a number of S-CSCFs for the purpose of scalability and redundancy. • SIP Application Servers (ASs) are SIP entities located either in the home network or in an external third party network. The role of Application Servers (AS) in IMS is to host and execute services on behalf of an end system depending on the application. The AS can interface with CSCF using the SIP. If located in the home network, it can interact with HSS via Sh, ISC or MAP interface. Application servers can be deployed in several modes of operation [11]: 1. AS acting as terminating UA or redirect server 2. AS acting as originating UA 3. AS acting as a SIP proxy 14
2.1. IP Multimedia Subsystem (IMS) 4. AS involving in third party call control/B2BUA mode Figure 2.2 shows possible roles of AS. • Media Resource Functions (MRFs) are located in the home network and perform media related functions such as media generation and processing in the home network. MRFs are further divided into a signaling plane mode called Media Resource Function Controllers (MRFC) and a media plane mode called Media Resource Function Processors (MRFP). • Breakout Gateway Control Functions (BGCFs) are in charge of routing functionalities based on telephone numbers. It is used for cases where an IMS UE initiates a call to a CS domain like PSTN. • PSTN gateways act as interfaces to/from PSTN/CS domains. PSTN gateways are decomposed into a Signaling Gateway (SGW), Media Gateway Control Function (MGCF) and Media Gateway (MGW). SGW interfaces the signaling plane of a CS network. MGCF is the central node of PSTN/CS domain that is in charge of call control protocol conversion and management of resources in a MGW, which provides interfaces the media plane of the PSTN/CS networks. IMS UE should meet some prerequisites in order to be able to operate in an IMS environment. Two access levels should be granted to the IMS UE. 1. Access to an IP Connectivity Access Network (IP-CAN). 2. Access to IMS via IMS level registration. Registrations in the above mentioned layers (IMS and IP-CAN) are independent of each other because in the IMS architecture IP-CAN and IMS application (SIP) are handled by separate layers. The location of the user can be determined at the IMS registration level through the IP address of the UE that is assigned by IP-CAN. In the following an abstract view of the required steps to operate in an IMS environment is summarized: 1. Establishment of IMS service contract through which the IMS service provider will authorize the UE to use the IMS service. 2. Obtaining an IP address and gaining access to IP-CAN. 3. P-CSCF address discovery that can take place as a separate procedure or as a part of IP-CAN connectivity. 4. IMS level registration where the IMS can locate the user through the IP address of the UE, authenticate, authorize session establishments and create security associations.
15
2.1. IP Multimedia Subsystem (IMS)
2.1.2
Home Subscriber Server
HSS is a key component of IMS which is an evolved version of HLR. In the current wireless networks, HLR is the essential component that provides access to mobile users in conjunction with the VLR and MSC. It also enables roaming within other networks. Although HLR stores the majority of subscribers‘ information, yet it is not a comprehensive database that stores all users profile information. HSS on the other hand provides a much wider range of features and is meant to act as a central master repository database. In addition to many of similar roles as HLR, HSS also provides routing information, maintains/tracks the location of a subscriber and keeps track of network resources. For the above mentioned reasons, the industry consensus is the deployment of a single logical repository database for all user related information. Migration from HLR to HSS practically can be done by initially enhancing the existing HLR elements to support additional end user requirements and interfaces with IMS network elements. HSS functionality will be more extensive and having such a repository database in a single location may not be feasible. HSS physical nodes may therefore need to be deployed in a more distributed (geographically) manner although HSS will logically be seen as a single repository network element. Figure 2.3 shows the interworking of HSS with different domains [7].
2.1.3
Service Layer
As shown in Figure 2.4, the Service layer consists of: 1. A Service Delivery Platform (SDP), 2. A centralized policy management function, and 3. Service enablers The SDP establishes a set of application programming interfaces (APIs) northbound to the application layer above by abstracting the complexity of the underlying network infrastructure from the application. The centralized policy management function authorizes or denies requests for network resources made by any application or user based on business rules, user profiles, and/or network resource availability. Service enablers perform tasks common to multiple applications necessary to deliver a service. For instance, a centralized charging agent that acts a service enabler in the service layer can facilitate consolidated billing and establish a single interface between the SDP and the network providers IT department. Service Delivery Platform Currently applications are tightly integrated with a networks infrastructure in the form of silos. The converged architecture should utilize common functions in each layer to reduce the number of network elements and interfaces and hence, simplify processes, reduce service development time and operation cost. The SDP facilitates this strategy by: 1. Abstracting the underlying network infrastructure for the applications using northbound common APIs, and 16
2.1. IP Multimedia Subsystem (IMS) 2. Interfacing the policy management module and service enablers with UEs via the IP network southbound. The SDP performs four major functions: 1. Interface with external applications and third party service providers to facilitate secure access to network resources and service enablers (e.g., location-based service (LBS), presence), which enables the UEs to access the external applications via the Parlay X [12] module that resides within the SDP. For example, the SDP via the Parlay X module performs bill settlement with external application or third party service providers. Similarly, Parlay X enables the network provider to securely share its network capabilities/resources in a controlled manner with enterprise customers, service integrators, and/or mobile virtual network operators (MVNOs). 2. Interconnecting internal applications and service enablers with various APIs. The Parlay group has defined a set of APIs for most mobile and landline service enablers. Also, 3GPP2 has outlined in Multimedia Domain (MMD) Rev. A [13] how IMS should interface the SDP through IMS Service Control (ISC) APIs. These APIs will establish an environment in which applications can discover and share information with each other. For example, for an application that directs a user to nearby bank machines, the SDP will enable it to: • Obtain the UEs location from a LBS application, • Retrieve relevant maps from a content server, and • Display a map on the UE with the locations of the user and nearby bank machines identified. 3. Providing charging information and billing records to a charging enabler. All application traffic traverses the SDP; therefore the SDP is optimally located to monitor and collect charging information for all applications. 4. Controlling access to network resources via a centralized policy enforcement function. Although a number of leading SDP vendors have integrated the policy enforcement function within their SDP product offering, telecom service provider should insist that this function be separated into a stand alone module for functional development purposes. The SDP enables applications to discover and to interwork with each other. This capability must also include discovering clients residing in UEs so that the SDP can invoke applications. Consider the bank machine locating application described above. This application may require downloading, or pushing, a client designed to display a map onto the UE. Determining whether the UE already has the client needed to complete a service is a function of the SDP middleware. Unfortunately, development of such SDP middleware has not been standardized and is vendor specific. Nonetheless, a network service provider can implement a common SDP middleware stack on all UEs to facilitate the convergence of its wireless and landline services.
17
2.1. IP Multimedia Subsystem (IMS) Policy Enforcement The Policy Enforcement (PE) module enables the service provider to manage and control access to its applications and network resources from a centralized location. It must also support and track the addition, modification, or removal of a resource (e.g., application, service enabler, and/or network element) as well as the policy associated with that resource. In addition, the PE module must support interrupting a service and/or accepting/rejecting a service request as a result of enforcing a policy. The PE module requires information from other modules within the Service Layer to perform these tasks. This information includes, but is not limited to, the following: • End user subscriptions • End user class (e.g., gold, silver or bronze) • Service Level Agreements (SLAs) between the network service provider and third party service providers • End user account status (e.g., online, offline) • End user personal data and preferences • Network resource availability and conditions • Regulatory and legislative variables and conditions (e.g., privacy, security, emergency) The PE module may approve, reject, terminate or negotiate changes to any service/resource request from a user or application. If the PE module rejects or terminates a service, it must inform the requestor/user of the cause in real-time. The PE module should retain a log of its policy decisions and record all of the events it observes (e.g., errors, resource request and responses, etc.). This data can then be relayed to the Support System Layer to facilitate trouble shooting and enhance customer service. Policy rules enforced by PE module are divided into three categories and each policy rule can be executed by a policy engine (service broker): • Service policy rules • User policy rules • Network policy rules These policies/rules can be communicated with other entities using the Generic User Profile (GUP) [14] framework. As shown in figure 2.5, when a user requests access to a service or application, the request is relayed to the policy enforcement function for verification and authorization. If the request is approved, the PE module checks and allocates the underlying network resources. In the following subsections we provide explanation to each of these categories and how they relate to HSS.
18
2.1. IP Multimedia Subsystem (IMS) Service Policy Module The service policy entity contains a set of business rules that authorize/deny: • Third party application service providers (ASPs) access to service providers network resources, and • End users access to other users (within or outside an operators domain) and services (provided internally by the service provider or externally by a third party ASP). This module also enables third party service providers to exchange service policy information with the network service provider. The service policy module should support a number of security protocols that enable a service provider to securely exchange information with third party ASPs, users, and trusted external service providers. If this exchange is between internal network users or within a trusted domain, a dedicated link or private network may be used. If the exchange is between un-trusted domains or through the Internet, encryption should be used. The appropriate security mechanisms/protocols/models supported by the SDP should be determined when the service policy function is implemented. The service policy module should only control the authorization process. It should not perform bill settlement or collect traffic information to facilitate billing. Instead these functions should be performed by a charging enabler and the service policy module should use this charging enabler to determine if: • A user has good credit before authorizing access to a service/resource, and • A service is billed using an online or offline charging mechanism.
User Policy Module User policies are determined by the data associated with a users profile as stored in the HSS. The user profile contains user preferences, service priorities, device configuration information, service subscription data, and network registration information. The user policy module does not authenticate a user; this task is performed by the Control Layer using the HSS. Instead, the user policy module authorizes access to network services/resources after a user is authenticated. Users profiles in existing CS mobile networks are centralized in their HLRs. The HSS described above should include HLR functionality. Users will subscribe to both CS and IP-based services for the foreseeable future. A single database that applies a users profile across both domains is therefore required to deliver services seamlessly across both domains. An HSS that integrates HLR functionality not only reduces the number of profiles associated with a user (thereby eliminating data discrepancies and duplication) but improves the end users experience by making it consistent. Network Policy Module The network policy module enforces the policies associated with the underlying network resources. It should support and interoperate with the charging and service discovery capabilities available within the SDP. Southbound, the network policy function must support a variety of network domains including a heterogeneous mix of access network infrastructures and access network service 19
2.1. IP Multimedia Subsystem (IMS) providers. Northbound the network policy function must support a variety of internal and external applications. All the while, the network policy module must be transparent to the users and applications and not preclude the deployment of any service enablers or limit the scalability of the converged network. The network policy function should have direct access to the Resource and Admission Control Function (RACF) in the transport layer. When a user or application requests a network resource, the network policy module should ask the RACF to determine in real-time if the requested network resource is available. And if the network policy function authorizes a requested service or application (based in part on the response of the RACF), it should secure the resources that satisfy the services QoS and SLA requirements through the RACF. Service Enablers Service enablers as shown in figure 2.4 link directly with clients on user devices to enable applications. For example, a presence server is a service-enabler that facilitates the delivery of presencebased call routing services orchestrated by applications hosted on servers within the Application Layer. Some service enablers will be common to both landline and wireless applications, e.g., presence, intelligent network (IN), unified messaging (UM), multimedia messaging service (MMS), connection manager, content download; other enablers will be relevant to only landline or wireless applications, but not both, e.g., wireless application protocol (WAP), short messaging service (SMS), LBS. A centralized charging agent should be deployed as a service enabler as emphasized by the Charging module within the group of service enablers shown in figure 2.4. This charging module or agent must inter-operate with: 1. Any charging client deployed in any application, 2. Other service enablers, 3. Any charging client deployed in any UE. This increases the complexity of the charging agent as explained below. Charging Agent The charging agent collects information in near real-time to support online (pre-paid) and offline (post-paid) billing systems. The charging agent also generates billing records that capture information such as volume, duration, application, and transaction, and communicates this information to the billing system via a single interface. Consolidating all communications with the billing system onto a single standardized link that interfaces with the charging agent is necessary to abstract the service providers billing system from any changes/upgrades to an application, service enabler, and/or network element. Otherwise such changes/ upgrades would require making time-consuming changes/upgrades to the billing system [15]. Figure 2.6 illustrates the relationship between the charging agent and an application, service enabler, network element, and service providers billing system. The charging client that can reside in the application server or network elements such as P-CSCF or gateways collects user billing information specific to an application, enabler and/or network element. The charging client forwards this information to the charging agent. The charging agent consolidates this information and communicates it to the billing system. 20
2.2. Integrating Heterogeneous Networks using IMS
2.2 2.2.1
Integrating Heterogeneous Networks using IMS Architecture
Packet switched networks are compatible with IMS and can therefore be integrated with IMS through logical interfaces that implement different instances of SIP. These interfaces have been standardized by 3GPP and shown in Figure 2.7. In this figure some of the most important logical interfaces are shown with dotted lines [8]. The Gm interface implements signaling communications between the IMS UE and the core network and offers procedures such as user registration, roaming, handover and service charging.
2.2.2
QoS Provisioning for Heterogeneous Wireless Access Interworking
Any application that requires a specific level of quality assurance from the network needs QoS support. QoS depends on type of service and can be categorized based on various characteristics. Examples of service types are best effort, streaming, interactive and conversational [16]. Native IP (IPv4) is connectionless and offers a best effort service. To identify traffic flows IPv4 offers two ways of traffic flow marking: (1) using the source and destination IP addresses and port numbers, and (2) using the Type of Service (TOS) field in the IPv4 header. Further, methods of Integrated Services (IntServ) (RFC 1633) and Differentiated Services (DiffServ)(RFC 2475) were proposed to meet the market demand for QoS. Two service classes are defined in IntServ: Guaranteed Service (GS) and Controlled Load Service (CLS). DiffServ has been proposed to deal with the scalability problem of IntServ. On the other hand, the next generation IP, IPv6, uses two fields in the IP header to mark the traffic: flow label field and traffic class field. If used with its full features, IPv6 is very powerful in flow identification. IPv6 can also mark an aggregate of flows and can theoretically identify a single customer. However, packet or flow identification is but the first step in solving the problem of QoS provisioning. QoS management functions and traffic control mechanisms need to be implemented in network nodes for QoS support. In the following a few solutions are briefly described. Over provisioning
of QoS is the act of throwing bandwidth at the problem. This becomes the
case when purchasing an oversupply of bandwidth is a simpler and easier solution to the problem of QoS provisioning. Although it can not be classified as a solution as it simply ignores bandwidth optimization and possible future trends of new services, it becomes useful in hazardous situations and military environments. Flow identification
as mentioned earlier is relevant in guaranteeing QoS. Different methods
to classify and identify packets are flow label and traffic class fields in IPv6, as well as TOS field, source/destination IP addresses and port numbers in IPv4. Resource reservation and call admission control (CAC) are needed to reserve the bandwidth and to allocate the resources based on CAC and policies. CAC decides whether or not a new or handoff call should be admitted or rejected based on the resources available and the applicable
21
2.2. Integrating Heterogeneous Networks using IMS Service Level Agreements (SLAs). Each call may be associated with a set of Service Level Specifications (SLS) given in terms of packet loss rate, delay and jitter for each flow or aggregate of flows. However, there is no clear method of identifying the amount of bandwidth and buffer available for a specific flow, except for peak bandwidth assignment. This method has led to many bandwidth allocation problems. The concept of equivalent bandwidth, defined as the minimum rate requirement to guarantee QoS for a single flow, has been introduced in the literature. Equivalent bandwidth techniques are often described by statistical characterization of traffic by parameters such as peak rate, mean rate and maximum burst size. Equivalent bandwidth techniques are hardly applicable to situations where aggregate flows of traffic are heterogeneous. Traffic shaping policies
limit flow rates to their committed rates. Two basic methods of traffic
shaping are leaky bucket and token bucket. Traffic scheduling determines the order that packets buffered in the transmission queue at an outgoing link of a packet switching node are sent over the link. One of the very basic algorithms is first-in-first-out (FIFO); however, more sophisticated traffic scheduling algorithms are needed to satisfy the delay and jitter QoS requirements of real-time flows such as voice and video. Queue management
is often linked to traffic scheduling where packet dropping strategies can be
applied when the transmit buffer is getting full. Other queue management strategies might involve partitioning the available buffer space so that some buffers may be dedicated to an individual traffic flow. QoS routing
is concerned with management of an end-to-end connection by routing it through
appropriate intermediate nodes to ensure that the QoS requirements of connection can be met. QoS routing is part of the CAC process, which also takes into account of traffic shaping and scheduling/buffer management in intermediate nodes. Provisioning QoS over interworking heterogeneous networks can be divided into the following steps: 1. CAC requests with specific QoS requirements should traverse end-to-end the interconnected networks that implement different technologies (e.g., cable, cellular, WLAN) and different protocols (i.e., ATM, TCP/IP). 2. QoS requirements should be interpreted according to the protocol and QoS mechanisms used in each individual network involved in the end-to-end connection. 3. The protocol stack of each of the interconnected networks consists of several layers each having a different scope for QoS provisioning. 4. Mapping should be done between the corresponding layers of the respective protocol stacks employed by the interconnected networks, such that the role played by each layer in provisioning QoS over the corresponding network can be harmonized across adjoining networks and end-to-end QoS can be guaranteed. 22
2.2. Integrating Heterogeneous Networks using IMS In the first two steps the QoS requirements are transferred to different network portions that implement their own technologies and protocols. It is referred to as Horizontal QoS mapping. Steps 3 and 4 show the concept of vertical QoS mapping that is based on the idea that a network is composed of functional layers and the overall QoS depends on the QoS achieved at each layer of the network. NGNs introduce new requirements for mobility management and provisioning of QoS in heterogeneous networks. QoS requirements include QoS guarantee among different domains/networks and end user perception of QoS. Some examples of mobility requirements are ability to change access point, ability to get access from any network and awareness of user availability. Mobility management with QoS support is of particular importance in NGN that is one of the driving forces of FMC, as it is desirable for users to be able to roam between fixed and mobile networks employing different technologies. There are three requirements to achieve end-to-end QoS in FMC [17]: 1. Capability of network controllers in individual networks to process resource reservations according to aggregate traffic demand, while maintaining fairness across individual traffic flows. 2. Mapping functions to convert the end-to-end QoS requirements to specific levels of QoS support in different types of networks. 3. Fast handover schemes to reduce the handover latency within a single domain and across heterogeneous access domains as a result of user roaming [18]. In access networks employing wireless technologies, QoS management needs to be supported via appropriate radio resource management (RRM) mechanisms. RRM schemes are designed to allocate scarce wireless resources to all users efficiently, fairly and to provide QoS support. The random nature of wireless channels due to path-loss, fading and shadowing brings many challenges in RRM. Due to the statistical nature of the wireless channel, soft or probabilistic QoS schemes may be more applicable in wireless access networks. For instance, RRM algorithms can provide soft QoS guarantees in an opportunistic way [17] so that in the short term users with good channel conditions are favored but in the long term all users receive the required throughput. In [17] QoS management for integrated WiFi and WiMAX services is discussed and a heterogeneous network architecture of FMC is demonstrated. Provisioning of end-to-end QoS in IMS over a WiMAX architecture is described in [19]. In this work a hybrid cross layer QoS scheme based on IMS over WiMAX is discussed. In this scheme, UE is responsible of initiating the QoS negotiation, resource reservation and cancellation. It is based on the following procedure: 1. Application layer QoS authentication at Policy Decision Function (PDF), 2. Vertical mapping of QoS for application layer and MAC layer at UE, 3. MAC layer and IP layer QoS mapping at the Access Service Network (ASN), 4. Intserv and DiffServ mapping at the Packet Data Gateway (PDG). NGNs conceptual model for QoS and mobility management consists of two main elements: Domain Policy Manager (DPM) that acts as a policy decision point and PE module which ensures 23
2.2. Integrating Heterogeneous Networks using IMS that only authorized IP flows are allowed to use the network resources that are reserved and allocated to them. The framework for end-to-end QoS support is presented in [20] released by 3GPP. Figure 2.8 shows a conceptual model for policy-based QoS provisioning in IMS. In this figure policies can be stored and retrieved to/from the policy repository database. PDF is in charge of retrieving policies from policy repository and controlling the PEP. The role of PEP is to take actions based on the decisions made by PDF. PEP is located in the Gateway GPRS Supporting Node (GGSN) because GGSN is in the data path. PEP and PDF communicate in two possible modes of communication, namely push mode and pull mode. In the push mode, communication is initiated by PDF and the decision is sent (pushed) to PEP for further actions. In the pull mode, communication is initiated by PEP to request a decision (pulling from PDF). As shown in the figure, several external IP networks are connected via data paths and may consist of several different domains. In order to provide end-to-end QoS support, it is required to provision QoS within each domain. An IP Bearer Service (BS) manager is used to manage the external IP BSs. As different domains may employ different technologies, translation functions may be needed to communicate QoS requirements to the other external BS managers. When a Packet Data Protocol (PDP) context is set up to enable a UE to access a packet switched network service, the UE shall have access to the following alternatives: • Basic GPRS IP connectivity service for which the communication is established based on the users subscription and local operators admission control, policies and roaming agreements. • Enhanced GPRS based services for which the bearer is used to support application layer services such as IM and would also be in charge of policy control decisions. In a situation where resources that are not owned by a certain network are required to provide QoS, interworking with external network that controls those resources is required. This interworking can be done in a number of ways [20]: 1. Signaling along the flow path (e.g., using the Resource Reservation Protocol, RSVP), 2. Packet marking or labeling along the flow path (e.g. DiffServ, MPLS), 3. Interaction between policy control and resource management elements 4. Border routers to enforce SLAs. 3GPP has also adopted another policy-based QoS solution in which sufficient resources to support a specified QoS are provided to authorized users. In this framework, policy rules are stored in a policy repository and PDF is able to retrieve appropriate policy rules. This action is triggered at the PE modules by contracted QoS-enabled IP services [21].
2.2.3
Mobility Support
Mobility management in NGN should be distinguished from mobility management in previous generations of networks in the sense that vertical handover, heterogeneity and roaming would also be involved. Support of global roaming in NGNs requires interworking of heterogeneous network 24
2.2. Integrating Heterogeneous Networks using IMS accesses with different mobility management techniques [18][22][23]. Different aspects of mobility can therefore be realized: Terminal mobility that is the capability of a mobile terminal to continuously remain connected to access the network in range. User mobility
refers to the ability of a mobile user to continuously access services from a network.
Service mobility
is the portability of a service which enables a user to access the same service
independent of location and terminal. Future wireless IP networks are expected to support basic mobility requirements such as: • Support of all forms of mobility and applications. • Support of mobility across heterogeneous wireless access networks. • Support of continuous service delivery and access. • Support of global roaming The most important components of mobility management are location management, packet delivery, handoff management, roaming and admission control. Location management schemes in current cellular networks were capable of the following procedures: location update and location discovery by paging. Location update strategies decide when a mobile terminal should perform location update and what information should be communicated to the network. The decision can be based on time, movement, distance or probability distribution function of time, movement and/or distance. Location discovery by paging is required when the network does not know the exact location of a user. The main goal in design of location discovery strategies is to minimize the paging cost. In the current cellular network the HLR and VLR databases are used to enable mobility management. For the case of handoff management in heterogeneous networks, in addition to triggering handoff and decision criteria, choosing the appropriate access mode based on application requirements and QoS mapping issues are of great importance [24]. Use of IPv6 can enhance the support of terminal mobility. At the time of IMS standardization, IPv6 was also being standardized and 3GPP considered the applicability of IPv6 to be deployed. Deployment of IPv4 in a large scale would lead to the need of Network Address Translations (NAT) as allocations of private IP addresses would be required. As the first IMS product reached the market, IPv6 had not taken off and this caused some efforts towards making IMS compatible with IPv4 via deployment of (IPv4 and IPv6) dual stack implementations of IMS UEs [9]. IMS is mainly based on MIPv6 and uses application layer SIP for signaling and control of real-time multimedia applications. SIP may also be used to support terminal mobility. Support of terminal mobility using SIP attracts many researchers to work in this area. Similar to email users that are identified by their email address, SIP users can be identified by their SIP URIs that works as a unique global Network Access Identifier (NAI). In terms of mobility support, the difference between MIP and SIP-based mobility is that unlike MIP, SIP servers are only in charge of setting up the sessions between users and once session are set up, data traffic will be communicated directly between end 25
2.2. Integrating Heterogeneous Networks using IMS users. This also solves the triangular routing problem of MIP. However, when a terminal moves to a new access network, it may need to register its new IP address with the SIP server and it can cause a high load on the home server. One of the solutions to this problem is to use a hierarchical registration similar to MIP. It is also difficult for a roaming terminal to keep a TCP connection alive while changing its IP address. This is a rather challenging problem and one approach is to consider using the Stream Control Transmission Protocol (SCTP) with its multi-homing features for end-to-end reliable transport. Other solutions are also summarized in [25].
26
2.2. Integrating Heterogeneous Networks using IMS
Application
AS
AS
SIP Dialog
IMS Core
CSCF
HSS
Application Servers acting as Terminating UA
SIP Dialog
IP Connectivity
Application
AS
AS SIP Dialog
SIP Dialog
IMS Core
CSCF
HSS
Application Server acting as SIP Proxy or 3rd party call control
SIP Dialog SIP Dialog
IP Connectivity
Application
AS
AS
SIP Dialog
IMS Core
CSCF
HSS
Application Server acting as an originating UA
SIP Dialog
IP Connectivity
Figure 2.2: Role of Application Servers
27
2.2. Integrating Heterogeneous Networks using IMS
CS Domain GMSC
MSC/VLR Diameter Sh
PS Domain
AS
GGSN
Gc MAP
SGSN
Gsm-SCF
Gr Si
Sh
IM-SSF
OSA SCS
Cx CSCF GGSN: Gateway GPRS Support Node SGSN: Service GPRS Support Node OSA SCS: Open Services Architecture Service Capability Server IM-SSF: IP Multimedia Service Switching Function CAMEL: Customised Applications for Mobile network Enhanced Logic
GMSC: Gateway Mobile Switching Center AS: Application Server CSCF: Call Session Control Function GUP: Generic User Profile
Figure 2.3: Functionality of HSS node at logical level across multiple domains
28
Web Services GW Trusted services
OSA/SDP SCS
RACF
User Inter
BPEL
Ge Ms n g
Poli cy Terminal Cap
Policies
Chargi ng Data
Account Man Connect
Session
Man
Mobili ty Presen ce
Parlay X
Frame Frame Chapte ‘work ’ work r#3
Security
Identity management Unified user profile
2.2. Integrating Heterogeneous Networks using IMS
Policy Enforcement Module Enablers
SMSC Device Manager
SIP Based
MMSC
Content Download
IN Presence
WAP
OSS/BSS Systems
Authentication/ Authorization
LBS
Charging
CS Network
Non-SIP Based Transport Layer
IP Network Cloud
TDM Network Cloud MG
Access Network Layer Fixed Wireless
QoS enforcement Traffic Statistic CAC
Device Layer
Figure 2.4: Details of Service Layer in SOA.
29
2.2. Integrating Heterogeneous Networks using IMS
Response Request
Service Policy
User Policy
Network Policy
Response Request
Figure 2.5: Policy-based Access Control.
30
2.2. Integrating Heterogeneous Networks using IMS
Charging Agent
Network Elements w/ Charging Function Landline & Wireless
Figure 2.6: Charging for Converged Services.
31
2.2. Integrating Heterogeneous Networks using IMS
Foreign IMS Networks Mk
Mm
Mm
I-CSCF PSTN / CS
BGCF Cs
Mg
Mw
Mj
AS
ISC
Dh
Sh
Cx
Mi Cx
Cs
-
Mn MGCF Mg Mw
IM MGW P-CSCF
Gm
WLAN Access Network
HSS
S-CSCF Dx Mw
P-CSCF
SLF
IMS Subsystem
Gm
3G Access Network
UE UE
Figure 2.7: Layout of logical interfaces for integration with IP core networks
32
2.2. Integrating Heterogeneous Networks using IMS
IMS Subsystem IMS Core Elements
MGCF
P-CSCF PDF
Policy Repository
PSTN / CS
Mn
MGW
UE
SGSN
GGSN PEP
Data path External IP Networks IP Bearer Service (BS)
Scope of PDP context
Figure 2.8: Network architecture for QoS conceptual models
33
Chapter 3
Novel Database Architecture and Signaling Scheme for IP-Based Heterogeneous Wireless Access Interworking 3 3.1
Introduction
In a transition to the future generation of wireless networks, convergence of IP based core networks with heterogeneous wireless access networks is inevitable. The Internet is composed of different domains operated by Internet Service Providers (ISPs) with different capabilities, policies and access networks. Next Generation Networks (NGNs) will employ standards and architectures that are based on the IP suite. The IP platform can integrate diverse access networks in a common scalable framework, and through IP Multimedia Subsystem (IMS), extend a wide range of multimedia services to subscribers over heterogeneous wireless and wireline access networks. The need for coexistence of diverse applications and wireless access technologies motivates integration of heterogeneous access networks with IMS. IMS paves the ways to deliver existing services in a more efficient manner while making the creation of new services possible for service providers. These have been pursued by separation of application and service plane from call control plane in IMS. Furthermore, IMS is also based on the vision of single logical view of a central repository database that is called Home Subscriber Server (HSS) and is one of the key components of IMS. To support the heterogeneity, multi-modal mobile devices that provide alternate means to access the Internet are becoming widely available [26]. Interworking architectures are proposed by 3rd Generation Partnership Project (3GPP) that show the interworking of wireless local area network (WLAN) and 3G cellular networks [27] [28]. In today‘s wireless networks each application, service or access network uses its own database for storing end user data. This requires a lot of effort by service providers to coordinate information exchanges across various databases, dealing with different authentication and registration procedures and hence leading to inconsistencies, duplication and data discrepancy. On the other hand, as the future of communication systems merges towards heterogeneity and deployment of multimodal User Equipment (UE), change of access mode (e.g. vertical handovers) within an overlaying network becomes very frequent in a fully overlapped coverage environment. Updating the repository database about current mode of access by mobile users can therefore lead 3
A version of this chapter has been published. P. TalebiFard and V.C.M. Leung, ”Novel Database Architecture and Signaling Scheme for IP-Based Heterogeneous Wireless Access Interworking”, Proc. ICST. WICON, Maui, Hawaii, Nov. 2008
34
3.1. Introduction to excessive signaling traffic. In our novel architecture we address this problem by introducing an Interface Agent (IA) to the 3GPP interworking architecture and proposing an efficient signaling scheme that decreases the signaling cost while improving delay and user experience. This chapter is focused on the idea of deploying a single logical database that can interwork with different access technologies. The problem of frequent changes of access mode is raised and a possible solution is proposed.
3.1.1
Location Management in 3G Cellular Systems
Location management is an important issue in mobility management. Mobility management involves call delivery, location management and handover management. It enables users to roam while simultaneously offering them incoming calls and seamless handovers. Location management involves location registration and call delivery. Call delivery procedure consists of locating a Mobile Terminal (MT) based on the available information in the database when a call to an MT is initiated. Location management also involves the process of discovering the geographical location and current attachment point of an MT. For the purpose of location management, the concept of Home Location Register (HLR) and Visitor Location Register (VLR) is used in the 3G networks. HLR and VLR are used to store the location information of MTs such as account numbers, subscribed features and preferences as well as access permissions and policies. In the literature, several schemes are proposed that are based on centralized database architectures or distributed database architectures [29][30]. For location management and call delivery the database architectures are divided into two categories of centralized and distributed. Centralized architecture
An example of a centralized architecture is dynamic hierarchical
structure [31]. One of the strategies in the centralized category is called per-user location caching strategy that is based on a multiple-copy location information strategy. In this method the user profiles are replicated in multiple databases[32]. Another strategy in centralized architecture is pointer forwarding strategy in which the user location reporting can be eliminated by setting up pointers from old VLR to the new VLR [33]. A similar approach called k-step pointer forwarding is also proposed in [34] by Weng and Chu. Distributed architecture This approach involves having multiple databases across different networks and can be considered as an extension of multiple-copy method. In this approach the location databases are connected in a tree form where multiple levels can be defined. The root at the highest level and leaves at the lowest level. Each MT is associated with a leaf [31]. Towards the integration of heterogeneous networks, one possible approach is to deploy a common HLR that interacts with VLRs in different access networks. However, the problem with this approach is that each individual network uses a different signaling format, authentication procedure and registration method [35]. To make roaming among different wireless access networks possible, the idea of Boundary Location Register (BLR) is proposed [36]. It involves some boundary interworking units that are connected to MSCs. This method is designed for networks with partial 35
3.1. Introduction
IMS Domain AS
Circuit Switched Domain
HSS OSA-SCF
HSS CSCF
S-CSCF
I-CSCF
MGCF
PSTN IM-SSF
P-CSCF
CAP
HSS
MAP
gsmSCF
3GPP PS Domain
HSS 3GPP PS Domain SGSN
Go GPRS / HSPA
Sc GGSN/ PDG
Gr SGSN
WLAN
WLAN WLAN
HSPA WLAN
WLAN
WLAN
WLAN
HSPA
WLAN
WLAN
HSPA
LA 1 GGSN: Gateway GPRS Support Node SGSN: Service GPRS Support Node CSCF: Call Session Control Function IM-SSF: IP Multimedia Service Switching Function gsmSCF: GSM Service Control Function
LA 2 HSPA: High Speed Packet Access AS: Application Server HSS: Home Subscriber Server MGCF: Media Gateway Control Function PDG: Packet Data Gateway
Figure 3.1: Interworking architecture that shows the physical distribution of HSS
36
3.1. Introduction overlaps at boundaries and may not be suitable for multiple networks that are fully overlapped [37]. In a heterogeneous network environment where multiple access networks fully overlap in their coverage, the following issues become important: • Through which access network the user profile information, location and access mode should be updated • In which network should the user profile be stored • How the database repositories (physical nodes) can be managed and distributed. The fully overlapped setting enables efficient ways to improve the capacity and quality of mobile users. To address the problem of heterogeneous access networks with different authentication procedures and registration operations the multi-tier HLR (MHLR) approach is proposed [37]. In the MHLR approach two methods of registration are introduced: Single Registration (SR) and Multiple Registration (MR). Under the SR method, the MT would only register through the lowest tier, whereas in MR method the MT would be able to concurrently register with multiple tiers. Detailed explanation of SR and MR is given in [35]. In a fully overlapped network setting, to solve the earlier mentioned problem about choosing the access network through which the MT can update its location, one possible approach would be updating the location information through all subscribed networks. However, this approach would make the task of call delivery more complex as it causes more data discrepancy. On the other hand it is not clear for the call delivery procedure that which one of the replicas is most recently updated and valid.
3.1.2
User Registration, Location Update and Access Mode changes
In this subsection we overview the existing schemes of user registration and location update as well as a base case approach for change of access mode notification. Figure 3.2 shows the registration procedure when a MT moves to a new LA. The registration procedure is as follows: 1. GGSN / PDG receives the registration request from the access network. 2. The request is forwarded to P-CSCF and the DNS query will be performed to find the I-CSCF. 3. I-CSCF will query HSS for capable S-CSCF and registers the user with the S-CSCF 4. After registration, all SIP signaling will be routed via the same route. When a MT is roaming within a LA and changes of access mode occurs, the core network can be notified using the following procedure as shown in figure 3.3: 1. GGSN / PDG receives the access mode change notification. 2. The decision of access grant permission will be made by PDF and enforced by PE module at the GGSN / PDG.
37
3.1. Introduction
GGSN / PDG
P-CSCF
I-CSCF
S-CSCF
Register – Location Update
DNS Query
Registration
Register Location Update
Change of Location Area (Location Update)
Query HSS
Register ACK ACK
ACK
Figure 3.2: Signaling diagram for location update and registration of a MT when it moves to a new LA
38
3.1. Introduction
GGSN /PDG
P-CSCF
I-CSCF
S-CSCF
Access Mode Change Access Mode Change Access Mode Change
Update HSS
ACK
ACK
ACK
Figure 3.3: Signaling diagram for change of access mode when a MT is roaming within a LA 3. If the access is granted, the notification will be conveyed to the I-CSCF for that domain. 4. I-CSCF will notify the S-CSCF. 5. S-CSCF will query HSS to update the access mode through which the user can be reached at.
3.1.3
IMS release 7 and 8 solution for multi-mode support
IMS registration in release 7 applies to changes of access mode or IP address. However, when the S-CSCF receives a registration request from a MT, it detects whether or not this MT has been registered before, based on its private ID which is practically a device ID. If the same MT is registering again, it considers it as an update and overwrites the previous registration. Similar procedure takes place as a MT changes its mode of access. In this case again the registration update will take place. This procedure eliminated the need for deregistration in case of change in access mode. For example, a MT may lose radio connectivity before deregistration. To eliminate the gap in reachability of a user, release 8 of IMS is making multiple registrations of a MT possible [38]. In this manner, a MT is able to register its public user identity with multiple IP addresses at the same time. A possible approach is to register the HSPA/GPRS interface with a different P-CSCF than the one for WLAN. It results in multiple P-CSCF attachments. As the 39
3.2. Proposed Scheme
Network Element S-CSCF I-CSCF P-CSCF
Table 3.1: Role of core elements of IMS Functionality and Behaviour Acts as a Registrar,Proxy Server & User Agent. As a registrar it stores user registration information in HSS Acts as a Proxy Server, User Agent. Locating/Assigning the S-CSCF to/for the subscriber Acts as a Proxy Server. Forwards SIP resister request to the I-CSCF
mode of access changes, S-CSCF will be notified of the current point of attachment. However, as the access mode changes the signaling traffic should traverse to reach S-CSCF. This chapter is organized as follows. In section 3.2 we explain the methods of user registration and change of access mode notification under the proposed method. In section 3.3 performance evaluation in terms of signaling cost and delay is presented. Section 3.4 concludes this chapter.
3.2
Proposed Scheme
Our assumptions for the proposed model is based on the interworking architecture shown in figure 3.1 where we also demonstrate the physical distribution of HSS [39]. The basic idea is to introduce an Interface Agent (IA) that stores the information about active interfaces for the users in a LA in its cache. The IA is a SIP entity that is co-located with the P-CSCF at the edge of the IMS, running on the same host as P-CSCF. IA can interact with other entities within P-CSCF directly through internal SIP transport since TCP/UDP, IP, MAC and Physical layers are bypassed. Some of the advantages of such approach are less processing time and more reliability. In the context of this work, we define the LA as a set of clusters or coverage areas that can be covered by different service providers and have different converges fully overlapped as we will explain later. In order to explain how our proposed scheme functions, we use the fully overlapped network setting which is a general and yet a more realistic and challenging case. In a LA there might be WLAN, WiMax and High Speed Packet Access (HSPA) coverage at the same time. A multimodal device is able to have multiple interfaces activated at the same time according to user’s preference or automatically controlled by the UE based on available bandwidth and power criteria. In deploying HSS as a central repository database, one possible approach can be that all HSS nodes of individual access networks would store all user related information such as user identification, account number, subscription data, preferences, location as well as current mode of access that the user can be reached at. The aforementioned approach may not be suitable for a fully overlapped network setting environment where multiple access coverages coexist. Mobile users often move within or outside a Location Area (LA). In either case, due to mobility of users, change of primary access mode can be very frequent and this may result in excessive signaling traffic for updating the database. In our proposed scheme, the role of IA is to store the information about active interfaces for the users in that LA in its cache. It interacts directly with the HSS physical node of that LA. Once the user moves outside the LA of a corresponding IA, the record for that user will be removed from the cache. 40
3.2. Proposed Scheme Our proposed architecture utilized the deployment of a single logical database that communicates with the IA. This single database can apply user profiles and store users’ locations across all domains. Figure 3.1 shows the interworking architecture with HSS physical nodes distribution at an abstract level. In the following we explain how our proposed method works. As mentioned earlier, in the context of IMS, HSS is the central repository database that stores user related information which includes user profile data (e.g. identification, contact information, preferences, security information, subscription information, etc.) It furthermore, includes the HLR functionalities. HSS is the industry consensus as a single logical database to store network and user related data. It does not however know about the access mode through which a user can be reached. It knows the LA in which the user is residing and with the help of Interrogating Call Session Control Function (I-CSCF) and Subscriber Locator Function (SLF) can locate the Proxy-CSCF(P-CSCF)and the IA for that LA. Table 3.1 summarizes the role of CSCF and core elements of IMS in the context of registration [7] [9]. In this manner the updates for change of access mode are not conveyed to the HSS or the core IMS network. In addition, it facilitates the access agnosticism of access mode from the core network. The role of IA is to cache the user data about the access mode that the user can be reached at. The final decision will be made by the P-CSCF and the network based on user‘s preferences or availability of network resources. This method improves the user experience, while eliminating data discrepancy and duplication. It also enables a faster call delivery and service time. Our proposed architecture best suits situations where multiple coverage coexist in a LA. It performs best when users roam within a LA. In designing the LAs, one should consider the upper limits on the number of LAs and therefore the number of HSS physical nodes and the way they are distributed. We will demonstrate later in our analysis that the proposed architecture reduces the signaling cost while users roam within a LA or outside a LA to a different or adjacent LA. Yet, the performance will be indifferent if this architecture is deployed in an environment where only single mode of access exists. However, in the future generation of wireless access networks, multimodal devices will be widely used and fully overlapped multiple coverage settings will exist. Therefore, the proposed method can be practically adapted for use in the future converged architectures. In the next section we demonstrate a user registration scenario based on the proposed architecture.
3.2.1
User Registration under the proposed method
In figure 3.4 we show the scenario for movement of a user from LA1 to LA2 in a situation where the two LAs may partially overlap if the LAs are covered by two different service providers. In this section we consider three situations for location update and registration: 1. When the MT is moving away from LA1 : When the user is in the overlapping area of the two LAs the process of registration in the new LA will be performed. In the meanwhile, the old IA (in LA1) will be notified and the entry for that user will be removed from cache. 2. When the MT enters LA2 : Once the MT enters LA2, along the registration procedure, information about the available access modes will be conveyed to the IA. If this is a new entry to the IA‘s cache, then HSS will be notified of the current location of the user. 41
3.2. Proposed Scheme
P-CSCF
P-CSCF IMS Core
IA
IA
HSPA WiMax WLAN
HSPA
WLAN
WiMax WLAN
WLAN WLAN
WLAN
WiMax
WLAN WLAN WiMax
WLAN
WLAN
WLAN MT
WLAN
MT
Figure 3.4: User movement and registration procedure
42
3.2. Proposed Scheme 3. When a MT resides in a LA: While residing in a LA, the MT might detect WLAN coverage or prefer to change the primary mode of access to WiMax, HSPA or WLAN. It is possible through deployment of multi-homed clients in which the primary IP address can be changed at any point in time. This only requires an update to the IA and not to the HSS. Multihoming can be made possible through deployment of Stream Control Transmission Protocol (SCTP) as the transport layer protocol. Furthermore, it is possible to consider the case of an SCTP association between the two end points (MT and P-CSCF or IA). Upon the first entry of an MT in a LA, in addition to conveying the location update to the HSS, it is also necessary to partially retrieve the user profile from HSS to determine the access networks that the user is subscribed to. A Policy Decision Function (PDF) is incorporated in P-CSCF and a Policy Enforcement (PE) module can also be incorporated in Gateway GPRS Support Node (GGSN). Policy enforcement decisions are conveyed to PE by PDF. PE and PDF can communicate in two possible modes of communication namely push mode and pull mode. In the push mode, communication is initiated by PDF and the decision is sent to PE module. In the pull mode communication is initiated by the PE module where a decision is requested[26]. The subscriber information about permitted access network types is cached to PDF of P-CSCF by HSS. Once the user enters a LA, it might need to authenticate itself to be granted access to its preferred available network. It therefore requires the PE module to operate in the pull mode and initiate a decision request from PDF. In this manner, the IA shall be informed by the P-CSCF about the current access mode through which the user can be reached. In the context of access mode permissions, in our proposed architecture the following two approaches can be realized: The access modes that a subscriber is allowed to connect can be conveyed to the IA and cached upon the first entry of an MT in a LA. Yet, it might impose some unnecessary load on IA that requires more processing power and storage. It might not be necessary for the IA to know about all subscribed access networks that a user is permitted to connect because in some LAs chances are that some coverage such as WLAN or WiMax are not available. On the other hand, the IA can pull the decision from the PDF of the P-CSCF upon the first occurrence of a user attempt to connect to a different Access Network (AN). This would happen if the UE detects such coverage and the user preferences allow initiation of access grant request. This approach is preferred because it imposes less amount of load on the IA node. To deploy this strategy we demonstrate a sample procedure for a scenario as described below. A subscriber is roaming within a LA where HSPA, WiMax and WLAN coverage are available. Upon its first entry to the LA, the user has registered with HSPA network. The user has subscribed to all ANs and should be granted access to any of them upon its request and availability of resources. Currently, the IA has a record of this user to be accessible via HSPA. As the MT moves, it initiates a request to GGSN/PDG (Packet Data Gateway) to be connected to a WLAN AN. As the IA does not have any previous record of this user to be connected to a WLAN AN, it will initiate a request to P-CSCF. The PDF would make a decision and query HSS if needed. Should the access permission be granted, the PDF will then push the decision to the PE module in the GGSN. In this scenario, PDF and PE communicate in push mode. If successful, the current point of attachment would be cached at the IA. Figure 3.5 demonstrates the above mentioned procedure. 43
3.2. Proposed Scheme
Figure 3.5: Communication of policy decisions for access mode grant permission
44
3.2. Proposed Scheme
IA @ P-CSCF
GGSN /PDG
Access Mode Change
Change of Access Mode
ACK
Figure 3.6: Signaling diagram for change of access mode under the proposed method
3.2.2
Change of Access Mode under the Proposed Method
As shown in figure 3.6 the procedure of notifying the core network of changes in access mode does not require HSS query but only an update to the IA in P-CSCF. As an example we consider the case where the MT moves to a new LA where HSPA and WLAN coverages coexist. As shown in figure 3.7, the difference between the two tasks of location update and change of access mode is that querying the HSS is eliminated in the later.
3.2.3
Implications on Access Mode changes for WLAN Re-Authentication
In a 3G-WLAN interworking environment, frequent changes of access mode to WLAN and within different Access Points (AP)s in WLAN hotspots, creates the problem of re-authentication in that the UE should be authenticated by the Home Authentication, Authorization and Accounting HAAA, HLR and HSS servers at the core network.3GPP has adopted EAP-Authentication and Key Agreement (EAPAKA) is the authentication method to authenticate a WLAN-UE in a 3G-WLAN interworking environment. This method is based on pre-shared secrets held by the WLAN-UE and the 3G home network. EAP-AKA supports two types of authentication, full authentication and fast re-authentication [40]. To secure the architecture against vulnerabilities, a localized fast re-authentication method is proposed in [41]. In this method the need to re-authenticate through HAAA is eliminated by delegating the WLAN AAA (WAAA) to carry the task of WLAN UE
45
3.2. Proposed Scheme
MT
RAN / WLAN AP
Register Location Update
GGSN / PDG
SGSN
RegisterLocation Update
RegisterLocation Update
IA @ P-CSCF
RegisterLocation Update
Registration
I-CSCF
S-CSCF
RegisterDNS Location Query Update
Change of Location Area (Location Update)
Query HSS
Register
ACK ACK ACK
ACK
ACK
ACK
Access Mode Change Change of Access Mode
Access Mode Change
Access Mode Change
Access Mode Change
ACK ACK
ACK
ACK
Figure 3.7: Signaling diagram for location update and access mode update in HSPA and WLAN
46
3.3. Performance Evaluation authentication locally.
3.3
Performance Evaluation
In this section we show the performance evaluation of the proposed model in terms of cost of signaling and delay as a result of access mode update. For the purpose of analysis in this work, we make the assumption that upon the first entry of a user in a LA, if authentication required by the network, the subscriber‘s profile information will be retrieved from HSS to determine the access networks that the user has subscribed.
3.3.1
Analysis of Signaling Cost
In the following, we model the cost of querying the IMS core and HSS based on the probability of user mobility within a LA and moving to an adjacent or different LA. In this analysis we only take the forced handovers into account (e.g. vertical handover from WLAN to HSPA when a user moves out of WLAN coverage to an area where WLAN coverage does not exist anymore). For the purpose of cost analysis we assume that LAs do not share any coverage (i.e there is no WLAN hotspot that crosses the boundary of a LA). We furthermore assume one HSS physical node per LA. We start the analysis by presenting the cost analysis for a typical base scenario. In a typical base scenario, all changes such as location change and access mode change will be conveyed to the HSS. Figure 3.8 shows the signaling diagram. The signaling cost of registration and location update, CReg in the typical base model is: CReg = 2 × (CU E−AN + CAN −SGSN + CSGSN −GGSN + CGGSN −P CSCF + CHSS−access )
(3.1)
Where Cx−y represents transmission costs between network element x and y. We denote CHSS−access as the cost of accessing the HSS through IMS. This would involve communicating with I-CSCF, S-CSCF, SLF and other core elements. Total signaling cost as a result of user mobility within a LA and movement to adjacent LAs can be written as:
CTbase = CReg × P r[Intra − LA − mobility] + CReg × P r[Inter − LA − mobility]
(3.2)
Where Pr[Inter-LA-mobility] and Pr[Intra-LA-mobility] are probabilities of moving outside a LA and within a LA respectively at a certain period of time given that the MT is moving and are explained in the following subsections.
3.3.2
Signaling Cost of Registration & Location Update
The following analysis is based on our proposed architecture and signaling scheme. Figure 3.7 shows the signaling diagrams. In this analysis we consider the signaling cost of location registration when a MT moves to a different LA. We define CReg as the cost of registration and location update. Cost
47
3.3. Performance Evaluation of location registration in a new LA, (CReg ) for the cases that the MT enters a HSPA coverage and the case that the MT enters a WLAN hotspot, can be written as:
hspa CReg = 2 × (CU E−RAN + CRAN −SGSN + CSGSN −GGSN + CGGSN −P CSCF + CHSS−access ) (3.3)
wlan CReg = 2 × (CU E−AP + CAP −SGSN + CSGSN −GGSN + CGGSN −P CSCF + CHSS−access )
(3.4)
An MT can move within a LA when it is active (whether it is on a call or not) or move outside a LA to an adjacent LA to LAi or a WLAN hotspot or a WiMax coverage within LAi at a certain time period. We define P r[Inter − LA − mobility] as the probability of an MT moving to an adjacent LA of LAi . We consider the whole system as M m LAs that consist of HSPA, WiMax and WLAN m coverage. Let Lm i be the set of LAs adjacent to LAi , GPi be the set of HSPA coverage inside LAi ,
W Mim be the set of WiMax coverage inside LAi and W Lm i be the set of WLAN hotspots inside the LAi . Let GPkadj be the set of HSPA coverage areas adjacent to HSPA coverage k, W Mkadj be the set of WiMax coverage areas adjacent to WiMax coverage k and W Ladj k be the set of WLAN hotspots m adjacent to hotspot k [42]. For instance in figure 3.9, M m = 9, Lm 6 = {1, 5, 7, 2}, GP8 = {1, 2, 3, 4},
W Lm 8 = {1, 2, 3, 4, 5, 6, 7}. The probability that an MT moves to an adjacent LA of LAi is:
P r[Inter − LA − mobility] =
X j∈Lm i
pL ij (
X
Pjlα +
l∈GP m i
X
β Pjm +
m∈W M m i
X n inW Lm i
γ Pjn )
(3.5)
α where pL ij is the probability of moving from LAi to LAj and Pjl is the probability of entering β in HSPA coverage l inside LAj , Pjm is the probability of entering in Wimax coverage m inside γ LAj and Pjn is the probability of entering in WLAN hotspot n inside LAj . For simplicity of our
analysis however we consider equal cost of location update for HSPA, WiMax and WLAN and we set it to be CReg . i.e. hspa wlan wimax ≈ CReg ≈ CReg CReg ≈ CReg
(3.6)
Therefore, the total signaling cost for location update is as follows:
s CReg = P r[Inter − LA − mobility] × CReg
3.3.3
(3.7)
Signaling Cost of Access Mode Update
Now we consider the case where the MT moves only within a LA but as it moves it might cross the boundaries of different coverage areas. For example it might perform a horizontal handover from WLAN hotspot i to an adjacent WLAN hotspot or perform a forced vertical handover from a WLAN hotspot to a WiMax or HSPA connection. In an overlapped network setting, forced handovers are: WLAN → WiMax, WiMax → HSPA or WLAN → HSPA. The cost of vertical
48
3.3. Performance Evaluation
MT
AN
Register Location Update
SGSN
Register Location Update
GGSN /PDG
Register Location Update
P-CSCF
I-CSCF
S-CSCF
Register Location Update
Registration
DNS Query
Change of Location Area (Location Update)
Register Location Update Query HSS
Register ACK ACK
ACK
ACK ACK ACK
Change of Access Mode
Access Mode Change
Access Mode Change
Access Mode Change
Access Mode Change
Access Mode Access Mode Change Change Update HSS
ACK
ACK
ACK
ACK
ACK ACK
Figure 3.8: Sinaling diagram for a typical base scenario
49
3.3. Performance Evaluation handover within an overlaying network of a LA is: Cho = 2 × (CU E−SGSN + CSGSN −GGSN + CGGSN −P CSCF )
(3.8)
We now define the probability of mobility of an MT which is the probability of an MT moving to adjacent coverage or hotspots within a LA or perform vertical handovers within a LA. The probability of mobility, P r[Intra − LA − mobility] at a certain period of time given that the user is moving is therefore: P r[Intra − LA − mobility] =
X
X
GP Pki +
i∈GP ik
WM Pkl +
l∈W M m k
X
WL Pkn
n∈W Lm k
(3.9)
+P r[F orced − V ertical − Handover] GP is the probability of moving (performing a horizontal handover) from HSPA coverage where Pki W M is the probability of moving from WiMax coverage k to l and P W L is the probability k to i, Pkl kn
of moving from WLAN hotspot k to n. P r[F orced − V ertical − Handover] can also be defined as follows: P r[F orced − V ertical − Handover] =
X
wimax→hspa Pkl
l∈GP m k
+
X
wlan→hspa + Pkj
j∈GP m k
X
wlan→wimax Pkn
(3.10)
n∈W M m k
where PijX→Y is the probability of VHO from X coverage i to Y coverage j (e.g. X and Y can be WiMax, HSPA or WLAN). We define the signaling cost as a result of an MT moving within a LA as follows: s Caccess−M ode−update = Cho × P r[Intra − LA − mobility]
(3.11)
Total cost of signaling in the proposed method is therefore: s s CTproposed = CReg + Caccess−M ode−update
(3.12)
s To obtain the signaling cost savings by using the proposed model, we consider CReg = Cdb−access .
We can therefore derive the savings in cost of signaling as follows: ∆Csav ≈ 2 × CHSS−access × P r[Intra − LA − mobility]
3.3.4
(3.13)
Analysis of Delay
In this section we compare the delay performance of a base case scenario and the proposed method. This analysis is based on figures 3.2, 3.3, and 3.6. To demonstrate the improvement in delay performance, we make the following assumptions: • we do not consider the queuing delay
50
3.3. Performance Evaluation • processing delay is not extracted from other delays but it is included in link transmission and query delays. • to focus on the delay within the IMS core network, we consider only the delays incurred from GGSN to IMS. • PDF entity has the record of all possible access modes that a user is subscribed to (i.e. the MT has already attempted all possible modes of access at least once) We denote the registration delay which is similar for base scenario and the proposed method, DReg , as follows: Query Link DReg = 2 × (DPLink DG−P CSCF + DDN S + DP CSCF −ICSCF Query Link +DHSS + DICSCF −SCSCF )
(3.14)
Link is the transmission link delay between network element x and y. D Query and D Query Where Dx−y DN S HSS
are delays of querying HSS and DNS respectively. Delay of change in access mode in the base scenario is as follows: Link Base = 2 × (DPLink Daccess−mode−update DG−P CSCF + DP CSCF −ICSCF + Query Link DICSCF −SCSCF + DHSS )
(3.15)
Delay of change in access mode in the proposed method is: P roposed internal−sip−transport = 2 × DPLink Daccess−mode−update DG−P CSCF + Dproc
(3.16)
internal−sip−transport is the delay of IA interaction between SIP servers running on the PWhere Dproc
CSCF host which consists of processing delays and interaction delays of SIP stack transport layer and upper layers (e.g. SIP Application). Assuming that the user is moving either within a LA or moving to an adjacent LA, we let the Base and D P roposed ) to be: average delay for base scenario and proposed method (DAvg Avg Base DAvg = DReg × P r[Inter − LA − M obility]+ Base Daccess−mode−update × P r[Intra − LA − M obility]
(3.17)
And P roposed DAvg = DReg × P r[Inter − LA − M obility]+ P roposed Daccess−mode−update × P r[Intra − LA − M obility]
3.3.5
(3.18)
Numerical Results
In this section we demonstrate some numerical results to provide a comparison between a typical base scenario and our proposed model in terms of signaling cost. As mentioned earlier we decomposed the probability of mobility into the probability of user mobility within an overlaying LA and outside a LA. In this analysis we considered the ratio of movements within a LA to the movements 51
3.3. Performance Evaluation
Table 3.2: Parameters for cost of signaling vs. probability of mobility Parameters Probabilty of Mobility Probability of LA changes Wireless link costs CHSS−access
Value 0.1 - 1 0.2 1 (unit of cost) 3 (units of cost)
Table 3.3: Cost savings as a result of change in ratio of Intra-LA to Inter-LA mobility Parameters Probability of mobility Probability of LA changes
Value 0.2, 0.5, 0.9 0.1 - 1
out of a LA to be 0.8 : 0.2 with the assumption that a movement occurs. Each of the elements in the cost of registration is considered as one unit of cost except the cost of HSS access that is three units of cost. The ratio of mobility within a LA and between different LAs depends on the size of the location area and with the assumption of one physical HSS node per LA, it also indicates the number of HSS physical nodes. We denote P r[mobility] as the probability that a user is mobile and γ as the ratio of mobility that indicates the portion of roaming within a LA with respect to the roaming to a different LA. We denote the savings in cost ∆Csav as follows: ∆Csav = CTproposed − CTbase
(3.19)
The graph of figure 3.11 shows the comparison of signaling cost between a typical base scenario and our proposed scheme. It demonstrates a better performance as the probability of user mobility increases. In other words, it shows more signaling cost savings. This graph is based on the assumption that 80% of user movements are within the LA and 20% chance of crossing a LA boundary that leads a location update. Table 3.2 shows values for the parameters that are used for this graph. The result of figure 3.12 shows that as we increase the number of HSS physical nodes or decrease the size of a LA, there are less signaling cost savings. We have demonstrated this scenario for the cases where probability of mobility is 0.9, 0.5 and 0.2. The numerical values are listed in table 3.3. As shown in the graph as the probability of changing the LA increases, there are less cost savings in signaling traffic. The extreme point in the graph can be interpreted as the situation similar to the cellular networks with only HSPA coverage that users roam within different registration areas. On the other hand the proposed scheme performs better in the areas that there are more heterogeneity of networks. In other words, if there is only single HSPA coverage in some LAs, deployment of an IA under the proposed scheme, would result in an indifferent cost and delay performance as in a typical base scenario. We next compare the delay performance of the proposed method and the typical base scenario presented earlier. The latency parameters are listed in table 3.4. The values represent some typical 52
3.3. Performance Evaluation
LA 8 WiMax 1
LA 3 LA i
WLAN 1
LA 7
LA 1 WLAN 3
WLAN 2 LA 4
WiMax 4
HSPA / GPRS 1
WLAN 4
WLAN 7
LA 5
LA 2
HSPA / GPRS 2
HSPA / GPRS 4 WLAN 5
LA 6
WiMax 3
WiMax 2
WLAN 6 HSPA / GPRS 3
Figure 3.9: Location areas and possible configuration of coverage within a LA
Table 3.4: Latency Parameters for transmission and query Parameters DNS Query HSS Query P-CSCF to I-CSCF I-CSCF to S-CSCF SIP interactions within co-locates SIP entities
Typical value 100 ms 300 ms 10 ms 10 ms 2 ms
53
3.4. Conclusion
Pr[mobility] vs. Average Delay
900 800
Delay (ms)
700 600 500 400 300 Base Scenario Proposed Model
200 100 0 0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Pr[Intra-LA-Mobility]
Figure 3.10: Average latency with changes in Pr[Inter-LA-Mobility] measured parameters based on SIP transmission [43]. As shown in graph of figure 3.10 delay performance has been improved. This improvement is more significant for the situations where the mobile user is mainly roaming within a LA.
3.4
Conclusion
In transitioning to the future generation of wireless networks, convergence of IP based core networks with heterogeneous wireless access networks is inevitable. IMS aims at providing a standardized solution for multimedia services in an access agnostic manner within a unified framework. In this chapter we looked at the database location management for interworking of wireless heterogeneous networks with IMS. HSS as a key component of IMS will take over the task of HLRs with additional functionalities. The industry‘s consensus is towards deployment of a unified repository database that holds the subscriber profile information across different domains. One of the issues in mobility of users in heterogeneous environments is through which network should the MT update its location and where should the location information and access mode information of subscribers be stored. Location update through all possible access modes to separate databases for each access network can cause excessive delay in call delivery and service time to the users while creating data duplication
54
3.4. Conclusion
Signaling Cost (units of cost)
14 Base Scenario
12
Proposed framework
10 8 6 4 2 0 0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Pr[Mobility]
Figure 3.11: Signaling cost vs. probability of mobility
55
3.4. Conclusion
6
Signaling Cost Saving
5
Pr[mobility]=0.9 Pr[mobility]=0.5
4
Pr[mobility]=0.2
3 2 1 0 0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Probabilty of Roaming within a Location Area
Figure 3.12: Signaling cost saving vs. probability of LA change and discrepancy. It also causes excessive signaling cost on the network. For the future wireless communication systems where multimodal devices deployed, having a single logical database that applies the subscribers profiles for all CS and Packet Switched (PS) domains would be advantageous. In this chapter we presented our proposed scheme that was motivated by the idea of a unified logical database. The problem of frequent changes of access mode as a result of high mobility is explained. To reduce the signaling cost as a result of frequent changes of access mode, we proposed the use of an IA to cache the location and mode of access for a user. We demonstrated the performance improvements for situations where user is traveling within and outside a LA. Furthermore, the case for access grant permission and policy decision procedure is illustrated.
56
Chapter 4
Call Delivery over IP-Based Heterogeneous Wireless Access Networks 4 4.1
Introduction
In this chapter the problem of call delivery in a heterogeneous access environment is discussed. An efficient call delivery scheme can make a significant contribution to delay performance and signaling costs. On the other hand, it is important to reduce the signaling cost of calls that would potentially fail. In the future communication systems, millions of calls are generated per second. However, only about 70% of the calls are responded to normally [44]. Some of the causes of a call failure can be that the users are busy, the users do not answer, or the calls get redirected and few calls can be failed due to user-selected errors. Mobility management is an important issue in the future wireless communication systems. It enables delivery of incoming calls, performing handovers and tracking the location of users. In this chapter, we address the problem of call delivery in heterogeneous wireless access networks. The proposed method of call delivery and signaling scheme is based on reverse virtual call setup. We also propose the deployment of an Application Server (For this purpose, it is called Call DeliveryAS, CDAS) to act as a proxy, originating or terminating agent.This method is aimed at addressing the issue of call delivery in a heterogeneous environment that interworks with IMS.It is based on partial retrieval of HSS to locate the user by existing paging network infrastructure and performing a reverse virtual call setup. One of the advantages of our proposed method is that the cost of call deliveries that would potentially fail can be reduced. Another advantage is that the overall delay for both successful and unsuccessful calls can be reduced significantly. This will lead to lesser waste of network resources as well as better user experience through lower service time. This chapter is organized as follows. In section 4.2 we briefly survey the previous work on call delivery in cellular systems explaining the idea of reverse virtual call setup and deployment of a call delivery application on SDP. In section 4.3 the problem of call delivery is explained and we introduce the interworking model that our proposal is based on. Section 4.4 consists of our proposed solution and illustrating possible scenarios of call/multimedia call setup. Section 4.5 provides performance evaluation of our proposed method with some graphical results. In this section we show our analytical model for cost and delay and compare our scheme with the base scenario for successful 4
A version of this chapter has been published. P. TalebiFard and V.C.M. Leung, ”Efficient Multimedia Call delivery over IP-Based Heterogeneous Wireless Access Networks”, Proc. ACM MobiWAC, Vancouver, BC, Oct. 2008.
57
4.2. Call Delivery and Mobility Management
CSCF
CSCF
P-CSCF
MGCF
MGW
Figure 4.1: Network configuration for the base scenario calls and failed calls. Section 4.7 addresses some implementation considerations regarding network selection and multimedia call setup. Section 4.8 concludes this chapter.
4.2
Call Delivery and Mobility Management
In this section we briefly explain call delivery and mobility management. We summarize the existing approach of originating and terminating a call. Figure 4.1 shows a typical network configuration that is used for the purpose of discussion in this section. For this base scenario, the two parties may be registered in the same IMS network or different IMS networks. For completeness, we demonstrate the case where the two IMS-capable Mobile Terminals (MT) use different IMS networks.
4.2.1
Call Delivery
Call delivery is one of the major tasks in location management. Location management involves location registration and call delivery. The call delivery procedure consists of locating a MT based on the available information in the database when a call to an MT is initiated. For the purpose of location management, the concept of Home Location Register (HLR) and Visitor Location Register (VLR) is used in the Third Generation (3G) networks. HLR and VLR are used to store the location information of MTs. In the literature, several schemes are proposed that are based on centralized database architectures or distributed database architectures [29][30]. 58
4.2. Call Delivery and Mobility Management In this section we show a typical registration method that is currently used, along with the base case scenarios for originating and terminating a call. Figure 4.2 shows a typical procedure for registering a user that attempts a registration in IMS. When the initial registration request reaches the P-CSCF it performs a DNS query to determine the I-CSCF for that domain. Once the I-CSCF is determined the request will be routed to the I-CSCF and HSS will be queried to determine the capable S-CSCF. The MT will then be registered to the capable S-CSCF and all the signaling to/from the MT will traverse the same path after registration. Initiating and terminating a call in a typical base scenario for the case that both ends support IMS and are registered in IMS involves the following as shown in figure 4.3 and figure 4.4: 1. Upon arrival of the call initiation request by P-CSCF the SIP request will be forwarded to the I-CSCF and S-CSCF. 2. The user profile will be looked up by a HSS query for Authentication, Authorization and Accounting (AAA) 3. The call initiation request would then be forwarded to the P-CSCF if the called party is registered in the same IMS network or the CSCF of the external IMS network if the called party is located in a different IMS. Terminating a call is performed with a similar procedure as follows: 1. When the I-CSCF of the called party’s IMS receives the session setup request, the request is forwarded to the capable S-CSCF. 2. The S-CSCF will query HSS for AAA of the called party to obtain subscription information. 3. The P-CSCF hosting the called user is located and the request is forwarded to the P-CSCF. Call delivery procedure for the example of figure 4.1 where one end is connected to P-CSCF that hosts a General Packet Radio Service (GPRS) network and the other end is a public switched telephone network (PSTN) is explained below as shown in figure 4.5. 1. The calling UE initiates a call by sending a call request to the Media Gateway Control Function (MGCF). 2. The MGCF forwards the request to the CSCF. I-CSCF would determine whether the request should be forwarded to any application server. This is done by querying HSS to retrieve the user profile for authentication and authorization purposes. 3. If the called party is located in an external IMS network, the request is forwarded to the CSCF of the terminating IMS. Otherwise, the P-CSCF to which the call shall be forwarded will be determined. 4. Upon arrival of the call request to the terminating IMS network at I-CSCF, the S-CSCF in charge will be determined. I-CSCF would also query the HSS for Authentication, Authorization and Accounting (AAA) of the called party as well as a user profile vector containing the possible application servers involved and/or the location of P-CSCF for that user. 59
4.2. Call Delivery and Mobility Management
P-CSCF
I-CSCF
HSS
S-CSCF
DNS Query
HSS Query for AAA and locating the capable S-CSCF
Register
Ack Ack
Figure 4.2: User registration in a typical base scenario 60
4.2. Call Delivery and Mobility Management
P-CSCF
I-CSCF
S-CSCF
HSS
CSCF
HSS Query - AAA
Initiating the call to the callee
Figure 4.3: Originating a call/session in a typical base scenario
61
4.2. Call Delivery and Mobility Management
P-CSCF
HSS
S-CSCF
I-CSCF
CSCF
AAA
Paging the MT
Figure 4.4: Terminating a call/session in a typical base scenario
62
4.2. Call Delivery and Mobility Management
IMS1
UE
PSTN
MGCF
I/S CSCF
IMS2
HSS
I/S CSCF
HSS
P-CSCF
UE
AAA
AAA
Figure 4.5: Signaling diagram for delivering a call in the base scenario for the example of GPRSPSTN 5. For a multimedia session establishment, the Policy Decision Function (PDF) in P-CSCF should retrieve policies from the policy repository database that leads to querying the HSS for retrieving user profile related information and negotiation of QoS. 6. Policy decision should be conveyed to the Policy Enforcement (PE) module in Gateway GPRS Support Node (GGSN). Communication between PDF and PE module can be done in push mode or pull mode [45].
4.2.2
Reverse Virtual Call Setup
A call delivery scheme based on RVC is proposed in [46] for cellular systems. The motivation of this work was to minimize the cost of failed call attempts. The presented algorithm is for a mobile terminated calls initiated from a fixed phone. In this work a proxy switch is introduced that has the capability to query the HLR. Instead of setting up the call through the terminating switch, the global paging network is used to page the MT. If the callee decides to accept the call, the MT will initiate a call setup to the caller via the proxy switch. Network resources therefore will not be allocated prior to acceptance of call by the callee. In this method, the user is paged with a signal that contains both end user IDs.
63
4.2. Call Delivery and Mobility Management
4.2.3
Call Delivery Application
The role of Application Servers (AS) in IMS is to host and execute services on behalf of an end system depending on the application. The AS can interface with CSCF using the SIP. AS can be located either in the home network or in an external network. If located in the home network, it can interact with HSS via Sh, ISC or MAP interface. There are different types of AS. SIP AS is in charge of executing IP multimedia services based on SIP. Open Service Access Service Capability Server (OSA-SCS) is an intermediate application server that interfaces with S-CSCF on one side and OSA application server and OSA Application Programming Interface (API) on the other side. IP Multimedia Service Switching Function (IMSSF) is an AS that interfaces with Customized Applications for Mobile network Enhanced Logic (CAMEL) application servers that were developed for GSM in IMS, using CAMEL Application Part (CAP). IM-SSF allows a GSM Service Control Function (gsmSCF) to control an IMS session. Furthermore, an AS that we call it Call Delivery Application Server (CDAS) can be developed to handle the task of call delivery. The purpose of this AS is to decide how to handle call delivery based on the logic that determines whether a subscriber is registered in IMS or might be residing in a network that supports IMS as well as handling calls to existing circuit switched networks. Like other application servers, CDAS can also interact with CSCF and HSS. One possible way of interacting with GSM networks is interfacing with IM-SSF through CSCF. We will explain the role CDAS in details in the call delivery procedure presented in section 4.4. CDAS in other words can be considered as a SIP entity co-located with one of the CSCFs within the core network. Towards deployment of the above idea, one possible approach is to have this SIP server co-located with one of the S-CSCFs capable of serving that request. It leads to allocating different S-CSCFs to be co-located with a CDAS. For example, it is possible to have different CDAS for video call requests, voice calls and etc. Having two servers running on the same host brings the following advantages of less processing time and more reliability. The servers within a common host can interact with other entities within S-CSCF directly through internal SIP transport since TCP/UDP, IP, MAC and Physical layers are bypassed. Call delivery applications running on a SIP server can interface with SDP (Parlay X module within SDP is in charge of this interface). one of the major function performed by the SDP consists of providing charging information and billing records to a charging enabler. All applications traverse the SDP; therefore the SDP is optimally located to monitor and collect charging information for all applications. This is an advantage for it facilitates the billing settlement. SDP on the other hand can facilitate the interconnecting of other internal applications. Once the User Equipment (UE) gains access to the IP Connectivity Access Network, it registers with the IMS network by sending its SIP registration request to the P-CSCF, which was assigned to the UE by the network (DHCP, PDP context activation). The P-CSCF forwards all messages it receives from the UE to the I-CSCF which serves at the user‘s home domain. The I-CSCF queries the HSS and selects a S-CSCF based on the services the user has subscribed to, and forwards the registration message to the selected capable S-CSCF. The S-CSCF is the service access point acting as a service dispatcher within the IMS network. It authenticates the user and registers it with the IMS network.
64
4.3.
4.3
System Architecture & Problem Formulation
System Architecture & Problem Formulation
One of the features that distinguishes voice calls from multimedia calls is that to set up a multimedia call the knowledge about the access mode for the called party is essential in order to guarantee a certain level of Quality of Service (QoS). On the other hand, in a multimedia call it is possible that the users establish multiple sessions such as voice, video or text. Existing solutions only address the case for delivering a voice call to a cellular subscriber. The case for the ratio of failed calls for voice-only sessions differs from the one for multimedia sessions due to the nature of the call and dependency on the available bandwidth. A multimedia session may demand excessive bandwidth that may require the user to have a Wireless Local Area Network (WLAN) connection. In such a case, the call will fail if the user is in a visiting network where WLAN access is not possible. The previously mentioned procedures of call delivery suffers from unnecessary usage of network resources for call attempts that fail which adds to excessive signaling delays and queuing costs. Since the percentage of failed calls are often high (30%-40%, it is therefore of great advantage to minimize the cost of calls that may fail. For multimedia calls such as video call, session initiation involves QoS negotiation and retrievals of user profile information for subscription details. It therefore, imposes a huge signaling traffic load on the core network. The system architecture that our proposed solution will be based on is shown in figure 4.6. In this figure our assumption for a possible interworking architecture is demonstrated. There is a possibility that each home or visited network does or does not support IMS. As shown in the figure, the visited network that does not support IMS, can be reached via the GPRS network. IM-SSF and gsmSCF together enable handling of IMS session control .Home and visited networks that support IMS can be located and reached via the P-CSCF. Considering the above mentioned interworking architecture, it is possible to realize two different roles for IMS. IMS can be configured to act as the main routing system where all calls to IMS, cellular or CS users are initiated via IMS and IMS makes the routing decision. IMS on the other hand, can act as a serving/visiting entity like MSC in the 3GPP. In this model, upon initiation of a call from a 3GPP network, IMS or one of its application servers will be queried to retrieve a routing directory number. System configuration for the cases where the home or visiting network supports IMS and does not support IMS differs in terms of the location of Proxy CSCF (P-CSCF). As mentioned earlier, IMS can be configured to act as a passive entity or an active entity. In our proposed method of call delivery, we suggest a strategy where IMS acts as the main routing system. In this manner, all calls are initially routed to IMS and the delivery decision is made by AS in charge based on whether or not the called party is registered in IMS or supports IMS.
4.4
Proposed Solution
Features of the proposed method are as follows: • Less query of database that contributes to lower processing and queuing costs. • It is suitable for interworking of heterogeneous networks that requires deployment of multi65
4.4. Proposed Solution
AS
HSS S-CSCF
IM-SSF
I-CSCF
MGCF
P-CSCF
P-CSCF
PSTN
CAP
Go
HLR/ VLR
MAP
gsmSCF
SGSN
Go
HLR/ VLR SGSN
GGSN
GPRS
SGSN
GGSN
Visited Network: Supports IMS
WLAN
WLAN
WLAN
WLAN
WLAN
WLAN
GPRS
GPRS WLAN
GPRS
WLAN
WLAN
WLAN
Visited Network: Does not support IMS
WLAN
Home Network: Supports IMS
GPRS
WLAN
WLAN
GPRS GPRS
Figure 4.6: Interworking model
66
4.4. Proposed Solution modal devices. • Suitable for future communication system requirements where users can setup a session with an application or vice versa. • MT can choose to return (reverse) the call with the preferred mode of access that is determined based on available network resources, subscription, user preferences and etc. • Lower signaling costs for successful and unsuccessful calls. • Better delay performance and end user experience • Integration of call delivery and session setup services within SDP that facilitates common billing. The basic idea behind our proposed method is that HSS will not be accessed to retrieve for user subscription information unless the session is going to be set up successfully. In this method, the caller party requests the session initiation/setup via IMS. Once the call delivery applications in charge are determined, the request will be handled by the AS. The AS will assign a temporary number or SIP address for this session and initiates the procedure of paging the called party. Once the potential location of the user is determined, the MT is paged using the paging subsystems. If the user is available to answer the call, the session response is reversed towards the AS that acts as an interim entity on behalf of the caller who initiated the session. This method in essence is a symmetric procedure and can be easily scaled for the session establishments triggered by application servers. It also implies that application servers should be capable of initiating a call and/or interacting with other application servers. Practical example of this is that in the future users can call an application and/or an application can call a user. The network setting that we consider in this work for wireless coverage area is a fully overlapped multi-tier setting in which GPRS/HSPA is the first tier and WLAN is the second (lower) tier. Figure 4.6 demonstrates our assumption. The reason for choosing this is that it is a more challenging and a more realistic case. To demonstrate the session initiation and termination, in this work we assume that MT is already registered with all possible access modes that the device supports and activated on the device. We furthermore, consider using the existing paging resources and infrastructure of cellular systems. It is also possible to develop a separate network optimized for paging purposes. It is important to note that paging does not lead to completing a call. As a result, the existing paging infrastructure can be reused and the alternative approach to develop a separate network can be a topic for the future research. A new registration procedure is also proposed. This method is mainly based on the idea of caching the initial Filtering Criteria(iFC) in the S-CSCF. iFC consists of basic information about user subscription. In this section, detailed explanation of the proposed method is presented. We show the user registration, initiating a call and terminating a call under the proposed method. For this purpose we only show the main components in the core network. We furthermore, consider the case for
67
4.4. Proposed Solution
P-CSCF
I-CSCF
HSS
S-CSCF CDAS
DNS Query
HSS Query: Determine the capable S-CSCF
Register
Cache the Initial Filter Criteria
Ack Ack
Figure 4.7: User registration in the proposed model delivering a multimedia call/session from PSTN to WLAN as well as from PSTN to GPRS. We furthermore address a more realistic case for a multimedia call delivery procedure.
4.4.1
Registration
As shown in figure 4.7 , MT registration under the proposed method is as follows: 1. When the registration request reaches the P-CSCF, DNS will be queried to determine the I-CSCF. 2. In the I-CSCF HSS will be queried to find the capable S-CSCFs. (There might be more than one capable S-CSCF based on the possible services a user is subscribed to) 3. The user will be registered to the S-CSCF and initial filtering criteria as explained earlier will be cached in the S-CSCF in which the user is registered.
4.4.2
Initiating a Call
As shown in figure 4.8 the caller party is accessing IMS through a P-CSCF in the edge of the network. We assume that the call initiation request has reached the P-CSCF. The session initiation request in the caller party originating network is as follows: 68
4.4. Proposed Solution
P-CSCF
I-CSCF
I-CSCF
S-CSCF CDAS
Processing based on the Initial Filter Criteria
Figure 4.8: Call initiation in the originating network for the proposed model 1. Upon receiving a call initiation request by P-CSCF, the request will be routed to the previously assigned I-CSCF and capable S-CSCF. 2. User will be authenticated based on the available information that was previously cached in S-CSCF as the initial filtering criteria. 3. Call delivery application that is running on the same host as S-CSCF will act a third party to initiate a call request towards the callee. It assigns a temporary SIP address and forwards the request to the CSCF of called user’s network.
4.4.3
Terminating a call
We demonstrate the procedure of terminating a call to the P-CSCF or the edge of the core network. Once the P-CSCF receives the paging request, the MT may be paged using the existing paging infrastructure for that overlaying network. As shown in figure 4.9, the procedure is as follows: 1. The request is received by the terminating network capable S-CSCF and the request is transmitted locally to the call delivery application. 2. SIP request will be routed to the I-CSCF. 3. The P-CSCF associated with the MT is located and the request is forwarded to that P-CSCF. 69
4.4. Proposed Solution
P-CSCF PDF
CDAS
S-CSCF
I-CSCF
HSS
CDAS
S-CSCF
Processing
Upon availability of the callee: Receiving the Call Reversal from MT
Query HSS for Access Accounting Authentication AAA
Binding
Figure 4.9: Call termination in the called party terminating network for the proposed model 4. The MT will be paged using the paging infrastructure of the network the user is connected. 5. If the user is available to respond the call, the call delivery application will be notified and binding procedure will take place. The MT will be reversing the call using the preferred mode of access. In the following we show examples of a case for session setup between a WLAN and PSTN and a case for session setup between GPRS and PSTN.
4.4.4
Call session setup between WLAN & PSTN
In this scenario the CS network connects to the IMS network via MGCF. In this call setup scenario, the AS plays the role of a SIP proxy. It means that from the originating UE point of view it acts as a terminating UA or a redirect server and from the terminating UE point of view, it appears as the originating UA. The CSCF performs a partial query to HSS for access authorization. The request will then be forwarded to the AS in charge (i.e. CDAS). While the request is being processed by the AS, HSS will be queried for additional information needed for call setup at the binding stage. The CDAS will then redirect the call to the CSCF of the called party. At the terminating IMS network, the CSCF will locate the P-CSCF through which the called party can be accessed and pages the user. In this scenario the user receives the paging via WLAN IP network. The UE will reverse the call presumably through the same access mode to the CDAS. The call will be revered 70
4.5. Performance Analysis and Evaluations
Originating Network
UE
PSTN
MGCF
I/S CSCF
HSS
Terminating Network
CDAS
I/S CSCF
HSS
P-CSCF
WLAN IP NW
UE
Access Authentication*
* Access Authentication involves partial query of HSS only for access. Authorization and accounting will be done concurrently with the CDAS attempt to deliver the call.
Figure 4.10: Signaling flow diagram for call delivery procedure for the example of PSTN to WLAN through IMS via the WLAN IP network and P-CSCF. Only at this stage will the HSS be queried. Figure 4.10 demonstrates this scenario.
4.4.5
Call session setup between GPRS & PSTN
For this scenario we will consider a different situation where the GPRS end user does not support IMS and both end users are connected to the same IMS network. Same procedure will be performed from the CS end point and the task will be forwarded to the AS in charge. In this case, we will consider the situation where the called party is visiting a network that does not support IMS. Upon initiating the call from AS, the P-CSCF for the called party network is located. After paging the user and querying the HLR, the call will be routed to the visiting network Serving GPRS Support Node SGSN via GGSN. Once the call has been initiated the IMS session control will be handled by IM-SSF via gsmSCF. The role of IM-SSF & gsmSCF in this method is in accordance with the 3GPP specifications in document TS23.228 [7].
4.5
Performance Analysis and Evaluations
In this section we present an analytical model for determining signaling cost and delay as a result of transmission link and database queries and compare the proposed scheme with the base scenario.
71
4.5. Performance Analysis and Evaluations Throughout the analysis we do not consider delays and costs of queuing and packet overhead for simplicity.
4.5.1
Cost Analysis
For the purpose of cost analysis we compare the signaling cost of delivering a call in the base case L scenario as shown in section 4.2 and for the proposed method as in section 4.4. Let Cx−y represent
the link transmission cost of wired and wireless link between x and y. We furthermore express the Q cost of querying HSS (CHSS ) for AAA as follows: Q Req Res CHSS = CHSS + CHSS
(4.1)
Req Res are costs of request and response for query of HSS. Where CHSS & CHSS
Considering both originating and terminating networks, the total cost according to the base scenario as shown in figures ?? and ??and ?? is:
Q L L + 2CIM CTBase = 2CPL−I + 2CI−S + CHSS S1−IM S2 Q L L + 2CS−P +2CI−S + CHSS
(4.2)
Assuming equal C L link transmission cost for all links we can simplify the above expression to the following: Q CTBase = 10C L + 2CHSS
(4.3)
For the proposed model the cost of signaling for successful calls and unsuccessful calls are slightly different. In the originating network HSS query will be performed if the required information in given by the iFC is not sufficient for setting up the session later. HSS query will be done concurrently while the request is being processed by the CDAS of the terminating side. In the terminating network, HSS query will be performed after the called party reverses the call to the originating CDAS. Therefore, in our proposed method the cost of signaling for a call that would potentially fail is lower. Cost of signaling for a successful call in the proposed model is as follows:
Q L CTP roposed−Success = 2CPL−I + 2CI−S + CHSS Q L L L +2CIM S1−IM S2 + 2CI−S + CHSS + 2CS−P
(4.4)
For the case of unsuccessful call setup we have: L L L L CTP roposed−f ail = 2CPL−I + 2CI−S + 2CIM S1−IM S2 + 2CI−S + 2CS−P
(4.5)
With the previously mentioned assumption of equal C L link transmission cost for all links we can simplify the above expressions to the following:
72
4.5. Performance Analysis and Evaluations
Q CTP roposed−Success = 10C L + 2CHSS
(4.6)
CTP roposed−f ail = 10C L
(4.7)
We assume that call failures are mainly caused by the called party. Cost savings of traffic and unnecessary querying of HSS can be represented in terms of percentage of calls that would potentially fail. In the following we denote the average cost of signaling in terms of theta, ratio of call that would fail. CTP roposed−Avg = θCTP roposed−f ail + (1 − θ)CTP roposed−Success
(4.8)
For the base case scenario, since the cost of signaling does not differ for the a successful and unsuccessful call, we ddenote CTBase−Avg = CTBase . Total savings in signaling cost (∆C) can be written as: ∆C = CTBase−Avg − CTP roposed−Avg
4.5.2
(4.9)
Delay Analysis
To compare our proposed method with the base scenario in terms of delay, we present a call setup delay analysis and demonstrate the delay for the case of a successful call and a failed call. Let DT be the total delay in the originating network and the terminating network that is a function of Q as link delay and HSS query HSS query and communication link delays. We denote DL and DHSS
delay. We first show the delay for a session setup in the base case scenario according to figures 4.3 and 4.4 the call setup delay can be written as follows:
Q L L DTBase = 2DPL −I + 2DI−S + DHSS + 2DIM S1−IM S2 Q L L + 2DS−P +2DI−S + DHSS
(4.10)
L Where Dx−y is the SIP transmission link delay between x and y. Assuming equal DL link
transmission delay for all SIP entities, we can simplify the above expression to the following: Q DTBase = 10DL + 2DHSS
(4.11)
For the proposed model, total delay for successful calls and unsuccessful calls are slightly different. In the originating network if HSS query is necessary according to the above mentioned criteria, it will be performed concurrently while the request is being processed by the CDAS of the terminating side. HSS query will be done in the terminating network may be performed after the called party reverses the call to the originating CDAS. Therefore, in our proposed method the call setup delay for a call that would potentially fail is lower. Since the call delivery application is assumed to be running on the same host as S-CSCF, we introduce an internal SIP stack delay (α). Call setup delay for a successful call in the proposed model is as follows: 73
4.6. Numerical Results
L DTP roposed−Success = 2DPL −I + 2DI−S + 2α Q L L L +2DIM S1−IM S2 + 2DI−S + DHSS + 2DS−P
(4.12)
For the case of an unsuccessful session setup we have:
L DTP roposed−f ail = 2DPL −I + 2DI−S + 2α L L L +2DIM S1−IM S2 + 2DI−S + 2DS−P
(4.13)
With the assumption of equal DL link transmission delay for all SIP entities, we can simplify the above expressions to the following: Q DTP roposed−Success = 10DL + 2α + DHSS
(4.14)
DTP roposed−f ail = 10DL + 2α
(4.15)
We was mentioned earlier, we assume that call failures are mainly caused by the called party. Average delay of a session setup can be represented in terms of percentage of calls that would potentially fail. In the following we denote the average cost of signaling in terms of theta, ratio of call that would fail. DTP roposed−Avg = θDTP roposed−f ail + (1 − θ)DTP roposed−Success
(4.16)
It can be shown that the time it takes for the caller to be notified of a failed call attempt can be significantly reduced. This means that once the caller is notified of a failed call, the reserved/allocated resources can be released which leads to significant reduction in waste of network resources.
4.6
Numerical Results
In this section we demonstrate some numerical results to provide a comparison between a typical base scenario and our proposed model in terms of signaling cost and delay. We consider a typical practical scenario for the purpose of evaluation. We assume equal link transmission cost of C L for all SIP entities where processing costs are assumed to be negligible for all SIP entities and CSCFs. Cost of HSS query is also assumed to be a constant value taking into account the cost of processing. As shown in table 4.1, C L for all CSCFs are set to be one unit of cost and HSS query is set to be 3 units of cost. We furthermore assume θ = 40% for chance of call delivery failure. Figure 4.11 shows the comparison for cost of call delivery in our proposed method and the base situation as well as potential cost savings as a result of avoiding unnecessary query of HSS. Since the proposed method is intended to avoid unnecessary usage of network resources for calls that fail, we also show the performance of this method when the call delivery failure varies. Graph
74
4.6. Numerical Results
Table 4.1: Signaling cost values (C L & C Q ) for transmission and query Parameters Q CHSS : HSS Query L CP −I : P-CSCF to I-CSCF L : I-CSCF to S-CSCF CI−S IMS1 to IMS2
Signalling cost for failure rate of 40%
2500
Baseline situation Proposed method Cost savings
2000 Total signaling cost
Typical value in (unit of cost) 3 1 1 1
1500
1000
500
0 10
20
30
40
50
60
70
80
90
100 110 120 130 140 150
Number of originating calls
Figure 4.11: Cost of signaling for base situation and proposed method versus the number of originating calls
75
4.6. Numerical Results
Figure 4.12: Savings in cost of signaling for the proposed scheme as the percentage of failed calls increases of figure 4.12 shows that as the ratio of failed calls increases more signaling cost is saved. We furthermore consider a typical practical scenario for the purpose of delay performance evaluation. We assume equal SIP link delay of DL for all SIP entities where processing delays are assumed to be negligible for all SIP entities and CSCFs. Delay of HSS query is also assumed to be a constant value taking into account the processing delay. As shown in table 4.2, DL for all CSCFs are set to be 100 ms and HSS query is set to be 300 ms [43]. We furthermore assume θ = 40% for chance of call delivery failure. Figure 4.13 compares the delay for a successful call delivery and an unsuccessful call delivery compared with the base situation. As shown in the figure, the delays of both successful and
Table 4.2: Latency Parameters (DL & DQ ) for transmission and query Parameters Q DDN S : DNS Query Q DHSS : HSS Query DPL −I : P-CSCF to I-CSCF L : I-CSCF to S-CSCF DI−S α: SIP interactions within co-locates SIP entities
Typical value 100 ms 300 ms 10 ms 10 ms 2 ms
76
4.7. Implementation Considerations
800 Base scenario Proposed method
700 600
Delay (ms)
500 400 300 200 100 0 Delay of a successful call
Delay of a failed call
Figure 4.13: Comparison of delay for successful and unsuccessful call for base situation and proposed method unsuccessful calls are lower than the base case scenario. In figure 4.14 we show the average delay experienced by a user when θ is varying. The average delay remains constant as the failure rate changes in the base case scenario. As shown in the figures 4.11 and 4.13, our proposed method leads to a better delay performance in comparison with the suggested base framework. It can be furthermore interpreted as a decrease in waste of network resources as well as lower service time and better user experience.
4.7
Implementation Considerations
In this section we address some practical implementation issues.
4.7.1
Best Network Selection
In heterogeneous networks where multimodal devices operate, the choice of best network is an important issue. Selection of best network may depend on the available resources, type of service (e.g. voice, video, etc.) or user preference. In NG wireless communication systems, mobile users will have their own preference for delivering the call. For instance, they may choose a preference 77
4.7. Implementation Considerations
800 700 600
Average Delay (m ms)
500 400 300 200
Proposed method Base Scenario
100 0 0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Probability of call failure
Figure 4.14: Average delay performance for proposed scheme when varying session setup failure rate θ
78
4.8. Conclusion on which subscribed network they want to receive the call. In this regard two scenarios can be considered: 1. The primary connection of the called MT is the best network 2. The primary connection of the called MT is not the best network For the first case, the call will be reversed through the existing primary connection. For the second case, the call is delivered to the previously best known network and the MT can update the system about the preferred mode of access and set the preferred access mode as the primary connection. The call can then be reversed through the preferred primary access mode upon availability of network resources.
4.7.2
QoS Aware Multimedia Call Setup
In this section we show a practical implementation of our proposed method and explain a more realistic case for a multimedia session setup that requires QoS negotiation and policy enforcement. As mentioned earlier, a PDF can be incorporated in the P-CSCF that interacts with PE module in either push or pull mode. The PE module enables the service provider to manage and control access to its applications and network resources from a centralized location. Policy rules in PE module are divided into three categories and each policy rule can be executed by a policy engine (service broker) [45]: • Service policy rules • User policy rules • Network policy rules Using the reverse call setup procedure requires that PE module and PDF communicate in the pull mode. In this manner, PE module would request the policy decision from PDF once the called party is available and reverses the call to the AS. Furthermore, PDF will query the policy repository database in HSS for the decision. When users request access to a service or application, the request is relayed to the policy enforcement function for verification and authorization. If the request is approved, the PE module checks and allocates the underlying network resources[45]. Figure 4.15 shows the message flow at the abstract level.
4.8
Conclusion
In transitioning to the future generation of wireless networks, convergence of IP based core networks with heterogeneous wireless access networks is inevitable. The future users of communication systems will subscribe to multiple services and domains and in the foreseeable future deployment of a single logical database will be the way in which all network operators will agree. Multimedia calls are becoming common and chances of failure of a multimedia call can lead to excessive delay 79
4.8. Conclusion
Application Servers CDAS
1
14
HSS -------------------
Policy Repository 9
2 13
10 P-CSCF PDF
MGCF 11
8
12
PSTN
3
PE GGSN RS GP
SGSN
4
Service Policy
User Policy 7 Network Policy
6 5 UE
Figure 4.15: Message flow for multimedia call setup
80
4.8. Conclusion in signaling and waste of network resources. In this work we presented a scheme to address the issue of call delivery in a heterogeneous wireless access network that interworks with IMS. The method is based on the idea of partial retrieval of HSS to locate the user by the existing paging network infrastructure and using a similar approach to the RVC and proposing the deployment of application servers for this purpose. We showed our proposed architecture and layout and demonstrated scenarios for call setup between GPRS, WLAN and PSTN. Furthermore, a more practical case of a multimedia call delivery is shown. In the performance evaluation, we showed that our method reduces the cost of failed calls and helps in better utilization of network resources. We also showed that the delay as a result of our proposed scheme is less than the base scenario which results in a better user experience, lower service time/delay and more efficient use of network resources. We presented some implementation considerations for practical situations and showed that our method can work with existing implementations. Our proposed method is also compatible with possible future deployment of M-SCTP as a transport layer protocol for future multimodal communication devices.
81
Chapter 5
Conclusion In the path to Telco 2.0 and transitioning to the future generation of wireless networks, convergence of IP based core networks with heterogeneous wireless access networks is inevitable. IMS aims at providing a standardized solution for multimedia services in an access agnostic manner within a unified framework. While IMS opens up new business and revenue generating perspectives for service providers and network operators, several business and technical issues are still to be solved. The future users of communication systems will subscribe to multiple services and domains and in the foreseeable future deployment of a single logical database will be the way in which all network operators will agree. Multimedia calls are becoming common and chances of failure of a multimedia call can lead to excessive delay in signaling and waste of network resources. NGNs and IMS will enhance existing wireless services such as voice calls, SMS, MMS, and enable new services such as IM, presence, PoC, UM, and multicast/broadcast. We have shown some examples advanced conversational communications that would be available to the future telecom users over an NGN composed of converged heterogeneous wireless and wireline access networks. IMS plays a major role in enabling these services. Towards the integration of heterogeneous networks with IMS, various challenges such as billing, QoS provisioning, mobility support, bandwidth management, resource allocation, customer profiling, security and location management issues have been presented. In this thesis we looked at the database location management for interworking of wireless heterogeneous networks with IMS. This part was motivated by the idea of a unified logical database. The problem of frequent changes of access mode as a result of high mobility is explained. To reduce the signaling cost as a result of frequent changes of access mode, we proposed the use of an IA to cache the location and mode of access for a user. We demonstrated the performance improvements for situations where user is traveling within and outside a LA. Furthermore, the case for access grant permission and policy decision procedure is illustrated. HSS as a key component of IMS will take over the task of HLRs with additional functionalities. The industry‘s consensus is towards deployment of a unified repository database that holds the subscriber profile information across different domains. One of the issues in mobility of users in heterogeneous environments is through which network should the MT update its location and where should the location information and access mode information of subscribers be stored. Location update through all possible access modes to separate databases for each access network can cause excessive delay in call delivery and service time to the users while creating data duplication and discrepancy. It also causes excessive signaling cost on the network. For the future wireless communication systems where multimodal devices deployed, having a single logical database that applies the subscribers profiles for all CS and Packet Switched (PS) domains would be advantageous. We furthermore presented a scheme to address the issue of call delivery in a heterogeneous 82
5.1. Design Issues, Technical Challenges, and Future Research Directions wireless access network that interworks with IMS. The method is based on the idea of partial retrieval of HSS to locate the user by the existing paging network infrastructure and using a similar approach to the RVC and proposing the deployment of application servers for this purpose. We showed our proposed architecture and layout and demonstrated scenarios for call setup between GPRS, WLAN and PSTN. Furthermore, a more practical case of a multimedia call delivery is shown. In the performance evaluation, we showed that our method reduces the cost of failed calls and helps in better utilization of network resources. We also showed that the delay as a result of our proposed scheme is less than the base scenario which results in a better user experience, lower service time/delay and more efficient use of network resources.
5.1
Design Issues, Technical Challenges, and Future Research Directions
5.1.1
IMS Session Setup in a Heterogeneous Mobile Environment
As mentioned earlier, SIP has been selected as the signaling protocol for multimedia sessions over IPbased mobile networks due to advantages such as simplicity, extensibility, flexibility and scalability. SIP session setup delay performance is shown to be acceptable over the land-based Internet [55]. However, in a converged network environment where wireless networks such as 3G and WLAN coexist, session setup may suffer from unexpected delay and performance degradation due to the limited throughput and/or reliability of the wireless channel causing multiple retransmission of SIP request and response messages, and the need to determine which one of several alternative access networks would best serve the session; i.e., the so-called network selection problem [56]. A network-assisted user-controlled network selection architecture may be suitable [57][58]. Multiattribute decision algorithms have been developed to enable network selection, and fuzzy logic has been applied to deal with imprecise network information [59].
5.1.2
Customer Profile and Identity Management
Profile management is the process of determining the various aspects to be specified in a users profile and how this information can be applied in order to tailor-make services according to the users preferences. Each user can potentially access applications and services via one of several UEs or access modes, and may do so using one of several identities (e.g., employee, private citizen). In order to better adapt and personalize services offered to users in this converged environment, it is important that the network provides an infrastructure capable of managing user identities and profiles in relations to the UE in use, the access mode, and other environmental factors, while enabling secure exchange of private subscriber data between the ASPs, network service providers and operators. Admission policies in the IMS can be enforced at different layers, including the service layer. Thus, it is important to define an access policy model that describes the elements that form part of a policy or serves as input to a policy, and apply this model to implement the admission policies using a policy language.
83
5.1. Design Issues, Technical Challenges, and Future Research Directions
5.1.3
Resource Management for QoS
Connection admission control (CAC), traffic policing, packet scheduling, and channel and power allocation are key resource management techniques necessary for QoS support. In wireless networks, mobility information can be exploited to enhance the performance of CAC [47] and scheduling [48], while the dynamics of the channel need to be taken into account to optimize resource application [49]. With the use of IMS to interconnect heterogeneous wireless networks that employ diverse resource management mechanisms and even dissimilar QoS measures, guaranteeing QoS in anticipation of user movements across different access networks, i.e., inter-technology or vertical handoffs, is challenging. Resource management mechanisms can be tuned to either encourage a vertical handoff, e.g., to make room for new calls, or to discourage a vertical handoff, e.g., when the targeted network is congested. In connections spanning multiple heterogeneous networks, endto-end QoS provisioning calls for techniques to harmonize the diverse QoS measures and allocate the degradations properly over the constituent network segments.
5.1.4
Seamless Mobility Management and Vertical Handover Support
Global Roaming is one of the future needs in mobile and wireless communications that enables users to initiate a call from any access network and seamlessly roam across heterogeneous wireless access networks. Another aspect of global seamless roaming is the continuity of a call or a session when a mobile terminal is moving to different coverage areas with different access network. This leads to initiation and performing a vertical handover. IMS should be able to manage various access related constraints at different layers that are imposed by heterogeneous access technologies. Providing a seamless vertical handover for applications that are sensitive to delay and have strict QoS requirement is a challenge and has attracted extensive research work [22].
5.1.5
Location Management – Tracking of User Locations
Mobility of users brings the challenge of location management. It is important that as users move within the coverage area of a specific wireless access network or roam between different access networks, their current locations can be determined for connection management purposes. With recent U.S. E911 requirements to fairly accurately report the location of VoIP and cellular callers, it opens up the opportunity of location-based services and applications. One example is Location aware PoC (LaPoC) [50]. Other location aware applications can be built around the existing applications and services enabled by IMS. Some examples are: • Combination of presence and location where the location of users can be seen by the other end. • Initiation of messaging based on a defined distance. • Location based context aware adaptation where the users communication decisions are based on the location (i.e. at work or at home). • Multimedia broadcasting based on the end users location.
84
5.1. Design Issues, Technical Challenges, and Future Research Directions Various SIP methods for communicating location information are being considered by the Internet Engineering Task Force (IETF) but no preferred standard has been chosen yet. In provisioning of user locations, [50] proposes the introduction of a service enabler entity called IMS Location Server (ILS) to be located at the Service Layer. ILS does not determine the locations of users, but it passes the location requests to positioning systems to determine the users locations. It furthermore acts as a client that interacts with other application servers via a SIP interface. Support of location and presence services at the Service Layer allows the application servers to support location aware applications. Investigation of new applications encompassing location and presence information is very much of interest to both research community and service providers.
5.1.6
Multicast/Broadcast Services Across Multiple Access Technologies
Currently, IMS only supports services based on unicast mode of transmission. Protocols and standards need to be developed in order for IMS to support multicast / broadcast and advanced group management functionalities. In [51], a system architecture for multicast/broadcast in IMS is proposed that is in line with Telecoms and Internet converged Services and Protocols for Advanced Networks (TISPAN) and NGN reference architectures. This design offers a converged logical framework that brings together a diversity of multicast-transport bearers such as Digital Video Broadcast for Handheld devices (DVB-H), Multimedia Broadcast Multicast Service (MBMS) and Broadband Multimedia Communications System (BMCS). Enabling multicast/broadcast services across multiple access technologies opens up many challenges in the areas of network selection, QoS management, session management and advanced multicast group management functionalities.
5.1.7
Issues with Call Admission Control
In the existing Call Admission Control (CAC) schemes the priority is given to handoff calls rather than new calls. For multimedia services, there are methods in which the connections are admitted whenever there are sufficient resources to support at least the mean data rate of the connection [42],[52]. In another work a joint CAC is proposed that takes the user‘s preferences into considerations [53]. From the proposed method it can be realized that the called party reverses the call via AN to the CDAS. The proposed call delivery scheme is compatible with the above mentioned CAC methods. In the following we explain how each of CAC schemes may be implemented with our proposed call delivery scheme. Many studies have focused on admission control for vertical handovers between cellular and WiFi networks. In these studies, handovers are assumed to have a higher priority than new calls. It therefore implies that chances of new calls not being admitted are greater than the one for handovers [42]. For CAC schemes that prioritize handover calls, the reverse call setup in our call delivery scheme will not be considered as a new call. We are currently working on developing an appropriate CAC scheme that suits best for this method. One possible approach is to treat the call reversals as if they are handover calls to minimize potential waste of network resources. For CAC schemes that are based on user‘s preference, my future work will address this problem by offering solutions that are jointly based on available resources to support a certain data rate while taking the user‘s preference into account. 85
5.1. Design Issues, Technical Challenges, and Future Research Directions
5.1.8
Charging Implications
In the proposed method of call delivery, although the call is set up in reverse direction, but still the calling party will be billed for the call and the called party will only be charged for possible roaming charges.
5.1.9
Migration from HLR to HSS
Towards adoption of IMS convergence solutions, service and network providers should migrate from HLRs to HSS. Since HSS is a more extensive database than HLR, it therefore needs to be geographically distributed. It furthermore, needs to support the existing and newly required functionalities, accommodate new subscribers while maintaining high availability and reliability as well as being simply manageable in a highly distributed framework. From the manufacturing point of view, however there may be upper limits on the size of a database. On the other hand, increasing the number of physical nodes may lead to a more complex management system. These can lead to researches on the optimal size of a database and the number of physical nodes.
5.1.10
Design Considerations for Mobility Support and QoS
There are several design considerations for mobility and end-to-end QoS support [54]: One of the considerations is the interoperability between the networks of different service providers. In a scenario where various administrative domains with different policies, management architectures and traffic mechanisms coexist, a common protocol would be needed for communication of end-to-end QoS requirements of user traffic. It is also necessary to respect the individualities of independent operation of traversed networks at the same time. Different autonomous networks and many proprietary solutions on the other hand, make this task difficult. One solution to make the system interoperable is to separate the signaling protocol from data. In a design proposed in [54] this concept is followed by decoupling the signaling plane from media plane. Another design consideration targets the choice of signaling protocol. An inter-domain signaling protocol can be used to enable the interaction of different domains, so that they can be integrated in a manner capable of provisioning QoS and supporting mobility functionalities. Solutions and examples that apply this concept are discussed in [54] by Gouveia and Magedanz. Finally, the integration level is also a factor that should be considered in a design. 3GPP defines two levels of integration between 3G cellular networks and WLANs: tightly coupled and loosely coupled integration. In tightly coupled integration, the WLAN access point is connected to the Serving GPRS Support Node (SGSN) and behaves like a node B (i.e., a 3G base station). In loosely coupled integration, the WLAN and 3G cellular networks are autonomous domains but may share a common Authentication, Authorization and Accounting (AAA) server that enables, e.g., a UE to be authenticated for WLAN access using its 3G cellular credential. In this case the two domains interwork via the respective gateways, namely WLAN Access Gateway (WAG) and Packet Data Gateway( PDG). 86
Bibliography [1] M. Marchese, ”QoS Over Heterogeneous Networks”, John Wiley & Sons, Inc, 2007 [2] T. Kourlas,”The Evolution of Networks beyond IP”, IEC Magazine ,2007 [3] I. Moraes,”Exploring Drivers for IMS Enhanced Services Platform Architectures and Applications”, IEC Magazine, 2007 [4] Y. Lin,”Wireless and Mobile All-IP Networks”, John Wiley & Sons, Inc (2005) [5] T. Macaulay ,”Securing Converged IP Networks”, Auerbach Publications, 2006 [6] ”Extensible Markup Language (XML)” ; Jan 2008, http://www.w3.org/XML [7] ”IP Multimedia Subsystem - stage 2,” TS 23.228, 3GPP, Rel. 7, June 2007. [8] Spyros L. Tompros, ”NGN networks; A new enabling technology or just a network integration solution?”, White paper appointed by VITAL consortium August 2007 [9] G. Camarillo, M. Garcia-Martin, ”The 3G IP Multimedia Subsystem (IMS): Merging the Internet and the Cellular Worlds”, Second Edition. John Wiley & Sons, 2006. [10] M. Handley, H. Schulzrinne, E. Schooler, J. Rosenberg, ”SIP: Session Initiation Protocol ”, RFC 2543, March 1999 [11] ”IP Multimedia (IP) session handling; IM call model, stage 2,” TS 23.218, 3GPP, Rel. 7, June 2007. [12] ”Open Service Access (OSA); Parlay X Web Services” , TS 29.199-1, 3GPP, Rel. 6 , Dec 2005 [13] ”All-IP Core Network Multimedia Domain Rev. A” , X.S0013, 3GPP2, Sept 2005 [14] ”Generic User Profile (GUP)”, TS 129.240, 3GPP, Rel. 7, June 2007 [15] ”Charging Architecture” , OMA-AD-Charging Draft v1.1, OMA, Oct 2007 [16] G. Gomez , R. Sanchez,”End-to-End Quality of Service over Cellular Networks: Data Services Performance Optimization in 2G/3G” ,John Wiley and Sons Ltd, April 2005 [17] W. Peng, C. Takeo, H. Jeyhsin, Y.J. Hung-Yu,”QoS Management and Peer-to-Peer Mobility in Fixed-Mobile Convergence”, Fujitsu Sci Tech , Japan (October 2006) [18] V.C.M. Leung, F. Yu, J. Zhang, H.C.B. Chan, ”SIP Signaling for Vertical Handovers in Heterogeneous Wireless Networks”, book chapter in SIP Handbook: Services, Technologies, and Security, S. Ahson and M. Ilyas, ed., CRC Press 2008 87
Chapter 5. Bibliography [19] W. Jiao, J. Chen, F. Liu, ”Provisioning end-to-end QoS under IMS over a WiMAX architecture: Research Articles”. Bell Lab. Tech. pp.115-121, May 2007 [20] ”End-to-End QoS Concept and Architecture” 3GPP TS 23.207, v7.0.0 Release 7, June 2007 [21] A. Elmangosh, M. Ashibani,F. Shatwan, ”Quality of Service Provisioning Issue of Accessing IP Multimedia Subsystem via Wireless LANS”, New Technologies, Mobility and Security, pp.133-143, Springer Netherlands 2007 [22] J. Zhang, H.C.B. Chan and V.C.M. Leung, ”A SIP-based Soft-handoff (S-SIP) Scheme for Heterogeneous Mobile Networks”, in Proc. IEEE WCNC07, Hong Kong, China, Mar. 2007 [23] J. Zhang, E. Stevens-Navarro, V.W.S. Wong, H.C.B. Chan and V.C.M. Leung, ”Protocols and Decision Processes for Vertical Handovers,” book chapter in Unlicensed Mobile Access Technology: Protocols, Architectures, Security, Standards and Applications, Y. Zhang, L. Yang, J. Ma, ed., Auerbach Publications, CRC Press 2008 [24] J. Zhang, H.C.B. Chan and V.C.M. Leung, ”A Location-Based Vertical Handoff Decision Algorithm for Heterogeneous Mobile Networks”, in Proc. IEEE Globecom06, San Francisco, CA, Nov. 2006 [25] J.C. Chen, T. Zhang,”IP-Based Next Generation Wireless Networks: Systems, Architectures, and Protocols” John Wiley & Sons, Inc, 2004 [26] P. TalebiFard, T. Wong and V.C.M. Leung, ”Integration of Heterogeneous Wireless Access Networks with IP-Based Core Networks”, book chapter in Heterogeneous Wireless Access Networks”, Ekram Hossain ed., Springer, 2008. [27] ”3GPP System to Wireless Local Area Network (WLAN); System Description”, TS 23.234, 3GPP, Rel. 7, June 2007. [28] ”Feasability Study of Mobilty between 3GPP-WLAN Interworking and 3GPP Systems” ,TR 23.827, 3GPP, Rel. 8, September 2007. [29] Y. Huh and C. Kim, ”New caching-based location management scheme in personal communication systems,” 15th International Conference on Information Networking, pp. 649–654, 2001. [30] D. Vergados, A. Panoutsakopoulos, and C. Douligeris, ”Location management in 3G networks using a 2-level distributed database architecture,” IEEE 18th International Symposium on Personal, Indoor and Mobile Radio Communications, pp. 1–5, September 2007. [31] J.S.M. Ho, I.F. Akyildiz, ”Dynamic hierarchical database architecture for location management in PCS networks,” IEEE/ACM Transactions on Networking,vol.5, no.5, pp.646-660, Oct 1997 [32] R. Jain, Y. B. Lin, C. Lo, S. Mohan, ”A caching strategy to reduce network impacts of PCS,” IEEE Journal on Selected Areas in Communications, vol.12, no.8, pp.1434-1444, Oct 1994 88
Chapter 5. Bibliography [33] R. Jain and Y. B. Lin, ”An Auxiliary User Location Strategy Employing Forwarding Pointers to Reduce Network Impacts of PCS,” Record of the 1995 IEEE Intl. Conf. on Communications, June 1995. [34] C.M Weng, C.H. Chu, ”K-step pointer forwarding strategy for location tracking in distributed HLR environment,” IEEE Proceedings-Communications, vol.150, no.3, pp. 207-213, June 2003 [35] Y.B. Lin, I. Chlamtac, ”Heterogeneous personal communications services: integration of PCS systems”, IEEE Communications Magazine, vol.34, no.9, pp.106-113, Sep 1996 [36] I.F. Akyildiz, and W. Wang, ”A Dynamic Location Management Scheme for Next-Generation Multitier PCS Systems”, IEEE Trans. Wireless Communications, vol.1, no.1, pp. 178-189, January 2002. [37] I.F. Akyildiz, J. Xie, S. Mohanty, ”A survey of mobility management in next-generation allIP-based wireless systems,” IEEE Wireless Communications, vol.11, no.4, pp. 16-28, Aug. 2004 [38] ”IP multimedia subsystem - Stage 2” ,TS 23.228, 3GPP, Rel. 8, Sept 2008 [39] P. TalebiFard and V.C.M. Leung, ”Efficient Multimedia Call delivery over IP-Based Heterogeneous Wireless Access Networks”, in Proc. ACM MobiWAC, Vancouver, BC, Oct. 2008. [40] 3rd Generation Partnership Project, 3GPP Technical Specifications, ”3G Security; WLAN interworking security (Release 7)”, 3GPP TS 33.234 v7.0.0, Mar. 2006. [41] Al Shidhani, A.; Leung, V.C.M., ”Local fast re-authentication protocol for 3G-WLAN interworking architecture,” Wireless Telecommunications Symposium, 2007. WTS 2007 , vol., no., pp.1-8, 26-28 April 2007. [42] J. Zhang, E. Stevens-Navarro, V.W.S. Wong, H.C.B. Chan and V.C.M. Leung, ”Protocols and Decision Processes for Vertical Handovers”, book chapter in Unlicensed Mobile Access Technology: Protocols, Architectures, Security, standards and Applications, Y. Zhang, L. Yang, J. Ma, ed., Auerbach Publications, CRC Press 2008. [43] S. Zaghloul, A. Jukan; W. Alanqar, ”Extending QoS from Radio Access to an All-IP Core in 3G Networks: An Operator’s Perspective,” Communications Magazine, IEEE , vol.45, no.9, pp.124-132, September 2007 [44] R. Gayraud, O. Jacques, April 2008; http://sipp.sourceforge.net/ims_bench. [45] P. TalebiFard, T. Wong, and V.C.M Leung, ”Integration of heterogeneous wireless access networks with IP-based core networks,” book chapter in Heterogeneous Wireless Access Networks, Ekram Hossain ed., Springer, 2008. [46] I. Chih-Lin, P. P. Gregory, and D. G. Richard, ”Pcs mobility management using the reverse virtual call setup algorithm”, IEEE/ACM Trans. Netw., vol. 5, no. 1, pp. 13–24, 1997.
89
Chapter 5. Bibliography [47] F. Yu, V.W.S. Wong and V.C.M. Leung, ”Performance enhancement of combining QoS provisioning and location management in wireless cellular networks”, IEEE Trans. Wireless Comm., vol. 4, no. 3, pp. 943-953, May 2005. [48] S.H. Ali, V. Krishnamurthy and V.C.M. Leung, ”Optimal and Approximate Mobility Assisted Opportunistic Scheduling in Cellular Networks”, IEEE Trans. Mobile Computing, vol. 6, no. 6, pp. 633-648, June 2007. [49] S.H. Ali, K.-D. Lee and V.C.M. Leung, ”Dynamic Resource Allocation in OFDMA Wireless Metropolitan Area Networks”, IEEE Wireless Comm., vol. 14, no. 1, pp. 6-13, Feb. 2007. [50] M. Mosmondor, L. Skorin-Kapov, R. Filjar,”Location Conveyance in the IP Multimedia Subsystem”, Volume 4003, Springer Berlin / Heidelberg, 2006 [51] A. Ikram,M. Zafar, N. Baker, R. Chiang,”IMS-MBMS Convergence for Next Generation Mobile Networks,” The 2007 International Conference on Next Generation Mobile Applications, Services and Technologies, 2007, pp.49-56, 12-14 Sept. 2007 [52] J. Roy, J. Juliet, V. Vaidehi, and S. Srikanth, ”Always best-connected QoS integration model for the wlan, wimax heterogeneous network,” First International Conference on Industrial and Information Systems, pp. 361–366, Aug. 2006. [53] O. E. Falowo and H. A. Chan, ”Joint call admission control for next generation wireless network”, Canadian Conference on Electrical and Computer Engineering, 2006. CCECE ’06., pp. 1151–1154, May 2006. [54] F. Carvalho de Gouveia, T. Magedanz, ”A Framework to Improve QoS and Mobility Management for Multimedia Applications in the IMS,” Seventh IEEE International Symposium on Multimedia (ISM’05),pp. 216-222, 2005 [55] Victor Y.H. Kueh, R. Tafazolli, Barry G. Evans, ”Performance analysis of session initiation protocol based call set-up over satellite-UMTS network”, Computer Communications Volume 28, Issue 12, , 18 July 2005, Pages 1416-1427 [56] F. Bari and V.C.M. Leung, ”Service Delivery over Heterogeneous Wireless Systems : Networks Selection Aspects”, in Proc. ACM IWCMC06, Vancouver, BC, July 2006 [57] F. Bari and V.C.M. Leung, ”Automated network selection in a heterogeneous wireless network environment”, IEEE Network, vol. 21, no. 1, pp. 34-40, Jan.-Feb. 2007 [58] F. Bari and V.C.M. Leung, ”Architectural aspects of automated network selection in heterogeneous wireless systems”, International Journal of Ad Hoc and Ubiquitous Computing, Feb. 2008. [59] F. Bari and V.C.M. Leung, ”Network Selection with Imprecise Information in Heterogeneous All IP Wireless Systems”, Wireless Internet Conference (WiCon), Austin, TX, Oct. 2007
90
Appendix A
Statement of Co-Authorship I am the first author and the main contributor of all the papers published that are of the main contribution to this thesis. chapters 3 and 4 are based on the two published conference papers5
6
where my contributions were: • identification and design of research problem • performing the research • data analyses • manuscript preparation Dr. Victor C.M. Leung supervised this thesis and is a co-author of all papers. Introduction and Background sections of this thesis are based on a book chapter7 where my contributions were: • preparing the outline of the chapter • performing the research • manuscript preparation Small portions of Introduction and Background sections are also a version of an invited paper8 . My contributions were: • preparing the outline and material for the paper • performing the research • manuscript preparation and final revisions
5
P. TalebiFard and V.C.M. Leung, ”Novel Database Architecture and Signaling Scheme for IP-Based Heterogeneous Wireless Access Interworking”, Proc. ICST WICON, Maui, Hawaii, Nov. 2008 6 P. TalebiFard and V.C.M. Leung, ”Efficient Multimedia Call delivery over IP-Based Heterogeneous Wireless Access Networks”, Proc. ACM MobiWAC, Vancouver, BC, Oct. 2008 7 P. TalebiFard, T. Wong, and V.C.M Leung, ”Integration of heterogeneous wireless access networks with IP-based core networks”, book chapter in Heterogeneous Wireless Access Networks, Ekram Hossain ed., Springer, 2008 8 V.C.M. Leung, T. Wong and P. TalebiFard, ”Breaking the Silos - Access and Service Convergence over the Mobile Internet”, Proc. ACM MSWiM, Vancouver, BC, Oct. 2008 (invited)
91
