620
IEEE Transactions on Consumer Electronics, Vol. 56, No. 2, May 2010
Design and Implementation of An Adaptive Middleware Based on The Universal Middleware Bridge for Heterogeneous Home Networks Yu-Seok Bae, Bong-Jin Oh, Kyeong-Deok Moon, Young-Guk Ha, and Sang-Wook Kim Abstract — As home networks become more complex
and dynamic, it is crucial to support seamless interoperability through automatic reconfiguration and robustness through the efficient handling of faults. This paper presents an adaptive middleware that provides an adaptive autonomic configuration for the heterogeneous home networks and an autonomous fault management that includes fault diagnosis and recovery from unexpected faults such as device plug-outs, network link failures, service failures, and other such incidents. The proposed adaptive middleware is based on the universal middleware bridge (UMB) to guarantee interoperability and robustness in the middleware layer appropriate to heterogeneous home networks. 1. Index Terms — Adaptive Middleware, UMB, AAC, AFM, Home Networks.
scheme that accounts for dynamic status changes in home networks. This paper proposes an adaptive middleware based on the UMB that supports adaptive autonomic configuration and autonomous fault management in the middleware layer in order to guarantee the seamless interconnection and robustness in heterogeneous home networks. The rest of this paper is organized as follows. Section II discusses related works on the interoperability and robustness of home networks. Section III briefly describes the UMB system while Section IV presents the proposed adaptive middleware architecture that supports an adaptive autonomic configuration and autonomous fault management. The implementation of the proposed architecture is discussed in Section V. Section VI is the summary and conclusion of this paper.
I. INTRODUCTION
II. RELATED WORKS
Due to the increasing number of interconnections of heterogeneous devices and networks in home network environments, various types of faults that cause failures in these networks commonly occur. Therefore, a fault management scheme that ensures the robustness of home networks is a vital issue and should be seriously considered to maintain the stability of heterogeneous home network environments. A considerable number of studies on fault management have been conducted in the network domain, but fault management studies in the home network domain have focused on the detection and handling of partial faults based on a rule-based model. Moreover, problems arising from cascading or propagated faults can induce a series of wrong actions and cause home networks to be finally paralyzed from critical failures [1]. Therefore, in order to ensure robustness and seamless interoperability in the home networks, it is essential to provide a fault management and automatic configuration management
Several studies have been conducted on the topic of the interoperability and robustness of home networks. The universal middleware bridge (UMB) system dynamically maps physical devices into virtually abstracted devices and guarantees seamless interoperability among heterogeneous home network devices [2, 3, 4]. Studies of the fault-tolerance capabilities essential for realizing high reliability in the UMB facility were established based on the TMO (Time-triggered Message-triggered Object) structuring approach [5, 6]. The framework for connecting home computing middleware developed at Waseda University enables one middleware to connect to another using a Virtual Service Gateway. However, this framework has some limitations that are related to dynamic service activation and different combination of the functions of the appliances in such networks [7]. A multi-agent-based solution for fault diagnosis systems for connected home provides distributed and highly autonomous characteristics, but the overhead of such a system may not be appropriate for home networks with limited resources [8]. The Trust4All project aims to define an open componentbased framework for the middleware layer in high-volume embedded appliances that enables robust and reliable operations while the software is being upgraded and extended [9].
1 This work was supported by the IT R&D Program of MKE/KEIT, [2009F-027-01, Development of Interoperable Home Network Middleware for Setting Home Network Heterogeneity]. Yu-Seok Bae, Bong-Jin Oh, and Kyeong-Deok Moon are with the Electronics and Telecommunications Research Institute, 161 Gajeong-dong, Yuseong-gu, Daejeon, Korea (e-mail: {baeys, bjoh, kdmoon}@etri.re.kr). Young-Guk Ha is with the Konkuk University, 1 Hwayang-dong, Gwangjin-gu, Korea (corresponding author, e-mail:
[email protected]). Sang-Wook Kim is with the Kyungpook National University, 1370 Sangyeok-dong, Buk-gu, Dae-gu, Korea (e-mail:
[email protected]).
Contributed Paper Manuscript received April 14, 2010 Current version published 06 29 2010; Electronic version published 07 06 2010.
0098 3063/10/$20.00 © 2010 IEEE
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
Y.-S. Bae et al.: Design and Implementation of An Adaptive Middlewar Based on The Universal Middleware Bridge for Heterogeneous Home Networks
III. UNIVERSAL MIDDLEWARE BRIDGE The UMB system provides a feasible solution for interoperability problems caused by the heterogeneity inherent in different types of home network middleware. Figure 1 shows a conceptual model of the UMB system.
Fig. 1. Conceptual model of the UMB system
The UMB dynamically maps physical devices in different middleware domains onto virtually abstracted devices in the UMB domain and enables all home devices overlaid on heterogeneous networks to be seen as virtually the same physical devices in the same middleware domain in order to detect and control each other. Therefore, the UMB not only provides an application-level interoperability framework but also provides UMB message protocols and Virtual Device Type (VDT) schemas for representing various devices in the UMB domain.
621
The main functions of the UMB-Core are managing UMBAdaptors (or Middleware Adaptors) and providing message interpretation and routing as well as dynamic connections for UMB Adaptors. A UMB-Adaptor is an adaptor that connects a home network middleware to the UMB-Core and communicates with other UMB-Adaptors. The UMB-Adaptor manages and controls devices in a local home network middleware domain such as UPnP or LonWorks, and provides device conversion and message translation between the home network middleware and the UMB system. For message transmission in the UMB system, a SOAP-based UMB message protocol is defined. IV. ADAPTIVE MIDDLEWARE ARCHITECTURE Figure 2 shows the proposed adaptive middleware architecture that provides capabilities for interoperability and robustness in heterogeneous home networks based on the UMB system. The Heterogeneity Abstraction Layer supports interoperability of heterogeneous home network middleware based on the UMB. It also composes the network topology and collects network traffic information. As shown in the Management Layer of the proposed architecture, the adaptive autonomic configurator (AAC) aims to provide automatic configuration management to reflect the dynamic status changes in home networks. The autonomous fault manager (AFM) deals with real-time fault diagnosis as well as fault evasion and fault recovery for robustness in home networks. In addition, the lean reliable communication broker (LCB) is a messaging system that enables reliable message transmission. A. Adaptive Autonomic Configurator The AAC consists of four types of components; a device manager (DM), a service manager (SM), a network manager (NM), and a resource manager (RM).
Fig. 2. Adaptive middleware architecture
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
622
IEEE Transactions on Consumer Electronics, Vol. 56, No. 2, May 2010
1) Device Manager The DM controls the UMB-Core and UMB-Adaptors in order to manage the interconnections of heterogeneous home network devices according to device status changes. In addition, a Device Agent (DA) is used to launch UMB facility and legacy software for UPnP devices such as an UPnP Media Server and an UPnP Media Renderer. The DM provides a uniform access mechanism for virtually abstract devices and manages status changes and event handling scheme for these devices. Figure 3 shows a working model of the DM with a device abstraction for physical home network devices.
2) Service Manager The SM controls the service lifecycle pertaining to the execution and termination of components for adaptive middleware and home network services. In addition, a Service Agent (SA) is used for service actuation and state monitoring. Figure 4 shows a working model of the SM undergoing a service download from a service repository along with its service actuation and monitoring operations.
Fig. 4. SM working model
Fig. 3. DM working model
The DM provides a device abstract service (DAS) coordinator and a message converter, and a device and event table manger for heterogeneous home networks. The DAS coordinator facilitates device abstraction providing a type of device surrogate via a device mapper that binds logical devices to DAS services. The message converter is responsible for message translations between UMB messages and messages in a local home network middleware. The device table maps physical devices to virtual abstracted devices and the event table maintains the status of the devices. The DM also reports device faults to the AFM about device faults arising from unexpected device plug-outs in home networks.
The SM interprets service logic, commands the SA to download services from the service repository and to launch them, and manages the service binding table of service status data. Likewise, the SM reports service failures to the AFM due to the unexpected termination of a component or service. 3) Network Manager The NM is responsible for collecting network topology and traffic information to guarantee QoS for reliable home network services. It consists of a Topology Investigator (TI) and a Traffic Analyzer (TA). The TI composes a network topology map using the LonTalk protocol and the LLTD (Link Layer Topology Discovery) protocol [10, 11]. The TA collects node traffic and link traffic by means of the Iperf [12]. Automatic discovery of the physical topology plays a crucial role in enhancing the manageability of networks [13]. Figure 5 shows a working model of the NM using the LLTD protocol. The former TI performs a series of actions. It discovers hosts by broadcasting advertisement message, detects segments with more than one host, analyzes edges for detecting shallow switches or island switches, and discovers gaps via gap dividing and resolution operation.
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
Y.-S. Bae et al.: Design and Implementation of An Adaptive Middlewar Based on The Universal Middleware Bridge for Heterogeneous Home Networks
623
The IRM table enables the monitoring of home network status changes at a glance by providing a cross-layered topology map and resource binding status information. Table I shows several parts of the IRM table maintained by the RM. TABLE I INTEGRATED RESOURCE MAP TABLE
Fig. 5. NM working model
The latter TA computes node traffic by intercepting packets using the PCAP (packet capture) library and estimates link traffic through a series of actions that include measuring the end-to-end bandwidth using Iperf, calculating the segment bandwidth, and estimating link traffics in a segment referring to the network topology. 4) Resource Manager The RM manages resource reservations for home network services and also provides an integrated resource map (IRM) table which collects real-time home contexts such as device information, the status of services and components, network topology and traffic information, and resource usage in home networks. The RM maintains the latest IRM table by interpreting messages received from AAC components according to dynamic status changes in home networks. The IRM tables are backed up in the repository manager (RPM) whenever it changes because they are referenced for fault diagnosis and recovery in the AFM. The RPM provides persistent database storages to store the IRM tables and a series of actions for fault diagnosis and recovery being performed in the AFM.
Device ID
Device Type
Loc.
Neighbor List
H1
Host
R1
H2,MS1
H2
Host
R1
H1,MS1
H3
Host
R2
MR1
H4
Host
R3
MS2,B1
MS1 MR1 MS2 B1
UPnPMS UPnPMR UPnPMS Bridge
R1 R2 R3 R3
H1,H2 H3 H4,B1
S1
Switch
R1
S2
Switch
R2
S3
Switch
R3
L1 W1
Light Window
R3 R3
Service ID SM AFM HTS DM UMB-C UPnP-A NM RM Mediatomb Xine LonWorks-A
Binding Device List
MS1,MR1
MR1 MS1 L1,W1
H1,H2, MS1,S2 S1,H3, MR1,S3 S2,MS2, B1,H4
B. Autonomous Fault Manager The AFM deals with real-time fault diagnosis as well as fault evasion and fault recovery to ensure robustness in home networks. The AFM obtains home contexts from the IRM table to the extent necessary for fault diagnosis and recovery. Figure 6 shows the message flows between the AAC components and the AFM.
Fig. 6. Message flows
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
624
IEEE Transactions on Consumer Electronics, Vol. 56, No. 2, May 2010
The AAC components notify the RM of dynamic status changes related to devices, services, and networks within home networks and deliver fault messages to the fault queue of the AFM. The AFM provides efficient fault management through fault message clustering, which reduces total fault recovery time. Figure 7 shows an overall fault management process in the AFM.
service that uses an UPnP Media Server and an UPnP Media Renderer in home networks. If the DM detects a loss of connectivity in an UPnP Media Server or an UPnP Media Renderer from the UMB, it delivers a device fault message concerning this event to the AFM. The AFM investigates whether it is a localized fault or a cascading fault and whether or not it can be recoverable. In other words, the AFM examines a faulted sub-tree of RDG where the device belongs whether the network link is alive by requesting a response for alive messages to the sub-tree nodes and determines fault ranges. If the fault is localized in a device and cannot be recovered, the AFM searches for another substitute device in the RDG, performs a device surrogate operation for fault evasion, and carries out fault recovery operations necessary to restore the normal streaming status prior to the fault. In the case of network partitioning, a partial home network subgroup is reorganized from the LCB messaging system layer, and the SM maintains the AAC components and the AFM for adaptive middleware. The AFM fetches the latest backup of the IRM table and delivers a re-sync message with the IRM table to the AAC components to synchronize home networks. It also executes fault evasion and recovery operations according to the result of fault diagnosis. All operations for fault diagnosis and recovery in the AFM are stored in the RPM and the Home Viewer service provides timeline-based visual simulations of status changes in home networks. C. Lean reliable Communication Broker The LCB messaging system plays an important role in the reliable message transmission among components and services. It is composed of a LCB-S (Slave) and a LCB-M (Master). Multiple LCB-Ss are connected to an LCB-M. Each middleware component uses LCB-S APIs in order to deliver control messages to other component through the LCB-M. Figure 8 shows a working model of the LCB messaging system.
Fig. 7. Overall fault management process
The AFM composes a real-time resource dependency graph (RDG) from the IRM table that reflects the home network status. It is useful for finding the location of faults and determining their propagation range. The RDG for fault diagnosis is also modified whenever the IRM table is changed. The AFM also creates a fault dependency graph (FDG) related to cascading faults from fault messages. The AFM analyzes fault messages in the fault queue by comparing the RDG with the FDG, and then finds the direct causes of faults. After that it determines the propagation range and performs fault evasion and fault recovery about recovery objects for seamless robustness of home networks. Media streaming service is a typical service often used in the home networks. The Home Theater Service is a streaming
Fig. 8. LCB working model
The LCB messaging system supports total ordering with a privilege-based algorithm and virtual synchrony for group communication to guarantee message consistency during message delivery operations using a token, a sequence number
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
Y.-S. Bae et al.: Design and Implementation of An Adaptive Middlewar Based on The Universal Middleware Bridge for Heterogeneous Home Networks
625
and a queue management scheme. In addition, it supports group management with reference to network partitioning and network merge environments. V. IMPLEMENTATION In order to verify the feasibility of the proposed middleware, a test bed was created with a residential gateway, hosts, UPnP Media Servers and UPnP Media Renderers, and LonWorks devices such as lights, valves, and a window. Figure 9 shows the test bed used to verify adaptive middleware in a heterogeneous home network. (b) Fig. 10. Screenshots
As shown in the Service Layer of the proposed architecture, the Home Viewer service communicates with the AFM and provides timeline-based visual simulations of fault diagnosis and recovery operations depending on status changes in home networks. VI. CONCLUSION
Fig. 9. Test bed for home networks
The adaptive middleware and the UMB system were implemented in the Linux environment. The Mapper service module for LLTD protocol, used for discovering hosts in home networks, was also implemented in the Linux. In addition, the libpcap library and Iperf were used for computing node and link traffic and the lightweight SQLite database was used for device management. Figure 10 (a) and Figure 10 (b) respectively show screenshots of the home network topology and autonomous fault diagnosis and recovery operations of the AFM.
The auto-configuration and fault management capabilities are important factors as they support ease-of-use and ease-ofmanagement in heterogeneous home networks. In this paper, we proposed the adaptive middleware architecture based on the UMB for heterogeneous home networks. The UMB system provides an effective solution for interoperability problems in heterogeneous home networks based on the concept of the UMB-Core and the UMB-Adaptor. The AAC provides adaptive auto-configuration according to dynamic status changes in home networks and the AFM offers efficient fault diagnosis and recovery through the RDG and FDG using fault message clustering. The proposed adaptive middleware based on the UMB with AAC components and the AFM in the middleware layer will provide a highly flexible solution for seamless interoperability and robustness in heterogeneous home networks. Future work in this area may include applying the proposed adaptive middleware to different ubiquitous computing environments. REFERENCES [1]
[2]
[3]
(a)
Y. Kitamura, M. Ueda, M. Ikeda, S. Kobori, O. Kakusho, and R. Mizoguchi, “A Model-based Diagnosis with Fault Event Models,” Proc. of Pacific Asian Conference on Expert Systems (PACES) 97, pp.275282, Feb. 1997. K. D. Moon, Y. H. Lee, C. E. Lee, and Y. S. Son, “Design of a Universal Middleware Bridge for Device Interoperability in Heterogeneous Home Network Middleware,” IEEE Transactions on Consumer Electronics, Vol. 51, No. 1, pp.314-318, Feb. 2005. Y. S. Bae, B. J. Oh, K. D. Moon and S. W. Kim, “ Architecture for Interoperability of Services between an ACAP Receiver and Home Networked Devices,” IEEE Transactions on Consumer Electronics, Vol. 52, No. 1, pp. 123-128, Feb. 2006.
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
626 [4]
[5]
[6]
[7]
[8] [9] [10] [11] [12] [13]
IEEE Transactions on Consumer Electronics, Vol. 56, No. 2, May 2010 D. H. Kim, C. H. Lee, J. H. Park, K. D. Moon, K. S. Lim, ”Scalable Message Translation Mechanism for the Environment of Heterogeneous Middleware,” IEEE Transactions on Consumer Electronics, Vol. 53, No. 1, Feb. 2007. K.. D. Moon, J. H. Park, K. H. Kim, L. C. Zheng and Q. Zhou,, "FaultTolerance in Universal Middleware Bridge," Proc. of 11th IEEE Symposium on Object Oriented Real-Time Distributed Computing (ISORC 2008), pp.471-477, 2008 K. H. Kim, Q. Zhou, J. Qian, K. D. Moon, J. H. Park, Y. S. Son, C. E. Lee, T. Y. Ku, "Realization of Fault-Tolerant Home Network Management Middleware with the TMO Structuring Approach and an Integration of Fault Detection and Reconfiguration Mechanisms," Proc. of 12th IEEE International Symposium on Object/Component/ServiceOriented Real-Time Distributed Computing (ISORC 2009), pp.188-197, 2009. E. Tokunaga, H. Ishikawa, M. Kurahashi, Y. Morimoto, and T. Nakajima, “A Framework for Connecting Home Computing Middleware,” Proc. of 22nd International Conference on Distributed Computing Systems (ICDCSW 02), pp. 765-770, 2002. Peter Utton and Eric Scharf, “A Fault Diagnosis System for the Connected Home,” IEEE Communications Magazine, Vol.42, pp.128134, Nov. 2004. R. Su, M.R.V. Caudron and J.J. Lukkien, “Adaptive runtime fault management for service instances in component-based software applications,” IET Software, Vol.1, No.1, pp.18-28, 2007. Echelon Co., LonTalk Protocol Specification Version 3.0, 1994. Microsoft Co., LLTD: Link Layer Topology Discovery Protocol Specification Version 1.0.8, May 2006. A. Tirumala, M. Gates, F. Qin, J. Dugan and J. Ferguson. Iperf - The TCP/UDP band-width measurement tool. M. H. Son, B. S. Joo, B. C. Kim, and J. Y. Lee, "Physical Topology Discovery for Metro Ethernet Networks," ETRI Journal, Vol. 27, No. 4, pp.355-366, Aug. 2005.
Bong-Jin Oh received B.S. and M.S. degrees in computer science from Pusan National University, Korea in 1993 and 1995 respectively. Since 1995, he has been with the Electronics and Telecommunications Research Institute (ETRI), where he develops home network middleware and data broadcasting middleware. He has been a PhD student at Chungnam National University since 2006. His research interests are home network middleware, data broadcasting middleware, IPTV, and pervasive computing. Kyeong-Deok Moon received B.S. and M.S. degrees in computer science from Hanyang University, Korea in 1990 and 1992 respectively. From 1992 to 1996, he was a researcher at System Engineering Research Institute where he worked on high performance computing and clustering computing. Since 1997, he has been a senior researcher and a director at the Electronics and Telecommunications Research Institute (ETRI), where he develops home network middleware and Java embedded architecture. He received a PhD degree from ICU in 2005. His research interests are home network middleware, Java, active networks, and pervasive computing. Young-Guk Ha received his BS and MS degrees in computer science from Konkuk University in 1993 and 1995 respectively. He received his PhD degree in computer science from Korea Advanced Institute of Science and Technology (KAIST) in 2006. He worked for Electronics and Telecommunications Research Institute (ETRI) from 1995 to 2008 as a senior member of the engineering staff and currently is an assistant professor in the Department of Computer Science and Engineering at Konkuk University. His research interests are in ubiquitous computing, home network and service middleware, and wireless sensor networks.
BIOGRAPHIES Yu-Seok Bae received B.S. and M.S. degrees in computer science from Kyungpook National University, Korea in 1995 and 1997 respectively. Since 1997, he has been with the Electronics and Telecommunications Research Institute (ETRI), where he develops home network middleware and data broadcasting middleware. He has been a PhD student at Kyungpook National University since 1997. His research interests are home network middleware, digital broadcasting middleware, IPTV, multimedia systems, and ubiquitous computing.
Sang-Wook Kim received a B.S degree in computer science from Kyungpook National University, Korea in 1979. He received M.S and Ph.D degrees in computer science from Seoul National University, Korea, in 1981 and 1989, respectively. Since 1988, he has been a professor in the Department of Computer Science at Kyungpook National University. His research interests include multimedia systems, multimedia contents security, and human and computer interaction.
Authorized licensed use limited to: Kyungpook National University. Downloaded on July 23,2010 at 01:27:05 UTC from IEEE Xplore. Restrictions apply.
