Improving Delay and Jitter Performance in Wireless ... - IEEE Xplore

642

IEEE TRANSACTIONS ON BROADCASTING, VOL. 55, NO. 3, SEPTEMBER 2009

Improving Delay and Jitter Performance in Wireless Mesh Networks for Mobile IPTV Services Bo Rong, Member, IEEE, Yi Qian, Senior Member, IEEE, Mahamat H. Guiagoussou, and Michel Kadoch, Senior Member, IEEE

Abstract—Wireless mesh networking has recently emerged as a promising technology for the next-generation wireless networks. In wireless mesh networks (WMNs), it is practically attractive to support the low-cost quality-of-service (QoS) guaranteed mobile TV service. To meet this need, our study addresses how to improve the delay and jitter performance of mobile IPTV services over IEEE 802.11 based WMN. Particularly, we first discuss the adaptation of IEEE 802.11 MAC layer to construct a WMN with emphasis on mobile IPTV service; we then develop an enhanced version of Guaranteed-Rate (GR) packet scheduling algorithm, namely virtual reserved rate GR (VRR-GR), to further reduce the delay and suppress the jitter in multiservice network environment. Simulation results show that our proposed approach can satisfyingly prioritize mobile IPTV services in WMN, while providing non-IPTV services with what they need as well. Index Terms—Delay, jitter, media access control, mobile IPTV, packet scheduling, quality-of-service, wireless mesh network.

I. INTRODUCTION ECENTLY, there has been considerable interest in mobile TV service provision from the academic and industrial communities [1], [2]. Today, 3G wireless systems support TV distribution with advanced streaming capabilities, either by unicast (e.g., High Speed Packet Access (HSPA)) [3], [4] or broadcast (e.g., Multimedia Broadcast Multicast Service (MBMS)) [5]. There are also alternative pure broadcast technologies, such as Digital Video Broadcasting-Handheld (DVB-H), Digital Multimedia Broadcasting (DMB), Integrated Services Digital Broadcasting-Terrestrial (ISDB-T), and Media Forward Link Only (MediaFLO), booming in different parts of the world [6]–[9]. Nevertheless, the mobile TV solutions stated above require considerable network construction investment from service providers, as well as high monthly charge from users. Therefore, it is practically attractive to develop a low-cost mobile

R

Manuscript received October 07, 2008; revised June 12, 2009. Current version published August 21, 2009. B. Rong is with the Communications Research Centre Canada, Ottawa, ON K2H 8S2, Canada (e-mail: [email protected]). Y. Qian is with the National Institute of Standards and Technology, Gaithersburg, MD 20899-8920, USA (e-mail: [email protected]). M. H. Guiagoussou is with the Mobile Wireless Content Delivery Platform, Sun Microsystems Inc., Montreal, QC H3A 3J6, Canada (e-mail: mahamat. [email protected]). M. Kadoch is with the Department of Electrical Engineering, Ecole de Technologie Superieure, Universite du Quebec, Montreal, QC H3C 1K3, Canada (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TBC.2009.2027739

TV technology. In the past few years, wireless mesh network (WMN) has drawn significant attention from research community and industry as a fast, easy, and inexpensive solution for broadband wireless access [10]–[12]. Motivated by this trend, we propose a novel approach of supporting mobile TV service over IEEE 802.11 based WMNs. Our approach is an extension of IPTV from wired networks to wireless networks, and thus a mobile IPTV technology. Successful deployment of IPTV services over WMNs requires excellent QoS for video and voice, including bandwidth, delay, jitter, packet loss rate, and so on [13]. Previous study showed that the bandwidth requirement could be satisfied by the resource reservation during routing process [14], and the packet loss rate could be guaranteed by application layer forward error correction (AL-FEC) [15]. This paper, thus, mainly addresses the improvement of delay and jitter performance. In this paper, we investigate the mobile IPTV oriented WMN where other types of traffic may also exist. Our work starts with the IEEE 802.11 MAC layer adaptation, so that IPTV services can have more media assess opportunities. Based on the MAC layer design, we further focus on developing a prioritized packet scheduling algorithm, which can give mobile IPTV high preference in multiservice environment and achieve improved delay and jitter performance. To meet the QoS requirements of multimedia services, a number of packet scheduling algorithms have been proposed in the literature, including Virtual Clock, Packet-by-Packet Generalized Processor Sharing (PGPS), Self Clocked Fair Queuing (SCFQ), and so on [16]–[20]. In [21], [22], Goyal et. al. defined the class of Guaranteed-Rate (GR) scheduling algorithms as the ones that allocate a given reserved rate to a flow of packets. GR scheduling algorithms guarantee a delay bound to a packet based on its expected arrival time. It has been proven that Virtual Clock, PGPS, and SCFQ all belong to GR. Conventional GR scheduling algorithms consider mainly the packet scheduling strategy for reserved rate service. This paper studies a multiservice environment, where mobile IPTV is dominating but other types of services are allowed as well. To handle the diversity of multiservice environment, we improve GR scheduling algorithms with the differentiating capability among distinct traffic classes. Particularly, we introduce the concept of virtual reserved rate to conventional GR scheduling algorithms and grant different service levels different virtual sub-capacities. As a result, the WMN gives top priority to serve mobile IPTV services, while other services are also accommodated with their essential needs. The rest of the paper is organized as follows. We first introduce the mobile IPTV deployment over WMN in Section II.

0018-9316/$26.00 © 2009 IEEE

RONG et al.: DELAY AND JITTER PERFORMANCE IN WIRELESS MESH NETWORKS FOR MOBILE IPTV SERVICES

643

Fig. 1. An example of a wireless mesh network.

We then propose an integrated framework of MAC and packet scheduling to reduce delay and suppress jitter in Section III. In Section IV, we review the conventional GR scheduling algorithm. In Section V, we discuss the diversity of services in multiservice network environment and develop a new approach of virtual reserved rate GR (VRR-GR). In Section VI, we present numerical results to demonstrate the advantages of our proposed approach, followed by Section VII to conclude the paper. II. MOBILE IPTV OVER WMN A. Wireless Mesh Networks As shown in Fig. 1, a WMN consists of two types of nodes: mesh routers and mesh clients. The mesh routers form an infrastructure of mesh backbone for mesh clients. In general, mesh routers have minimal mobility and operate just like a network of fixed routers, except being connected by wireless links through wireless technologies such as IEEE 802.11. We can observe from Fig. 1 that, a WMN can access the Internet through a gateway mesh router, which is connected to the IP core network with physical wires. In a WMN, every mesh router is equipped with a traffic aggregation device (similar to an 802.11 access point) that interacts with individual mesh clients. The mesh router relays aggregated data traffic of mesh clients to and from the IP core network. Typically, a mesh router has multiple wireless interfaces to communicate with other mesh routers, and each wireless interface works corresponding to one wireless channel. These wireless channels have different characteristics, because wireless interfaces are running on different frequencies and built on either the same or different wireless access technologies, e.g., IEEE 802.11a/b/g/n. It is also possible that, directional antennas are employed on some interfaces to establish wireless channels over long distance. Unlike many other wireless networks, such as mobile ad hoc networks (MANET) and wireless sensor networks (WSN), WMN in general has the support of infrastructure. In other words, a backbone can be built among wireless mesh routers.

Therefore, it is reasonable to assume that the wireless link between two mesh routers has fixed bandwidth capacity [23]–[25].

B. Mobile IPTV Deployment in WMN IPTV defines the way of provisioning real-time television services over IP networks with various mechanisms implemented to ensure the appropriate level of quality [26]–[31]. Originally, IPTV was proposed to serve the users of fixed terminals, such as set-top box and desktop computer. As the requirement of mobility support rises up, it is an inevitable trend to extend IPTV technology from wired network to wireless network. In this paper, we investigate the mobile IPTV deployment in WMN, which enables roaming users to receive TV programs anywhere on their handhelds or laptop computers. A variety of delivery architectures have been proposed to support IPTV service in wired network. These architectures can be roughly classified into the following categories [32]: 1) native IP multicast; 2) application-level infrastructure overlays advocated by CDN companies, e.g., Akamai; 3) peer-to-peer technologies, such as P2P multicast tree and mesh-pull P2P streaming. So far, in terms of the number of simultaneous users, the most successful IPTV deployments in the Internet utilize mesh-pull P2P streaming architecture. In WMN, however, we consider native IP multicast as the optimal architecture for mobile IPTV service. As an example, in Fig. 1 we could deploy several content servers right behind gateway mesh routers, from which the TV programs are delivered to WMN users using native IP multicast. The major advantage of native IP multicast comes from its capability of saving network resources. Particularly, native IP multicast allows the source to send a packet only once, even if it needs to be delivered to a large number of receivers. It is well known that, native IP multicast is generally not considered as practical in the Internet, because it requires router updating across the whole network [33]. However, native IP multicast becomes a feasible solution in WMN, because the service provider can control all the mesh routers in the network.

644


Fig. 2. An integrated framework of MAC and packet scheduling.

Although we promote native IP multicast as the optimal solution, in the initial stage of deployment, all wired delivery architectures may coexist in WMN, due to the limitation of content resources and user hobbies. III. INTEGRATED FRAMEWORK OF MAC AND PACKET SCHEDULING The packet delay in WMN is mainly caused by MAC layer latency and packet scheduling latency. In the following, we first introduce the enhanced distributed channel access (EDCA), which is standardized in IEEE 802.11e [34] to support multimedia services with tight QoS requirements. We then propose our integrated MAC and packet scheduling framework to provide mobile IPTV services with low delay and jitter in multiservice environment. A. IEEE 802.11 MAC Layer Currently, distributed coordination function (DCF) [35] is the most popular IEEE 802.11 MAC protocol in use. DCF is based on the scheme of carrier sense multiple access with collision avoidance (CSMA/CA), which does not differentiate traffic types. As a result, a station might have to wait for an arbitrarily long time to send a packet, and multimedia services may suffer intolerable delay and jitter. To solve this problem, IEEE 802.11e proposes EDCA as an enhanced version of DCF. EDCA supports the QoS by introducing four access categories (ACs). Each packet arrives at the MAC layer with a priority from higher layer, and is mapped to an AC according to the priority. AC3, AC2, AC1, and AC0 are for voice, video, best effort data, and background traffic, respectively. To differa set entiate the traffic types, EDCA grants AC of specific parameters, including minimum contention window , maximum contention window , and arbitration inter-frame space . With above parameters, the support of QoS can be achieved by differentiating the probability of channel access among different ACs [36]–[38]. B. Integrated Framework This paper focuses on providing mobile IPTV service in WMN. As shown in Fig. 2, we develop an integrated frame-

work, which involves both MAC and packet scheduling to reduce the delay and suppress the jitter of mobile IPTV services. In the integrated framework, we begin with the rearrange of EDCA in MAC layer, so that mobile IPTV service could be given the highest priority. Based on original EDCA, we propose an IPTV-EDCA, which places mobile IPTV services in AC3 and non-IPTV voice and video services in AC2 while keeping AC1 and AC0 as same as in the IEEE 802.11e standard. In addition to MAC layer, packet scheduling is another issue that can significantly influence the delay and jitter performance. Regarding the ACs of IPTV-EDCA, we define four scheduling categories (SCs), i.e., SC3, SC2, SC1, and SC0 in packet scheduling. To grant different SCs with different priorities, we extend the original GR packet scheduling algorithm to VRR-GR algorithm. Next, we review the original GR and propose the VRR-GR in Sections IV and V, respectively. IV. BASICS OF GR PACKET SCHEDULING ALGORITHMS GR packet scheduling algorithms were firstly defined by Goyal et. al. in their seminal work on determining end-to-end delay bound [21]. Assuming that packet is the unit of data transmission at the network level, we refer to the sequence of packets transmitted by a source as a flow. Each packet of a flow is then served by a sequence of network nodes (switching elements, i.e., mesh routers in WMN) along the path from the source to the destination. Consider a general case where flow is switched by network node . To provide guaranteed performance, network node al(in bits/second) as requested locates flow a reserved rate by the source of flow . Network node is called a GR network node if it complies with the service discipline of work-conserving and non-preemptive and uses the Guaranteed-Rate clock value of a packet as scheduling priority. In other words, when GR network node switches a new packet, the packet in queue with the smallest Guaranteed-Rate clock value is selected for service. Let and represent the th packet of flow and its length, respectively. Let represent the arrival time of packet at GR network node . Then, Guaranteed-Rate clock value for


packet by

at network node , denoted by

, is given

645

the path of a flow belongs to GR, then the end-to-end delay of is bounded by packet (5)

(1) and . where With the definition of Guaranteed-Rate clock value, a scheduling algorithm at network node belongs to class GR for flow , if it guarantees that packet will be transmitted by . Here, is a constant which depends on the scheduling algorithm and the network node. For example, , the virtual clock algorithm belongs to GR with and represent the capacity of network node where and the maximum length of packet served by network node , respectively. As shown in [21], [22], strict delay bounding is a salient feabe the total ture of GR packet scheduling algorithms. Let number of network nodes along the path of a flow, be the th network node on the path, network node 0 be the source, and netbe the destination. Since the packet arrives at work node and network node guarthe first network node at time will be transmitted by antees that packet , the end-to-end delay of , denoted by , is bounded by

As a statistical relaxation of leaky bucket, exponentially bounded burstiness (EBB) process characterization has been proposed in [43]. A flow conforms to EBB if the deviation probability of a source from the average rate decreases expobe the prefactor and be the decay rate of nentially. Let the exponential decay function, then a flow is an EBB process , if with parameters

(6) If flow conforms to EBB with parameters and the scheduling algorithm at each of the network nodes on the path of a flow belong to GR, then the end-to-end delay of packet is bounded by

(7) (2) is the propagation delay between network node where and the destination. depends on From (1) we observe that , which in turn depends on by . Applying can be replaced with this argument recursively, in (2) by

In addition to delay, jitter is another important issue to multimedia services [44]. Jitter can be derived from delay by (8) Since , , we have . The value depends on and thus varies from time to time. Let of , we then define as a random variable ranged in with probability density function (pdf) . As a result, the average jitter of flow is given by

(9)

(3) Since is completely determined by the traffic arrival characteristics of the source and the rate associated with the flow, the end-to-end delay can be determined if source specification is known. For instance, leaky bucket is a source traffic specification that bounds the maximum deviation from the average rate [39]–[42]. be a function that denotes the bits of flow that Let . A flow conforms to leaky bucket arrive in the interval and average rate if with burst size

(4) If flow conforms to a leaky bucket with parameters , and the scheduling algorithm at each of the network node on

V. VRR-GR PACKET SCHEDULING ALGORITHM FOR MULTISERVICE NETWORKS A. Architecture Design Multiservice networks are replacing older generation of single service networks. Instead of using three different networks for data, voice, and video, a multiservice network is an all-in-one platform for broadband, phone, TV services. Our study focuses on a typical multiservice environment, i.e., mobile IPTV in WMN with other possible services. Particularly, we consider that the services may have various QoS requirements, and thus the WMN has to support four scheduling categories: SC3 for mobile IPTV, SC2 for non-IPTV voice/video, SC1 for best effort data, and SC0 for background traffic.

646


Fig. 3. The architecture of extended GR packet scheduling.

Traditional GR scheduling algorithms address mainly the reserved rate service. To accommodate the diversity of multiservice environment, we emphasize the differentiating capability during package scheduling process in this paper. Fig. 3 demonstrates the architecture of extended GR packet scheduling architecture. It shows that four categories of services, are scheduled together in the four priority queues, where the Guaranteed-Rate clock value of a packet is utilized as scheduling priority. The architecture in Fig. 3 is suitable for both unicast and multicast. The only difference is the scheduled packet will be transferred to one outgoing interface for unicast, but perhaps multiple outgoing interfaces for multicast. Above difference does not affect the performance of VRR-GR algorithm. Let be the total capacity of the network node , the capacity allocated to different service levels can then be represented by , where , , , and are the sub-capacity allocated to SC3, SC2, SC1, and SC0, respectively. For SC3 and SC2, each connection is viewed as a flow and guaranteed with a fixed bandwidth (the reserved rate as required by the flow). For SC1 and SC0, the fixed bandwidth is allocated to all the connections, thus each connection can only share a part of it, which may vary from time to time. Therefore, one contribution of our approach is to aggregate low-priority traffic into low-priority packet scheduling classes and share the bandwidth of the class among the constituents. It is worth noting that differentiated services or DiffServ also addresses the traffic aggregation and service differentiation. However, DiffServ elaborates mainly on the per-hop behavior (PHB) with the differentiated

service levels incompatible to IEEE 802.11 MAC layer. Furthermore, it still needs a mature underlying packet scheduling algorithm to guarantee the desired PHB in DiffServ. B. Virtual Reserved Rate GR Scheduling Algorithm We develop a virtual reserved rate GR (VRR-GR) scheduling algorithm to determine the Guaranteed-Rate clock value of each flow in the priority queue of Fig. 3. The major difference between VRR-GR and conventional GR is that the former uses virtual reserved rate instead of real reserved rate to calculate the Guaranteed-Rate clock value. Here, the virtual reserved rate comes from the concept of virtual sub-capacity, which is defined as (10) where , , and represent the virtual sub-capacity, real on network sub-capacity, and virtual offset-capacity of node , respectively. Bandwidth and delay bound are two important aspects to be considered when designing VRR-GR algorithm. Real sub-capacity decides the bandwidth that a flow finally obtains, and virtual sub-capacity decides the induced delay during packet scheduling. Using conventional GR as a reference, virtual offset-capacity explains the additional reserved rate that VRR-GR grants to a flow during packet scheduling. It is worth noting that virtual offset-capacity is introduced to influence the delay bound but not the bandwidth of a flow.


In VRR-GR, virtual sub-capacity and virtual offset-capacity must conform to the following constraint (11) In our design, virtual offset-capacities, i.e., , , , , serve as priority indicator in VRR-GR algorithm. Larger vir, , tual offset-capacity means higher scheduling priority. , may take positive or negative values. Positive/negative value results in lower/higher delay bound than conventional GR can achieve. In practice, we usually configure the parameters as , , , , so that mobile IPTV service can borrow some packet scheduling resource from best effort data and background traffic. Assuming the reserved rate of a SC3 or SC2 flow is , then for SC3, the virtual reserved rate of this flow is for SC2. For SC1 or SC0, we consider all or the connections in that service category as one flow. That is, the for SC1, or virtual reserved rate is for SC0. In this paper, we set , which means that the SC0 packets will be scheduled only when the network node is idle. Moreover, inside the SC0 packets, network node employs the strategy of first in first out (FIFO). As shown in Fig. 3, we assume that the packets are shaped and policed before entering the WMN, so that the following admission control rule is guaranteed: (12) where denotes the set of flows that are active at time . It is clear that, if above condition holds, VRR-GR scheduling algorithm can provide SC1 and SC0 with their real sub-capaciand , respectively. ties, i.e., C. Delay and Jitter Gain of VRR-GR Scheduling Algorithm It is a salient feature that VRR-GR scheduling algorithm can provide mobile IPTV service with lower delay bound than conventional GR. Using (2) the delay bound of VRR-GR scheduling algorithm is given by

(13) Accordingly, we define the delay gain of VRR-GR scheduling algorithm on mobile IPTV service as

(14) In addition to delay, VRR-GR can suppress jitter as well. Let , then the jitter of VRR-GR scheme, rep, can be defined as a random variable ranged resented by

647

with probability density function . We furin ther assume that the probability density functions of conventional GR and VRR-GR are subject to the property of scaling . It then results similarity, i.e., , which implies a jitter suppression in . from VRR-GR as VI. SIMULATION RESULTS In this section, we present numerical results to demonstrate the advantages of our proposed integrated MAC and packet scheduling framework. Particularly, we first study the performance of VRR-GR scheduling algorithm to find the appropriate parameter configuration for mobile IPTV in multiservice environment. We then evaluate the overall performance of our integrated framework by combining VRR-GR with IPTV-EDCA. As discussed in previous sections, GR packet scheduling could be practically implemented in the form of Virtual Clock, PGPS, or SCFQ. In our simulation study, we employ Virtual Clock to obtain the numerical results of conventional GR and VRR-GR. A. Performance of VRR-GR Scheduling Algorithm We have implemented a simulation using OPNET Modeler 11.0 to evaluate the performance of VRR-GR through a line of nodes (mesh routers in WMN). Fig. 4 illustrates that our OPNET model consists of three classes of components, , VRR-GR nodes, and including traffic sources a sink. The traffic sources generate packets according to special representing the source of mobile profiles, with IPTV, non-IPTV voice/video services, best effort data, and VRR-GR nodes are background traffic, respectively. The programmed to be identical and to schedule the packets using VRR-GR scheduling algorithm according to the configuration. The sink node is placed at the end of the line to recycle the packets. Fig. 5 demonstrates the average packet delay of while configuring mobile IPTV and non-IPTV voice/video services as leaky bucket traffic sources. In this simulation scenario, we characterize the OPNET model in Fig. 4 with the following features. nodes, 1) VRR-GR nodes: The simulation involves where each of them has a total capacity of 500 Mbps, with 50% allocated to SC3, 15% allocated to SC2, 30% allocated to SC1, and 5% allocated to SC0. Moreover, we vary , the virtual offset-capacity of SC3, from 25 Mbps to 175 Mbps, while keeping and . , we know Due to the constraint is varying from 0 Mbps to 150 Mbps accordingly. 2) Traffic source: and are configured to be leaky bucket traffic sources as shown in Tables I and II. consists of 30% of 500 Mbps traffic, which includes ftp, email, and http with packet size randomly ranged in (0.5 Kbytes, consists of 5% of 500 Mbps traffic, 1.2 Kbytes). which includes probing, backscatter, and misaddressed traffic with packet size randomly ranged in (0.5 Kbytes, 1.0 Kbytes). 3) Others: To simplify the simulation, we assume that the propagation delay of each link is 0 second.

648


Fig. 4. OPNET simulation model of VRR-GR packet scheduling algorithm.

Fig. 5. Average packet delay through a line of VRR-GR nodes with leaky bucket traffic sources.

TABLE I TRAFFIC LOAD CONFIGURATION OF T (LEAKY BUCKET TRAFFIC THAT OCCUPIES 50% OF 500 Mbps)

TABLE II TRAFFIC LOAD CONFIGURATION OF T (LEAKY BUCKET TRAFFIC THAT OCCUPIES 15% OF 500 Mbps)

Fig. 5 shows that mobile IPTV services have lower increases. This phenomenon implies packet delay as VRR-GR has better performance than conventional GR, noting that VRR-GR degenerates to conventional GR when . Moreover, as grows from 25 Mbps to 175 Mbps, the packet delay of non-IPTV voice/video and background traffic keeps almost the same, since

Fig. 6. Average packet jitter through a line of VRR-GR nodes with leaky bucket traffic sources.

and are fixed to be 0 Mbps and 25 Mbps respectively; increases, as is the packet delay of best effort data varying from 0 Mbps to 150 Mbps. In practice, we usually , so that best effort data can have set virtual sub-capacity. Besides packet delay, jitter is another important QoS parameter. Fig. 6 illustrates the average jitter of with the same simulating configuration that generates Fig. 5. It reveals that the jitter of is suppressed considerably as grows. We then conclude that VRR-GR can provide mobile IPTV services with better jitter performance than conventional GR. Similar to Figs. 5–8 demonstrate the OPNET simulation results of delay and jitter using EBB traffic source. This simulation uses the parameter configuration the same as Figs. 5 and 6, except that leaky bucket traffic sources are changed to EBB . Figs. 7 and 8 traffic sources with show that EBB traffic source incurs larger delay and jitter than leaky bucket traffic source, since the former is not regulated as strictly as the leaky bucket source. However, in case of EBB, it still holds that VRR-GR has better performance than conventional GR in terms of delay and jitter. In summary, Figs. 5–8 show that in a network of five nodes is a


649

Fig. 7. Average packet delay through a line of VRR-GR nodes with EBB traffic sources.

Fig. 9. Joint MAC layer and packet scheduling delay/jitter of mobile IPTV services while varying the number of network nodes.

Fig. 8. Average packet jitter through a line of VRR-GR nodes with EBB traffic sources.

suitable parameter configuration to achieve low delay and jitter for IPTV services while satisfying non-IPTV services with their requirements as well. This conclusion from five node network is also true for the network of other size, although the numerical results are not shown here due to the limit of space. B. Performance of Integrated MAC and Packet Scheduling Framework Finally, we demonstrate the advantage of our proposed integrated MAC and packet scheduling framework for mobile IPTV service. We assume that in Fig. 4 Node 1 is a traffic aggregation device, which utilizes 802.11 MAC layer to collect traffic . Once the traffic is aggregated, it is transmitted through all the network nodes (mesh routers in WMN) using a certain packet scheduling algorithm until reaching the sink. In our simulation study, we compare the performance of two transmission schemes: 1) DCF GR: DCF in MAC layer and conventional GR in packet scheduling (conventional GR is equal to VRR-GR of zero virtual offset-capacities), 2) IPTV-EDCA+VRR-GR: IPTV-EDCA in MAC layer and

VRR-GR in packet scheduling. Specifically, we assume that in Fig. 4 the number of nodes varies from three to nine; we configure MAC layer protocols, such as DCF and IPTV-EDCA, with the recommended parameter values from standards [34], [35]; we configure VRR-GR nodes with , whereas other parameters, such as traffic load, node capacity, etc., are the same as in previous leaky bucket simulation scenario. Numerical results are presented in Fig. 9 to demonstrate the delay and jitter performance of mobile IPTV service for both IPTV-EDCA+VRR-GR and DCF+GR, while varying the number of nodes in Fig. 4 from three to nine. It shows that our proposed IPTV-EDCA+VRR-GR framework can provide mobile IPTV services with better delay and jitter performance than conventional DCF+GR approach. To further compare IPTV-EDCA+VRR-GR and DCF+GR, Fig. 10 demonstrates the delay and jitter gains for mobile IPTV services from MAC layer and packet scheduling, respectively. To clearly present the numerical results, we define MAC layer and packet scheduling delay gain ratio delay gain ratio as

(15)

where and represent the MAC layer delay of DCF and represent the and IPTV-EDCA respectively, packet scheduling delay of conventional GR and VRR-GR respectively.

650


Fig. 10. MAC layer and packet scheduling delay/jitter gain ratio for mobile IPTV services while varying the number of network nodes.

Likewise, we define MAC layer jitter gain ratio packet scheduling jitter gain ratio as

and

(16)

where and represent the MAC layer jitter of DCF and and represent the packet IPTV-EDCA respectively, scheduling jitter of conventional GR and VRR-GR respectively. Fig. 10 shows that IPTV-EDCA VRR-GR has steady delay and jitter gain ratio against DCF GR from MAC layer and packet scheduling. The MAC layer gains keep stable regardless of the number of network nodes, because MAC layer interaction happens only at the first mesh router. The packet scheduling gains are quite stable as well, due to the robustness of VRR-GR algorithm. Moreover, as the number of network nodes grows, it is clear that packet scheduling will play more and more important role in terms of delay and jitter. In other words, VRR-GR has practical significance especially in large-scale WMN. VII. CONCLUSIONS In this paper, we have addressed the support of mobile IPTV service over IEEE 802.11 based WMNs. Specifically, we have proposed an integrated framework that combines MAC and packet scheduling to reduce the delay and jitter of mobile IPTV services. Within the framework, we have proposed a VRR-GR packet scheduling algorithm to further prioritize mobile IPTV in multiservice environment. Simulation study demonstrates that our proposed approach can achieve satisfying delay and jitter performance in mobile IPTV over WMN. REFERENCES [1] D. Sandham, “Mobile TV proves a hit,” IEE Communications Engineer, vol. 4, no. 1, p. 12, Feb./Mar. 2006.

[2] D. Sandham, “What next for mobile TV?,” IET Communications Engineer, vol. 4, no. 2, pp. 10–13, Apr./May 2006. [3] D. Mulvey, “HSPA,” IET Communications Engineer, vol. 5, no. 1, pp. 38–41, Feb./Mar. 2007. [4] S. Aissa and G. Aniba, “Queuing models for dimensioning interactive and streaming services in high-speed downlink packet access networks,” IEEE Trans. Broadcasting, vol. 53, no. 3, pp. 619–627, Sep. 2007. [5] A. M. C. Correia, J. C. M. Silva, N. M. B. Souto, L. A. C. Silva, A. B. Boal, and A. B. Soares, “Multi-resolution broadcast/multicast systems for MBMS,” IEEE Trans. Broadcasting, vol. 53, no. 1, pp. 224–234, Mar. 2007. [6] M. Kornfeld and G. May, “DVB-H and IP datacast-broadcast to handheld devices,” IEEE Trans. Broadcasting, vol. 53, no. 1, pp. 161–170, Mar. 2007. [7] S.-J. Lee, S. W. Lee, K.-W. Kim, and J.-S. Seo, “Personal and mobile satellite DMB services in Korea,” IEEE Trans. Broadcasting, vol. 53, no. 1, pp. 179–187, Mar. 2007. [8] G. Bedicks, Jr., F. Yamada, F. Sukys, C. E. S. Dantas, L. T. M. Raunheitte, and C. Akamine, “Results of the ISDB-T system tests, as part of digital TV study carried out in Brazil,” IEEE Trans. Broadcasting, vol. 52, no. 1, pp. 38–44, Mar. 2006. [9] M. R. Chari, F. Ling, A. Mantravadi, R. Krishnamoorthi, R. Vijayan, G. K. Walker, and R. Chandhok, “FLO physical layer: An overview,” IEEE Trans. Broadcasting, vol. 53, no. 1, pp. 145–160, Mar. 2007. [10] I. F. Akyildiz and X. Wang, “A survey on wireless mesh networks,” IEEE Communications Magazine, vol. 43, no. 9, pp. S23–S30, Sep. 2005. [11] A. Raniwala and T. Chiueh, “Architecture and algorithms for an IEEE 802.11-based multi-channel wireless mesh network,” in IEEE INFOCOM 2005, Mar. 2005, vol. 3, pp. 2223–2234. [12] H. Jiang, W. Zhuang, X. Shen, A. Abdrabou, and P. Wang, “Differentiated services for wireless mesh backbone,” IEEE Communications Magazine, vol. 44, no. 7, pp. 113–119, Jul. 2006. [13] Y. Xiao, X. Du, J. Zhang, F. Hu, and S. Guizani, “Internet protocol television (IPTV): The killer application for the next-generation internet,” IEEE Communications Magazine, vol. 45, no. 11, pp. 126–134, Nov. 2007. [14] X. Chen, B. Li, and Y. Fang, “A dynamic multiple-threshold bandwidth reservation (DMTBR) scheme for QoS provisioning in multimedia wireless networks,” IEEE Trans. Wireless Communications, vol. 4, no. 2, pp. 583–592, Mar. 2005. [15] [9] ETSI TS 102 034 v1.3.1, “Transport of MPEG 2 Transport Stream (TS) Based DVB Services over IP Based Networks,” DVB Blue Book A086rev5, Oct. 2007 [Online]. Available: http://www.dvb.org/technology/bluebooks [16] L. Zhang, “Virtual clock: a new traffic control algorithm for packetswitching networks,” in Proc. ACM SIGCOMM’90, pp. 19–29. [17] H. Zhang and S. Keshav, “Comparison of rate-based service disciplines,” in Proc. ACM SIGCOMM’91, pp. 113–121. [18] A.K. Parekh, “A Generalized Processor-Sharing Approach to Flow Control in Integrated Services Networks,” PhD thesis, Department of Electrical Engineering and Computer Science, MIT, Cambridge, Mass, 1992. [19] S. J. Golestani, “A self-clocked fair queueing scheme for high speed applications,” in Proc. INFOCOM’94, 1994. [20] Q. Zheng and K. Shin, “On the ability of establishing real-time channels in point-to-point packet-switching networks,” IEEE Trans. Communications, vol. 42, pp. 1096–1105, Mar. 1994. [21] P. Goyal, S. S. Lam, and H. M. Vin, “Determining end-to-end delay bounds in heterogeneous networks,” in The 5th International Workshop on Network and Operating System Support for Digital Audio and Video, 1995, pp. 273–284. [22] P. Goyal and H. M. Vin, “Generalized guaranteed rate scheduling algorithms: A framework,” IEEE/ACM Trans. Networking, vol. 5, no. 4, pp. 561–571, Aug. 1997. [23] H. Zhu, M. Li, I. Chlamtac, and B. Prabhakaran, “A survey of quality of service in IEEE 802.11 networks,” IEEE Wireless Communications, vol. 11, no. 4, pp. 6–14, Aug. 2004. [24] B. Rong, Y. Qian, K. Lu, and R. Q. Hu, “Enhanced QoS multicast routing in wireless mesh networks,” IEEE Trans. Wireless Communications, vol. 7, no. 6, pp. 2119–2130, Jun. 2008. [25] D. Wu and R. Negi, “Effective capacity: A wireless link model for support of quality of service,” IEEE Trans. Wireless Communications, vol. 2, no. 4, pp. 630–643, Jul. 2003.


[26] U. Jennehag, T. Zhang, and S. Pettersson, “Improving transmission efficiency in H.264 based IPTV systems,” IEEE Trans. Broadcasting, vol. 53, no. 1, pt. 1, pp. 69–78, Mar. 2007. [27] S. Chand and H. Om, “Modeling of Buffer storage in video transmission,” IEEE Trans. Broadcasting, vol. 53, no. 4, pp. 774–779, Dec. 2007. [28] H. Joo, H. Song, D.-B. Lee, and I. Lee, “An effective IPTV channel control algorithm considering channel zapping time and network utilization,” IEEE Trans. Broadcasting, vol. 54, no. 2, pp. 208–216, Jun. 2008. [29] W. Sun, K. Lin, and Y. Guan, “Performance analysis of a finite duration multichannel delivery method in IPTV,” IEEE Trans. Broadcasting, vol. 54, no. 3, pt. 1, pp. 419–429, Sep. 2008. [30] W. Li, T. Herfet, C. Jacquenet, H. Liu, J. Maisonneuve, and T. Stockhammer, “From the guest editors: IPTV in multimedia broadcasting,” IEEE Trans. Broadcasting, vol. 55, no. 2, pp. 311–314, Jun. 2009. [31] R. Sharpe, J. Heiles, H. Liu, M. Deschanel, Y. Wu, J. Maisonneuve, and W. Li, “An overview of IPTV standards development,” IEEE Trans. Broadcasting, vol. 55, no. 2, pp. 315–328, Jun. 2009. [32] X. Hei, C. Liang, J. Liang, Y. Liu, and K. W. Ross, “A measurement study of a large-scale P2P IPTV system,” IEEE Trans. Multimedia, vol. 9, no. 8, pp. 1672–1687, Dec. 2007. [33] A. El-Sayed, V. Roca, and L. Mathy, “A survey of proposals for an alternative group communication service,” IEEE Network, vol. 17, no. 1, pp. 46–51, Jan./Feb. 2003. [34] Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications Amendment: Medium Access Control (MAC) Quality of Service Enhancements, IEEE Standard 802:11e, IEEE 802.11 WG, Nov. 2005. [35] Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specification, , IEEE 802.11 WG, Aug. 1999, IEEE Standard. [36] H. Jiang, P. Wang, and W. Zhuang, “A distributed channel access scheme with guaranteed priority and enhanced fairness,” IEEE Trans. Wireless Communications, vol. 6, no. 6, pp. 2114–2125, Jun. 2007. [37] P. Wang, H. Jiang, and W. Zhuang, “IEEE 802.11e enhancement for voice service,” IEEE Wireless Communications, vol. 13, no. 1, pp. 30–35, Feb. 2006. [38] S. Kim, R. Huang, and Y. Fang, “Deterministic priority channel access scheme for QoS support in IEEE 802.11e wireless LANs,” IEEE Trans. Vehicular Technology, vol. 58, no. 2, pp. 855–864, Feb. 2009. [39] M. Butto, E. Cavallero, and A. Tonietti, “Effectiveness of the “leaky bucket” policing mechanism in ATM networks,” IEEE Journal on Selected Areas in Communications, vol. 9, no. 3, pp. 335–342, Apr. 1991. [40] K. Sohraby and M. Sidi, “On the performance of bursty and modulated sources subject to leaky bucket rate-based access control schemes,” IEEE Trans. Communications, vol. 42, no. 234, pt. 1, pp. 477–487, Feb.–Apr. 1994. [41] V. Anantharam and T. Konstantopoulos, “Burst reduction properties of the leaky bucket flow control scheme in ATM networks,” IEEE Trans. Communications, vol. 42, no. 12, pp. 3085–3089, Dec. 1994. [42] N. Yamanaka, Y. Sato, and K. Sato, “Performance limitation of the leaky bucket algorithm for ATM networks,” IEEE Trans. Communications, vol. 43, no. 8, pp. 2298–2300, Aug. 1995. [43] O. Yaron and M. Sidi, “Generalized processor sharing networks with exponentially bounded burstiness arrivals,” in IEEE INFOCOM ’94, Jun. 1994, pp. 628–634. [44] S. R. Gulliver and G. Ghinea, “The perceptual and attentive impact of delay and jitter in multimedia delivery,” IEEE Trans. Broadcasting, vol. 53, no. 2, pp. 449–458, Jun. 2007. Bo Rong (M’07) received the B.S. degree from Shandong University in 1993, the M.S. degree from Beijing University of Aeronautics and Astronautics in 1997, and the Ph.D. degree from Beijing University of Posts and Telecommunications in 2001. He is currently a Research Scientist with Communications Research Centre Canada, Ottawa, ON. He is also an Adjunct Professor at Ecole de technologie superieure (ETS), Universite du Quebec, Canada. His research interests include modeling, simulation, and performance analysis of next-generation wireless networks.

651

Yi Qian (M’95–SM’07) received a Ph.D. degree in electrical engineering with a concentration in telecommunication networks from Clemson University, in Clemson, South Carolina. He is with the National Institute of Standards and Technology, in Gaithersburg, Maryland. His current research interests include information assurance, network security, network management, network design, network modeling, simulation and performance analysis for next generation wireless networks, wireless sensor networks, broadband satellite networks, optical networks, high-speed networks and the Internet. He has publications and patents in all these areas. He was an assistant professor in the Department of Electrical and Computer Engineering, University of Puerto Rico at Mayaguez (UPRM) between July 2003 and July 2007. At UPRM, he taught courses on wireless networks, network design, network management, and network performance analysis. His research and curriculum development efforts were funded among others by National Science Foundation, General Motor, IBM, and PRIDCO, with more than $2 millions in total award amount during the four years at UPRM. Prior to joining UPRM in July 2003, he worked for several start-up companies and consulting firms, in the areas of voice over IP, fiber optical switching, Internet packet video, network optimizations, and network planning as a technical advisor and a senior consultant. He also worked several years for the Wireless Systems Engineering Department, Nortel Networks in Richardson, Texas as a senior member of scientific staff and as a technical advisor. While at Nortel, he was a project leader for various wireless and satellite network product design projects, customer consulting projects, and advanced technology research projects. He was also in charge of a wireless standard development and evaluation project in Nortel. Dr. Yi Qian is a member of ACM and a senior member of IEEE.

Mahamat H. Guiagoussou received the B.S. degree and M.S. degree from University Of Yaounde (Cameroon) in 1987 and 1988 respectively. He is the lead customer engineer of mobile wireless content delivery platform at Sun Microsystems Inc. His current research interests focus on Java technology, mobile agent, and mobile ad hoc network.

Michel Kadoch (S’67–M’77–SM’04) received the B. Eng from Sir George Williams University in 1971, the M. Eng from Carleton University in 1974, MBA from McGill University in 1983, and the Ph.D from Concordia University in 1991. He is a full professor at Ecole de technologie superieure (ETS), Universite du Quebec, Canada. He is also an adjunct professor at Concordia University, Montreal, Quebec, Canada. Prof. Kadoch is serving as reviewers for a number of journals and conferences, as well as for NSERC grants. His current research interests include crosslayer design, reliable multicast in wireless ad hoc and WiMAX networks. He has publications and patents in all these areas.