In this chapter, dependencies between the different protocol layers and across the hops through WMNs will be exploited to deliver video streaming with high QoE. Specifically, a cross-layer optimization approach is applied to a replication strategy for video on-demand over WMNs. Additionally, the authors consider a distributed implementation algorithm, namely Mod&Timer, to handle the placement of replicas for saving storage resources of the network. Simulation results are shown to demonstrate that the proposed cross-layer design outperforms many existing schemes in terms of QoE. More importantly, compared to other strategies without cross-layer interaction, the proposed cross-layer design satisfies the heterogeneous bandwidth constraint of end users.
TopIntroduction
Wireless mesh networks (WMNs) (Akyildiz, et al., 2005) have gained an important role in broadband wireless networks to serve a large number of users having high demands for multimedia services. WMNs rely on multihop wireless backbones consisting of relatively stationary mesh routers and gateways as illustrated in Figure 1. The salient advantages of WMNs, such as self-organization, self-configuration, self-healing, reliability, and scalability, provide an easy and economical network deployment and management. With the rapid growth of multimedia applications, many studies have focused on how to improve the efficiency and the performance of delivering multimedia streaming over WMNs by using feasible techniques such as scheduling, dropping, routing, caching, replication, and cross-layer design.
Figure 1. Components of a typical wireless mesh network
Multimedia streaming applications allow the users to initiate and commence playback simultaneously instead of waiting for downloading to complete. In general, multimedia streaming applications can be subcategorized into three scenarios: 1) On-demand streaming of stored multimedia, e.g., Video on-Demand (VoD), 2) live streaming, e.g., Internet live sport networks, Internet radio stations, and 3) real-time interactive streaming, e.g., on-line games, video conference, E-education. These three scenarios have several QoE objectives in common such as small distortions in the reconstructed application at the receiver, smooth playback, and continuous playback. However, live and real-time interactive streaming has their own characteristics and technical challenges including stringent startup delay constraint and different interactive requirements. In this chapter, we mainly focus on VoD over WMNs and evaluate its QoE performance as well as network resources for this scenario.
Aiming at improving the streaming performance of VoD, caching and replication strategies have been considered as efficient techniques to solve the problem of data accessibility for WMNs (Ghandeharizadeh, et al., 2004; Jin, 2004; Aazami, et al., 2004; Yu, et al., 2005; Hara & Madria, 2006; Jin & Wang, 2005; Ding, et al., 2008). The metrics to evaluate data accessibility performance such as delay, throughput, packet loss, and connectivity are related to service integrity parameters and statistics of the network, medium access control, and physical layers. However, service integrity parameters do not often correlate well with QoE obtained from mechanisms at the application layer such as 1) minimum average reconstructed distortion as quantified by Peak Signal-to-Noise Ratio (PSNR) or for high reconstructed quality playback, 2) minimum fluctuation of reconstructed distortion among received video frames for smooth playback, and 3) in time arrival of received video frames for continuous playback.
In this chapter, we will therefore consider the interrelationships between statistics of the network, medium access control, and physical layers and QoE metrics at the application layer including reconstructed distortion, smooth playback, and continuous playback. Specifically, a cross-layer design is proposed for the replication strategy of VoD streaming over WMNs. The cross-layer design exploits the mutual dependence among protocol layers and across hops to deliver video streaming with high QoE. Moreover, by utilizing the proposed cross-layer design, QoE performance can be optimized by formulating a constrained nonlinear optimization problem in terms of network resources, e.g., bandwidth and storage capacity.