Book Description

WMNs comprise nodes that are able to receive and forward the data to other destinations in the networks. Consequently, WMNs are able to dynamically self-organize and self-configure [5]. Each node itself creates and maintains the connectivity with its neighbors. The availability of ad-hoc mode on popular IEEE 802.11 allows low-cost implementation of WMNs. Nevertheless, WMNs have two major drawbacks: interference and scalability as discussed in [6]. (D1) Interference : The independent behaviour and arbitrary deployment of nodes in WMNs can create an extremely high interference environment, which leads to degradation in the quality of wireless connections. For instance, the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) mechanism of IEEE 802.11 (CSMA/CA) has long delays and low resource utilization in dense networks [7]. Recent advancements in physical (PHY) and medium control access (MAC) layers, such as multiple-input multiple-output (MIMO) and multiple channels MAC, can overcome this challenge. The deployment of some solutions are unable in practice because of specific requirements of hardware. Moreover, some implementations such as multiple channel MAC requires high synchronization, which is difficult in WMNs [8]. (D2) Scalability: Multi-hop communication are able to improve coverage and band-width availability in wireless networks [9]. However, it has scalability issues as discussed in [10, 11]. It means that the performance of networks deteriorates significantly when the size of networks grows. PHY layer may experience an extremely noisy medium, thus causing throughput degradation at MAC layer. Moreover, the noisy environment increases the packet loss rate, which impacts significantly to network and transport layers. The existing solutions at PHY or MAC layer can solve the interference problem mentioned in D1. Meanwhile, the scalability of WMNs could be tackled by routing solutions [11]. Routing algorithms are responsible for computing routes so as to convey data through multiple hops until reaching the destinations. As shown in [11], the shortest-path routes, which are the default solutions of conventional routing algorithms, usually have more interference. The solution, subsequently, is finding other routes that have less interference. These routes could be optimal or sub-optimal with given objectives and arguments. The arguments of routing problems comprise of network-oriented metrics and User-oriented metrics. Network-oriented metrics, also called as Quality of Service (QoS) metrics, are derived from the network directly such as bandwidth, delay, jitter, etc. Meanwhile, User-oriented metrics, also called as Quality of Experience (QoE) metrics, are based on users' experience such as mean opinion score (MOS). They represent the level of satisfaction of a users. The good perception of users is the major objective of video streaming services. Most of existing routing algorithms give routing decisions based on single or combination of network-oriented metrics. For example, the routing algorithms in [12, 13, 14] determine routes based on the bandwidth and congestion. Nevertheless, network-oriented metrics may not be well-correlated to users' experience [15, 16, 17, 18]. In other words, users may not be satisfied even with optimal network-oriented metric routes. As a result, it is necessary to develop routing algorithms that take user-oriented metrics into account. This thesis addresses the routing of video streaming over WMNs and proposes novel routing algorithms. These routing algorithms give routing decisions based on the perception of users. To do that, the proposed solution has to address two challenges as follows :(M1) estimate users' perception in real-time and (M2) find optimal or sub-optimal routes efficiently.