Opportunistic routing in wireless mesh networks

Advances in communication and networking technologies are rapidly making ubiquitous network connectivity a reality. In recent years, Wireless Mesh Networks (WMNs) have already become very popular and been receiving an increasing amount of attention by the research community. Basically, a WMN consists of simple mesh routers and mesh clients, where mesh routers form the backbone of WMN. Due to the limited transmission range of the radio, many pairs of nodes in WMN may not be able to communicate directly, hence they need other intermediate nodes to forward packets for them. Routing in such networks is an important issue and it poses great challenges. Opportunistic Routing (OR) has been investigated in recent years as a way to increase the performance of WMNs by exploiting its broadcast nature. In OR, in contrast to traditional routing, instead of pre-selecting a single specic node to be the next-hop as a forwarder for a packet, an ordered set of nodes (referred to as candidates) is selected as the potential next-hop forwarders. Thus, the source can use multiple potential paths to deliver the packets to the destination. More specically, when the current node transmits a packet, all the candidates that successfully receive it will coordinate with each other to determine which one will actually forward it, while the others will simply discard the packet. This dissertation studies the properties, performance, maximum gain, candidate selection algorithms and multicast delivery issues about Opportunistic Routing in WMNs. Firstly, we focus on the performance analysis of OR by proposing a Discrete Time Markov Chain (DTMC). This model can be used to evaluate OR in terms of expected number of transmissions from the source to the destination. Secondly, we apply our Markov model to compare relevant candidate selection algorithms that have been proposed in the literature. They range from non-optimum, but simple, to optimum, but with a high computational cost. Thirdly, the set of candidates which a node uses and priority order of them have a signicant impact on the performance of OR. Therefore, using a good metric and algorithm to select and order the candidates are key factors in designing an OR protocol. As the next contribution we propose a new metric that measures the expected distance progress of sending a packet using a set of candidates. Based on this metric we propose a candidate selection algorithm which its performance is very close to the optimum algorithm although our algorithm runs much faster. Fourthly, we have investigated the maximum gain that can be obtained using OR. We have obtained some equations that yield the distances of the candidates in OR such that the per transmission progress towards the destination is maximized. Based on these equations we have proposed a novel candidate selection algorithm. Our new algorithm only needs the geographical location of nodes. The performance of our proposal is very close to the optimum candidate selection algorithm although our algorithm runs much faster. Finally, using OR to support multicast is an other issue that we have investigated in this thesis. We do so by proposing a new multicast protocol which uses OR. Unlike traditional multicast protocols, there is no designated next-hop forwarder for each destination in our protocol, thus the delivery ratio is maximized by taking advantage of spatial diversity.Postprint (published version