A space-efficient wireless mesh network : architecture, frequency planning and routing

A pure multi-hop Wireless Mesh Network (WMN) is well known to be not scalable, because the space reuse efficiency drops severely with an increasing number of concurrent transmissions, as long as all the transmissions share the same spectrum. In this thesis, we introduce a scalable wireless mesh network, equipped with directional antennas that operate on three orthogonal channels to partition the network into multiple adjacent interference-free regions, in order to enhance the spatial utilization efficiency. We show both in theory and by simulation that, this WMN architecture performs significantly better than a pure multi-hop WMN. Based on this architecture, we study two peripheral problems: 1. Since the proposed WMN requires precise deployment of Access Point (AP) nodes, we design an algorithm to compensate the performance degradation if the locations of AP nodes can not be exactly as proposed. 2. Intelligent utilization of partially overlapping channels produces higher efficiency in resource allocation and may be an option when orthogonal channels are not accessible; we analyze when and how we are able to apply partially overlapping channels in this WMN. Noticing that none of the existing routing algorithms suits the proposed WMN well, we also design a routing method named Location-Hashed Link State (LHLS) routing, which depends on the AP nodes to cache the position and routing information of mobile clients for mobility support. Using LHLS, the amount of broadcasted routing control packets is lowered. Furthermore, LHLS provides very smooth AP association exchange for a mobile client node moving across adjacent AP regions even at a high mobility rate, and therefore causes less packet loss during the AP handover. Simulation results demonstrate the performance advantages of LHLS over AODV in all respects of throughput capacity, packet delay and control overheads