On the QoS tree construction in WiMAX mesh networks based on genetic algorithm approach

We study the influence of tree's depth and nodes' fan-out on the performance of WiMax mesh networks. For a given tree topology, we first analytically obtain per-node delay and per-node throughput. Then among plenty of tree topologies, extractable from a given network's graph, we search feasible trees which fulfil some per-node and network QoS requirements. Since the searching space is potentially very huge, we use a genetic algorithm in order to explore enough good delay and throughput trade-off. We use the Pruefer code tree representation followed by novel genetic operators. Moreover, by using proper fitness functions, we are able to investigate any desired delay and throughput trade-off in a unified framework. Employing genetic algorithm approach leads to the exploration of extremely wide search space in a reasonably short time, which results in overall scalability and accuracy of the proposed tree exploration algorithm. Due to fast convergence, the proposed genetic algorithm is a good candidate to be implemented in a real-life Base Station (BS) in order to construct adaptive tree topologies based on nodes' traffic demand.