MOGAMESH: A multi-objective algorithm for node placement in wireless mesh networks based on genetic algorithms

The optimal placement of mesh nodes of an infrastructure wireless mesh network is considered. The optimization is performed w.r.t. two objectives, i.e. installation cost and coverage probability. Mesh nodes with directional antennae are assumed. In this regard, we propose MOGAMESH, a 2-stages multi-objective evolutionary optimization algorithm which tries to optimize the two objectives by means of genetic algorithms, where individuals or solutions are represented by network graphs. In the first stage, candidate network topologies are found by letting a population of graphs evolve. Iteratively, a non-dominated sorting technique classifies individuals which are then selected for the evolution process. Solutions are individuals belonging to the Pareto front, i.e. the set of all non-dominated solutions. In the second stage, a link elimination algorithm further reduces the number of links of the network. In this way, MOGAMESH can provide the network designer with the maximum number of devices to install for every mesh node along with candidate network topologies. We analyze the performance of MOGAMESH for realistic instances of a wireless mesh network with increasing user density.