Optimal placement of UAVs of an aerial mesh network in an emergency situation

Abstract

There has been a rapid increase in usage of unmanned aerial vehicles (UAVs) in different application areas that are unfriendly to humans. These UAVs have been used in Aerial Mesh Networks (AMNs) that act as backbone network to support communication in a post-disaster scenario. However, there may be limited available number of UAV nodes that need to be utilized efficiently to improve the performance of such networks. Here, we consider three important objectives of the network i.e. target coverage, Quality of Service and energy consumption by the network that need to be optimized efficiently to improve the performance. Yet, it is a grueling task to optimize all of these conflicting objectives at the same time, which is affected by the height of UAVs. To optimize more than one conflicting objectives, we used metaheuristic based multi-objective optimization algorithms i.e. Multi-objective Particle Swarm Optimization (MOPSO), Non-dominated Sorting Genetic Algorithm II (NSGA-II), Strength Pareto Evolutionary Algorithm 2 (SPEA2) and Pareto Envelope-based Selection Algorithm II (PESA-II), which suggest the optimal placement of UAVs. These algorithms are compared based on four performance metrics i.e. generational distance, diversification metric, spread of non-dominant solutions and percentage of domination in three different scenarios. The rigorous experiments are performed by each algorithm in small, medium and large-scale scenarios to compare their results. The ANOVA???s validation test suggests that SPEA2 performs better than others in small-scale scenarios while NSGA-II performs better than others in medium and large-scale scenarios. However, MOPSO has lowest average execution time, after that NSGA-II, then PESA-II and then SPEA2.