Multi-objective Placement of Unmanned Aerial Vehicles as Communication Relays Based on Clustering Method

This work presents a clustering based relay node placement method for using unmanned aerial vehicles as communication relays to from a two-tiered communication network for multiple agents in the operation region. Network performance and relay UAV number are jointly considered at the same level, which turns the problem to be multi-objective relay node placement (MORNP) problem with NP-hard computational complexity. Instead of finding Pareto solutions to the multi-objective optimization problem, this work attempts to achieved balanced network performance and relay UAV number. A one-step approach (OSA) algorithm is proposed, which contains two main aspects: optimal relay position seeking and channel association optimization. A set of initial relay positions are generated by finding the local density peaks of discretization positions in the mission area. It has proved that the channel association matrix is unique with specified relay positions, and the optimal relay positions could be given by solving a nonlinear programming problem. Then the optimal placement result could be achieved by iterating the channel association and optimal relay position seeking progresses. Simulation results validate the feasibility of the proposed OSA algorithm, and comparisons results show that the OSA algorithm requires less computational time than K-means strategy based algorithm, and less relay UAV numbers than the fast search and find density of peaks algorithm.