Evolutionary Path Planning for Multiple UAVs in Message Ferry Networks Applying Genetic Algorithm

Abstract

Message ferry networks are disconnected wireless networks where one or several mobile nodes, which are called message ferries, travel among stationary and isolated wireless nodes and deliver their messages. This paper presents an evolutionary algorithm to plan the path of multiple UAVs which act as message ferries to deliver messages among isolated wireless nodes. To enable message delivery among isolated nodes, a network model is proposed which allows message exchange between any two nodes in the network by enabling multi-hop delivery of messages through the sink node. Then, the problem of path planning for multiple message ferry UAVs is modeled similarly to the mTSP with a different objective. The genetic algorithm is applied as a heuristic method to find the path planning solution. The proposed genetic algorithm builds clusters of nodes, assigns UAVs to clusters and defines the schedule for UAVs to visit the nodes inside the clusters based on the traffic flows between nodes and the load of messages in nodes to optimize the average weighted delay of message delivery in the network. The results show that the proposed genetic algorithm performs as good as the exhaustive search for small networks but with much shorter execution times. Moreover, it outperforms the traditional mTSP solution in terms of message delivery delay in multi ferry networks.