Comparative Analysis of Machine Learning Application on Routing Protocols in FANET

Flying ad hoc networks (FANET) are defined by their energetic topology and the absence of fixed infrastructure, which makes efficient routing an important challenge. Traditional routing protocols faced an indeed imposing problem, namely they could not at all efficiently react to network speed fluctuations that often cause topology changing that in turn results into low throughput, high delay, packet loss, and disconnection. This is helped by emerging methods like Machine Learning (ML), which are able to learn by maximizing learning and self-learn. In this paper, we review quite a number of powerful ML techniques and their application to routing, documenting the key strengths and weaknesses of all. We also review how the above stated ML approaches have been applied to distinct current work.