A QoS Adaptive Multi-path Reinforcement Learning Routing Algorithm for MANET

The goals of QoS routing are in general twofold: selecting routes with satisfied QoS requirement, and achieving global efficiency in resource utilization. The prediction of these goals in real time is quite difficult, making the effectiveness of "traditional" protocols based on analytical models questionable. In this paper we first discuss some key design considerations in providing QoS routing support, and present a review of previous work addressing the problem of route selection in interaction with QoS constraints. We then devise a solution based on swarm intelligence paradigm based on reinforcement learning approach that we find more adapted for this kind of problems. Finally, we discuss some possible future directions for providing efficient QoS routing mechanisms in wireless ad hoc networks.