Cooperative agent delay routing approach for manets

Real time and multimedia applications are widely deployed on mobile ad hoc networks in the last few years. Such applications have, often, hard requirements in terms of QoS metrics. We are interested in this paper to end-to-end delay required by some applications. We propose a new adaptive routing approach called based on reinforcement learning. Our approach is built around three parts: delay estimation, routing and flooding optimization. From simulation results on NS-2, we note that our adaptive routing algorithm performs best performance compared to QAODV and DOLSR.