Formal Modeling of Reinforcement Learning Algorithms Applied for Mobile Ad Hoc Network

− A mobile ad hoc network represents a state of system dynamics where the topology changes rapidly. This is further complicated with the necessity of real world mobility abstraction, represented by mobility models. The other major players in this system are wireless links, often supplemented by limitations in terms of bandwidth, link breakage, etc. With the above mentioned characteristics, it a design challenge to achieve efficient routing performance, as it is central to the design of routing protocols. Most of the traditional routing algorithms for mobile ad hoc networks lack the capability to meet all of the above mentioned challenges. Thus we are proposing a new routing algorithm based on reinforcement learning. We use the generalization of prioritized sweeping which optimizes the performance of Q-Routing. We complement this further by using formal methods to model the system as a distributed optimization problem and verify its correctness mathematically. Index Terms − Mobile ad hoc network, Routing Optimization, Prioritized Sweeping, Model Checking.