Multi-objective reinforcement learning based routing in cognitive radio networks: Walking in a random maze

The routing procedure in cognitive radio networks with dynamic spectrum activities is studied. The spectrum statistics are assumed to be unknown. Moreover, the performance is measured using multiple metrics like average delay and packet loss rate. To address the challenges of randomness, uncertainty and multiple metrics, the multi-objective reinforcement learning algorithm is applied for the routing in cognitive radio networks. The effectiveness of the learning procedure is demonstrated by numerical simulations.