Cognitive Network Management with Reinforcement Learning for Wireless Mesh Networks

We present a framework of cognitive network management by means of an autonomic reconfiguration scheme. We propose a network architecture that enables intelligent services to meet QoS requirements, by adding autonomous intelligence, based on reinforcement learning, to the network management agents. The management system is shown to be better able to reconfigure its policy strategy around areas of interest and adapt to changes. We present preliminary simulation results showing our autonomous reconfiguration approach successfully improves the performance of the original AODV protocol in a heterogeneous network environment.