Fuzzy Link Cost Estimation based Adaptive Tree Algorithm for Routing Optimization in Wireless Sensor Networks using Reinforcement Learning

Background and Objective: Internet of things, being a technological advancement in collaboration with traditional wireless sensor networks, are characterized by a highly complex, largescale, dynamically changing, heterogeneous and asymmetric wireless network. These constraints make routing in such wireless sensor networks a challenging task. In this paper, fuzzy link cost estimation based Adaptive Tree routing algorithm (Fuzzy AT) has been introduced which uses the concept of fuzzy logic based link cost estimation, obtained from physical layer and mac layer parameters such as residual energy, packet drop rate, RSSI and total packet receiving time from source node to sink node. The performance of this algorithm has been evaluated with traditional reinforcement learning based algorithms such as real-time search, adaptive tree, ant-based flooded forward routing and Constrained Flooding algorithms on the basis of performance metrics like throughput, latency, energy consumption, energy efficiency and network lifetime. Results and Conclusion: The simulation results reveal that Fuzzy AT algorithm is most appropriate reinforcement learning based routing algorithm among all five algorithms for ensuring energy efficient, reliable, error-free, real-time compatible and QoS aware routing in dynamically changing, asymmetric and unreliable wireless environment of Wireless Sensor Networks.