Design of energy-aware QoS routing algorithm in wireless sensor networks using reinforcement learning

Nowadays a major class of wireless sensor network (WSN) applications required a minimum quality of service parameters to be satisfied while the wireless sensor nodes might be mobile. Most of the standard WSN routing algorithms greedily choose the neighbor node with the best quality of service (QoS) parameter(s) as a next hop. However, the data packet might be able to be routed through other neighbors as it might require less QoS. So the energy of the neighbor node with the best QoS will deplete earlier than other nodes which will result in the reduction of network lifetime. Therefore, it is important for WSN QoS routing protocols to efficiently balance energy and other resources consumption throughout the network. In this paper, we proposed EQR-RL, energy-aware QoS routing protocol in WSNs using reinforcement learning. We compare the network performance of our proposed protocol with three other protocols (QoS-AODV, RSSI and RL-QRP). The packet delivery ratio, average end-to-end delay and impact of the different traffic load on average end-to-end delay are investigated. Simulation results indicate the superiority of our proposed protocol over two others by considering different network traffic load and node mobility in terms of average end-to-end delay and packet delivery ratio.