A Hybrid Routing Algorithm using Artificial Neural Network and Swarm intelligence for Wireless Sensor Networks

In Wireless Sensor Networks, reducing energy consumption is a crucial aspect while designing routing algorithms in order to increase throughput, improve network lifetime, and promise efficient network operations. Herein, a hybrid approach using Artificial Neural Networks (ANN) combined with Particle Swarm Optimization (PSO) is employed to develop an energy-efficient routing algorithm where, ANN with MLP Regressor is trained to find the shortest path and PSO to compute the optimized sink position. Results prove that the proposed ANNMLP-PSO outperforms as compared to Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Grey Wolf Optimization (GWO), and Reposition Particle swarm Optimization (RA-RPSO) for a network size of 50 nodes up to 300 nodes in terms of performance metrics: network lifetime, non-functional sensor nodes, energy usage, and packet delivery ratio. For a network size of 300 nodes, ANNMLP-PSO shows 0.43J of energy consumption whereas GA uses 0.88J, GWO uses 0.72J, PSO uses 0.82J, and RPSO uses 0.70J. At the end of 3000 rounds the count of dead nodes is 195 in the proposed method, while it is 300, 270, 280, and 290 nodes in the other methods. In this method, ANN is used to identify the shortest path between sensor nodes whereas PSO is employed to achieve optimized sink positioning. Due to the optimized sink position, the number of retransmissions and distance between the sink and SNs are decreased which results in overall reduction in energy usage.