Wireless Sensor Networks (WSNs) have gained popularity among industries and enterprises especially in fields, such as industrial automation, environmental monitoring as well as surveillance applications; however, the performance of WSN is frequently limited because of the restricted amount of available energy by the sensor nodes. To address this issue, this paper presents a Hybrid Neural based Cluster (HNC) protocol which combines energy-aware cluster-based routing with neural network-driven optimization to overcome the problems associated with sensor nodes. In the first stage of HNC protocol operation, the protocol engages a probabilistic threshold-based mechanism to select cluster heads based upon residual energy to maintain an equilibrium across all cluster heads during operation. Subsequently, a Feed Forward Neural Network (FFNN) model trained off-line using historical data from the network will be employed to improve the quality and efficiency of inter-cluster communication paths. HNC has been tested using MATLAB using simulated data from both LEACH and PEGASIS protocols as a point of reference. The results demonstrate that HNC out performs both conventional protocols described within the scope of this report, providing new benchmarks within WSN's, with regards to first node to fail and overall network life span. The HNC protocol also produces higher throughput and packet delivery ratio (92%). Furthermore, average energy consumed per round has been reduced by 0.025J. All of the results confirm that HNC is an effective method of providing reliable and energy-efficient communications in Wireless Sensor Networks.
