Stochastic Gradient Descent Deep Neural Network Classifier for Traffic Aware Resource Efficient Routing in WSN

Abstract

        Congestion in Wireless Sensor Networks (WSN) increases data loss rate and energy consumptionfor data transmission. Few research works are designed in existing works for carried outing traffic and resource optimized routing in WSN with help of different techniques. But, routing performance of conventionaltechniques was poor as congestion control and resource utilization of sensor nodes was not solved effectively when considering large size of WSN. In order to overcome such drawbacks, Stochastic Gradient Descent Deep Neural Network Classification based Routing (SGDDNNC-R) technique is proposed. The SGDDNNC-R technique is designed with the key objective of performingtraffic and resource optimized routing in WSN. The SGDDNNC-R technique introduced Stochastic Gradient Descent Deep Neural Network (SGD-DNN) Classifier algorithm thatincludes multi hidden layersin order to classify sensor node in WSN with less time and better accuracy performance.After node classification, SGDDNNC-R technique performs a route path discovery process.During this process, SGDDNNC-R technique selectsoptimal sensor nodes (i.e. nodes with minimal resource utilization) with minimum distancein order tobroadcast data packets from source to destination node in WSN.  The data transmitted through discovered best route path using SGDDNNC-R technique successfully delivered to destination node in a wireless network without any traffic and also utilizes minimal energy, bandwidth, and end-to-end delay.  Thus, SGDDNNC-R technique achieves congestion aware and resource optimized routing in WSN as compared to state-of-the-art works. The SGDDNNC-R technique performsa simulation process using factors such as energy consumption rate, bandwidth utilization rate and data loss rate and end-to-end delay with respect to a various number of sensor nodes and data packets. The simulation result depicts that the SGDDNNC-R technique is able to minimize the data loss rate and energy consumption of data delivery in WSN as compared to state-of-the-art-works.