Energy Efficient Data Aggregation with Dynamic Mobile Sink-Based Path Optimization in Large Scale WSNs Using Reinforcement Learning

During the past decades, Wireless Sensor Networks (WSNs) have become extensively used due to their prominent number of applications. The use of WSNs is a domineering need for future radical areas commencing from smart home to military surveillance in which hundreds or thousands of sensor nodes are positioned. The usage of mobile sink in those large scale WSNs, for data aggregation amends the functioning of the network by bringing down the energy conservation, amending the network lifetime and data transmission time lag between the nodes. In this paper Center of Energy -Reinforcement Learning based On-Demand Transition State Update algorithm (CERL-ODTST) is proposed to dynamically update mobile sink traversal path. Initially cluster formation and cluster head election for large scale WSNs are done by novel center of energy method. Clustering and data aggregation techniques are applied to reduce the amount of data transmission hence decreasing the energy consumption in the network. In this context cluster heads aggregates data from cluster members which are collected by mobile sinks. The amount of data transmission can be significantly reduced by using Machine Learning algorithms like neural networks and swarm intelligence and also using the distributive features of the network. It offers a reasonable study of the functioning of diverse methods to support the engineers for projecting suitable machine learning based results for grouping the nodes and data aggregation applications. Compared to traditional methods, in the proposed CERL-ODTST, reinforcement learning is used for intra cluster data aggregation to improve aggregation efficiency in the whole network. The implementation results show that proposed CERL-ODTST performs well in terms of overall tour length, energy efficiency and reduces the transmission delay hence increases network lifetime.