Energy Efficiency Random Forest Classification based Data Mustering in Wireless Sensor Networks

Abstract

Energy consumption while data transmission in wireless sensor networks is the major issue, which degrades the network lifetime. So the longer the energy, the longer the network lifetime can be achieved. To enhance the network lifetime, have to reduce energy consumption. Data mustering is one of the methods to reduce the energy consumption of sensor nodes to transmit the right data in the right path. The number of research work revealed better energy consumption for data collection in wireless sensor networks. However, the accuracy of data mustering was not increased and energy consumption during data mustering was not reduced. T o surpass such constraints, an Energy Efficiency Random Forest Classification based Data Mustering (EERFCDM) technique is proposed. To perform the classification of nodes, each node of the residual energy is calculated initially. After that, the bagging method is applied in the EERFCDM technique, which enhanced the boundary line between sensor nodes of various classes. The bagging method fusing results of all the models obtained from the bootstrap sample are trained independently. Based on the results, the majority polling considers as final output, which can reduce the misclassification error. After that classification process, primary sensor nodes that are those who have less energy send the data to the secondary sensor nodes those having more energy by calculating Manhattan distance. Later, secondary sensor nodes send the data to the sink node or base station. Consequently, High energy efficiency can be achieved through this method. From the simulation results, compared with existing techniques, the proposed EERFCDM method reduces the energy consumption, delay and data mustering time, meanwhile, the accuracy of data mustering and network lifetime also increased.