A Novel Data Gathering Approach Using Machine Learning Technique In Wireless Sensor Networks

Abstract

In recent years, intensive research is performed in the field of newlineWireless Sensor Networks (WSN), especially for monitoring, newlinecharacterizing, and tracking large physical environments and conditions newlinesuch as, temperature, wind, pressure, and humidity. Due to its newlineversatility, it has received significant attention and has been deployed in newlineapplications such as natural disaster relief, military target tracking, newlinehazardous environment exploration and wildlife monitoring. To process newlinedata at remote central unit (e.g., sink), many of these application use newlinesensors to periodically sense and send the sensory information. After newlinedeployment, in many of these applications, the critical task is in newlineclassifying the huge amount of sensed data and maintaining the sensors. newlineTherefore energy efficient data collection protocol has become utmost newlineimportance in these networks. With restricted energy storage capability newlineof sensors and bandwidth scarcity, it is crucial to jointly consider newlineguaranteed data delivery incorporated with efficient data gathering in newlineWSN. Moreover, challenging deployment environments pose intricacies newlinein reliable data transmission in these networks. newlineThough many improvements are done, lifetime of the network newlineremains a major issue in WSN technology. To develop a performancecentric newlineautomated system, the work in this dissertation dedicatedly newlineconcentrates on building a smart network management with reconfigurable newlinesensors. It primarily focuses in achieving successful data newlinegathering for critical applications using genetic algorithm and machine newlinelearning techniques. newlinevi newlineThe first stage of the dissertation involves implementing a newlinenovel quotProactive Event and Time driven (Pro_ET)quot protocol that newlineexploits on-demand reconfigurable smart collectors for efficient energy newlinemanagement using weighted fairness queuing (WFQ) mechanism. The newlinerole of the Smart Collector (SC) is to self-organize by itself to exhibit newlinethe capability to gather and aggregate data efficiently both during newlinecritical and non-critical occasions. Implementing Pro_ET shows newlinepromisin