Energy Efficient Machine Learning Technique for Smart Data Collection in Wireless Sensor Networks

Abstract

Performance-centric automated system plays a vital role in next-generation wireless networks. With reduction in size and cost, sensor devices have led to envision a world of ubiquitous wireless sensor networks. Inherent behavior of resource-constrained sensors results in energy-consuming hot spots (such as communication overhead, severe packet collision, network congestion and packet loss) causing premature death of sensor nodes and entire network. In this paper, a novel ‘Monkey Tree Search-based Location-Aware Smart Collector (MTS_LASC)’ that exploits fauna inspired Monkey Tree Search (MTS) behavioral model is explored. The MTS_LASC is an extremely dynamic and fascinating phenomenon comprising distributed smart collectors and a centralized meta-heuristic MTS engine used for solving hard and complex problem. The distributed smart collector is embedded with a client MTS module. It is capable of analyzing, categorizing and aggregating data collected from sensors and disseminating them to the sink using fuzzy inference mechanism, whereas the centralized MTS engine exploits meta-heuristic search to facilitate comprehensive situation awareness through energy-efficient route among multiple paths for crucial decision making in Internet of Things-based applications. Simulation results reveal promising gains with higher delivery ratio by significantly reducing redundant packet transmission and maintaining fidelity through data aggregation. Performance analysis shows that MTS_LASC remains stable even in high traffic-constrained setup as energy degrades more slowly resulting in prolonged network lifetime. By improving the life prospects of the sensor network commendably, the proposed scheme reflects high potential on practical implementation.