Adaptive Data-Centric Clustering with Sensor Networks for Energy Efficient IoT Applications

A wireless sensor network (WSN) typically involves deploying multiple nodes in an area to measure environmental parameters. WSNs are getting enveloped within the realm of IoT which significantly increases their scale of deployment. The end-objective of deploying a sensor network is to get valuable data about a region irrespective of the physical configuration used for measurement. We propose an Adaptive Data-centric Clustering algorithm for Sensor networks (ADCS), a hierarchical algorithm where user-specific data requirements are factored into the clustering decisions. Specifically, similarity in parameter variations are used as a criteria for optimization. We have deployed an eKo-based sensor network in north-eastern India to measure environmental parameters as part of a precision agriculture application. Data from this network is used to develop models to rigorously compare the performance of three variants of ADCS: ADCS-DB, ADCS-KM and ADCS-AG and arrive at useful recommendations for deployment planning.