Predictive Modeling for Energy Conservation in Wireless Sensor Networks (WSNs)

Energy conservation in wireless sensor nodes has posed a big problem and limits the applications and deployment of wireless sensor networks. This study proposes, an alogrithm to minimize energy consumption in the nodes so as to extend the life time of the sensors. It does this by generating a prediction model from the readings received from the sensor nodes in the sensed field for sometime. The readings are then used to generate a model at the base station and sent back to the sensor nodes. The software for the generation of the prediction model is developed using Visual Basic Version 6.0. The software so developed allows the monitoring entity to set various sleeping modes for the sensor nodes. The nodes compare their recent readings with the predicted model received from the base station and transmit their readings only when they are below or above a certain specified threshold. The experiment was set up using Rene mote fitted with temperature sensor. Five of such Rene motes were placed in different locations with the 6th as the base station. The set up was run for 1 h in the first instance in the update mode with the frequency of 1, 5 and 10 sec. It was then repeated for another 1 h using the prediction model. The result showed the research achieved a considerable energy reduction in the sensor nodes. The energy conservation obtained varied from 23-97% and depend on the frequency of sensing.