Adaptive Sensing Using Data Prediction for Wireless Sensor Networks
Vipan Arora; Dr. T. P. Sharma
Wireless sensor networks are generally deployed in inaccessible terrains for monitoring certain physical parameters like temperature, humidity, radiations, vibrations etc. As large numbers of sensor nodes are sensing the region, it is likely that sensed data may be both spatially and temporally correlated. These correlations can be exploited to reduce communication cost by reducing data exchange by setting sampling rate of sensor nodes appropriately. In this paper, we propose an Adaptive Sensing Using Data Prediction (ASDP) scheme that controls the sensing rates of sensor nodes. To do this, scheme utilizes the local estimation at every cluster head and finds correlation among different values at cluster head level. The strategy thrives at dynamically altering the sensing frequency of sensor nodes based on this correlation. Highly correlated values depict the static nature of the event under observations whereas highly uncorrelated values points towards very dynamic event. Thus, by dynamically setting the sensing frequency, generation of redundant data in case of static event can be minimized. Also, missing some important readings due to low sensing frequency can also be taken care of. Simulation results show that ASDP provides substantial energy saving as compared to other adaptive sensing schemes.