Online drift correction in wireless sensor networks using spatio-temporal modeling

Wireless sensor networks are deployed for the purpose of sensing and monitoring an area of interest. Sensors in the sensor network can suffer from both random and systematic bias problems. Even when the sensors are properly calibrated at the time of their deployment, they develop drift in their readings leading to erroneous inferences being made by the network. The drift in this context is defined as a slow, unidirectional, long-term change in the sensor measurements. In this paper we present a novel algorithm for detecting and correcting sensors drifts by utilising the spatio-temporal correlation between neigbouring sensors. Based on the assumption that neighbouring sensors have correlated measurements and that the instantiation of drift in a sensor is uncorrelated with other sensors, each sensor runs a support vector regression algorithm on its neigbours' corrected readings to obtain a predicted value for its measurements. It then uses this predicted data to self-assess its measurement and detect and correct its drift using a Kalman filter. The algorithm is run recursively and is totally decentralized. We demonstrate using real data obtained from the Intel Berkeley Laboratory that our algorithm successfully suppresses drifts developed in sensors and thereby prolongs the effective lifetime of the network.