Position estimation of nodes moving in a wireless sensor network

Abstract

This thesis investigates the use of measured radio signal strength indicator (RSSI) in wireless sensor networks (WSNs) to provide an estimated distance. Estimated distances are arrived at using experimental calibration of a free space model to find the best free space model exponent and transmission power level over distances up to 7.78 m. We use these distances to estimate the position of a moving wireless sensor node. A new application called MobiPos for indoor distance determination using TelosB module has been developed and tested. In our MobiPos application, the moving node collects RSSI values from TestFtsp messages sent by up to eight stationary nodes. The moving node, in turn, collects valid RSSI observations from the same set, and formulates a special MULTI RSSI message that is sent to a stationary gateway node. The gateway node collects and time stamps received MULTI RSSI messages for post processing using a least squares position estimation process. We experimentally verified our research on a 23.92 m long mobile wireless sensor network testbed with up to six stationary nodes and one moving node. With an average velocity of 0.19 m/s, and 131 received MULTI RSSI messages over a test period of 127 seconds , we found that 64 observation sets (49%) converged with reasonable position estimates compared to approximate true positions. The average position difference for these 64 estimated positions compared to their approximate true positions was 5.03 m.