A Decentralized Positioning Method for Wireless Sensor Networks Based on Weighted Interpolation

A decentralized sensor positioning algorithm is proposed using an adaptive weighted-interpolation method. The proposed method utilizes in-network processing among sensors to compute the location of the target, which is in contrast to most existing algorithms that rely on the joint processing of raw measurements from all sensors at a central server. Specifically, the target location is computed by taking the weighted average of the local estimates based on the sensors' reliability. The average is attained iteratively with each iteration being performed by a different sensor in the network. During each iteration, a sensor computes a new estimate of the target's location based on its own observation and the most recent update passed over by the sensor responsible for the previous iteration. The newest location estimate and the update process is circulated among the sensors in the close-vicinity of the target, similar to that of a token- ring topology. A message-passing protocol is proposed for the inter-sensor communication and is used to adaptively select the participating sensors as the target moves around the area. Energy and bandwidth efficiency is achieved since the system need not expend large amounts of resources in transmitting the raw data to the central server. Simulation results demonstrate the fast convergence of the iterative method and the effectiveness of the proposed positioning scheme compared to other methods.