Fusion of weighted decisions in wireless sensor networks

Detection of an event occurrence is one of the main tasks of a wireless sensor network (WSN) in many applications. Sensors observe the environment and send either their raw measurements or their decisions about the event occurrence to a fusion centre (FC) where the final decision is made. It is desired to obtain a decision fusion scheme in which detection performance is optimised while considering resource limitations of WSNs. In this study, a weighted decision fusion (WDF) scheme is proposed. In WDF, each sensor that detects the event sends its quantised estimated signal-to-noise ratio to the FC if it is more than a pre-specified value. The FC uses the information in order to weigh each sensor´s decision and makes the final decision using an adaptive threshold scheme. More efficient use of bandwidth and adaptive thresholding at the FC result in considerable improvement. The substantial improvement of detection performance compared with existing methods is shown by analysis and simulation.