Maximum likelihood detection with intermittent observations

The decentralized detection performance, using wireless passive sensor networks, is analyzed according to the minimum probability of error criterion. Passive sensors communicate their measurements to the reader using data network packets, and therefore, the two main phenomena affecting the detection performance are packet loss and packet delay. In this paper, we formulate the decentralized detection problem with passive sensors and show that the optimal decision rule with packet loss is the likelihood ratio test. We present a comparative analysis study between detection with ideal and non-ideal channels, for the problem of DC level detection in white Gaussian noise. We validate the analytical results using Monte Carlo simulation study. Finally, we present a simple scheme for adaptive detector design, to restore the original detection performance, with the cost of increasing the delay for detection.