Improving Receiver Detection Performance Through NLOS/LOS Vision

This paper explores the benefits of sensor-based line-of-sight/non-line-of-sight (LOS/NLOS) information on the detection performance of an on-off keying (OOK) communication link. Bayes risk and composite likelihood ratio test (LRT) methods are used to derive the optimal decision rule for minimizing probability of error in a dynamic LOS/NLOS channel. By exploring three varying degrees of knowledge, it is shown that one can achieve improved constant-false-alarm-rate (CFAR) detection performance in an ensemble of trials over the uniformly-most-powerful (UMP) test when labeled information about LOS is provided at the receiver. It is also shown that, in the presence of prior knowledge of signal presence, one can benefit from LOS statistics when minimizing probability of error.