We propose a nonparametric detector for signal detection with distribu
ted sensors. A constant signal in additive noise is to be detected whe
re the only assumption made on the noise is that it has a symmetric pr
obability density function. The proposed scheme uses Wilcoxon statisti
cs followed by an M-level quantizer to generate the local decisions. E
quations for solving the optimum thresholds for the quantizers are pre
sented. The fusion center generates the final decision based on a line
ar function of the local decisions, Numerical results show that this d
etector provides better performance than other nonparametric distribut
ed detectors with the same transmission bandwidth requirement. (C) 199
7 Elsevier Science B.V.