Wideband Signal Detection Algorithm in Local Stationary Colored Gaussian Reverberation Model Background

For wideband signal detection in the local stationary colored Gaussian reverberation model background, the effect of auto regressive (AR) model parameters on local stationarity of reverberation is studied, and the simulation data of reverberation are obtained. Block matched filter and block pre-whiten matched filter for detection of linear frequency modulation(LFM) and hyperbolic frequency modulation(HFM) wideband signals in reverberation environments are presented. It is found that the colored Gaussian reverberation model can exactly describe the real reverberations generated by LFM and HFM via analyzing their local stationarity. The results of simulation and detection of wideband signal show that the two filters can detect the wideband signal at low signal to reverberation ratio(SRR), specifically the block pre-whiten matched filter can detect HFM signal at SRR ＞-19 dB, and the performance of block pre-whiten matched filter is much better than that of block matched filter.