Multi-channel signal detection using time-varying estimation techniques

We propose to improve the performance of a generalized coherence (GC) detector using noise-suppressed estimates obtained from time varying (TV) techniques. In source detection and localization, the presence of a common but unknown signal must be detected using data from several noisy channels. If one of the channel outputs has a sufficiently high SNR, then it could be pre-processed to improve detector performance. This work uses matching pursuit decomposition (MPD), and instantaneous frequency (IF) estimation to form an estimate of the highest SNR channel output. The signal estimate is then processed by a GC detector or a generalized likelihood ratio test (GLRT) detector to detect the presence of the signal on the remaining, noisier channels. Detector performance is shown to be significantly improved via simulations.