Comments on "A further analysis of decorrelation performance of spatial smoothing technique for real multipath sources"

Yang and Tsai (see ibid., vol.40, no.8, p.2109-12, 1992) analyzed and discussed the decorrelation performance of spatial smoothing based on a first-order prediction approximation to time-delay narrowband signals when nonzero bandwidth signals impinge on the array. Based on the approximation model, they showed that the spatial smoothing progressively reduces the correlation between signals until certain smoothing is reached and then builds up the correlation to that of the original on further smoothing. The present authors first show that the prediction model used in Yang and Tsai is valid only for short time delays. In large arrays, the time delay (with respect to the reference sensor) becomes very large and the approximation fails. This failure leads to erroneous results, such as a decrease of the correlation initially and then an increase beyond so-called optimum smoothing. The present authors show that this result is due to modeling error and not due to nonzero bandwidth signals. Further, using the true signal model for nonzero bandwidth signals, it is shown that the spatial smoothing progressively decorrelates the signals.<>