Performance analysis of a novel matched filter detector with reduced out-of-band response for narrowband signals

Reduced spectral response in adjacent spectral bins is very desirable for detectors since it provides the capability to detect narrowband signals in the presence of unknown interference and colored noise. Recently, a new method of detecting narrowband signals which uses a close approximation to the matched filter is proposed by Raja Kumar. The impulse response of this filter turns out to be longer than that of the matched filter. In order to support this extra length of filter response, it is proposed that the signal be extrapolated by prediction before the filtering operation. In this paper, the characteristics of this detector are mathematically analyzed and the performance of the detector in the presence of additive white Gaussian noise (AWGN) is studied. The detection method uses predicted signals that are obtained typically from short time data. Hence, in this paper, a method of analyzing the statistics of predicted signals from short time data is presented, and it is applied for the performance analysis of the present detector. The accuracy of the present detection analysis is demonstrated through theoretical and simulation studies. Also the analysis and performance study is extended to fluctuating targets, and it is shown that the proposed method yields near-optimum performance for Swerling´s Case-1 and Case-3 targets while giving the excellent advantage of low out-of-bin response