Research On Signal Detection Using IIR Notch Filter In Colored Noise

Signal detection and estimation in noise is the key problem of signal procession. It has wide applications in radar, sonar, oscillation measurement, biomedicine, seismic signal processing. Traditional methods of signal detection and estimation were based on the assumption that the background noise is white. But in practice, this assumption is not always true. When the noise is not white, the traditional signal detection methods will be deteriorative and even invalid. So, it causes the development of signal detection in colored noise. The dissertation was devoted to satisfy this demand.In this dissertation, high order statistic theory and adaptive technique is used to detect and enhance line spectrum. The main results and achievement are summarized as follows:A direct form IIR notch filter based on the FOMC(forth order mixed cumulation) of the observation sample is introduced. Compared with traditional autocorrelation method, the IIR notch filter based on FOMC can not only suppress the effect of Gaussian noise, but also suppress the other non-Gaussian noise, like Raleigh noise, uniformly distributed noise, etc. Computer simulation result proved the correctness.A new method which can detect and estimate the signal with fewer computation is proposed, and it can also track the signal frequency.A complex algorithm of gradient-based second-order adaptive IIR notch filter is proposed. The convergence characteristics about gradient algorithm and simplified algorithm were investigated. It has been shown that the convergence speed does not depend on the variance of input white noise; and the simplified algorithm converges faster than the gradient algorithm. The effects of colored input noise generated by first-order autoregressive model to the convergence speed were presented. The theoretical analyses show that the initialization of the adaptive process is related with convergence speed in colored noise.A new cascaded structure is proposed. It can exactly estimate the frequency of the input noisy signal, either the background noise is white or nonwhite.An adaptive algorithm based on the power of forth order cumulant of output signal is proposed. Compared with original simplified gradient adaptive algorithm, the...