| A narrow band, Gaussian-plus-shot noise process has been chosen as an impulsive noise model. The shot noise process has been generalized to control the randomness of arrival of impulses. Some first order statistics, including Amplitude Probability Distribution (APD) and Average Level Crossing Rate (ALCR), have been evaluated through analysis and simulation.Optimum non-linear receiver structures have been discussed, and optimum ZMNL for weak signal detection has been evaluated. Four simple ZMNL's such as the hard-limiter, the soft-limiter, the hole-puncher, and the ZMNL of algebraic type have been interduced to approximate the optimum ZMNL, and some attempts have been made to analyze the suboptimum nonlinear receiver. It has been shown, by simulation, that the simple ZMNL is quite effective in mitigating the effect of impulsive noise. The performance advantage is more apparent the more impulsive the noise is.The performance of fixed equalizers in impulsive noise has been discussed, and some simulation results have been provided. Also, the possibility of using ZMNL's in the receiver front-end has been investigated. It turns out that the performance of fixed equalizers, if the the ISI is not severe, is quite robust to the impulsive noise in the sense that no substantial performance degradation can be observed by combined effect of ISI and impulsive noise.Finally, a description of the Interactive Communication Simulator (ICS) upon which much of this work was done is provided. Some suggestions for future research have also been provided.The performance of a linear matched filter receiver operating in impulsive noise for various modulation schemes has been extensively studied. The modulation schemes considered include BPSK, QPSK, MSK, DPSK, and BFSK. The results indicate that the performance of a linear receiver can be severely degraded even with a small impulsive noise component. Also, the well known 3 dB advantage of DPSK over BFSK does not necessarily hold in an impulsive noise environment. A variety of numerical techniques for evaluating the probability of error performance of a linear receiver for coherent modulation has been surveyed, and the range of model parameters for which each numerical technique is applicable has been determined. |
