A Study of the Effect of Sample Dependence and Sample Size on Some Nonparametric Signal Detectors.

摘要

Most nonparametric techniques of hypothesis testing and estimation used for signal detection require the availability of a large number of independent samples. However in non-Gaussian noise environments independence can be obtained only by sampling at a rate much less than the Nyquist rate. Hence in many practical communication situations,where nonparametric procedures should have advantage over parametric procedures,it is not possible to satisfy the requirement of independent samples without seriously reducing the data rate. Effect of sample dependence and sample size on the performance (in terms of probability of error) or M-ary optimum Gaussian,Mann-Whitney and Sign detectors is evaluated under Gaussian and Cauchy (jamming) noise environments. The evaluation is performed by a hybrid simulation of a communication system with facility to vary the sampling rate. An increasing energy ternary (M=3) signal set is used. (Author)