Noise-improved Nonlinear Multiple Signal Detection

In communications and other fields, traditional reducing noise results in information lossinevitably and degenerates signal quality. However, noise can bring beneficial results to signalprocessing in some nonlinear systems in certain conditions ,i.e., the signal transmission or signalprocessing through certain nonlinear systems can be improved by adding noises, which is knownas stochastic resonance (SR).Focusing on ternary signal as the study object, this paper discusses four representative noisesand Gaussian mixture noise to improve multiple signal detection based on a maximum aposterior probability (MAP) criterion. This thesis is composed of five parts as follows:The first section shows the study background and progress of SR at home and abroad, andthen introduces several important kinds of stochastic resonance models and stochastic resonancetheories simply.The second section contains some basic knowledge about statistical signal detection theoryas follows: typical statistical model of binary signal detection, hypothesis testing theory,Bayesian criterion, derived bayesian criterions and multiple hypothesis testing.The third section discusses SR in nonlinear multiple signal detection for four representativenoises (uniform noise, Gaussian noise, Laplace noise, and Cauchy noise) based on a maximum aposterior probability (MAP) criterion.Under Gaussian mixture noise, the fourth section studies nonlinear statistic to improvemultiple signal detection and how the parameters affect the results of SR and supra-thresholdstochastic resonance (SSR).The fifth section gives the summary and scope of this paper.