Noise models for detection
Abstract
Binary detection problems involve the section of one of two statistical environments, described as an hypothesis H and an alternative K. The decision is based on a statistic of an observation sequence x of length m. The present investigation is concerned with a deterministic signal in additive noise. Several methods of generating families of multivariate noise densities are considered. It is pointed out that methods which produce large families of densities are of particular interest since they are more variable and may provide a closer fit to the actual noise. Closed forms are considered along with differential equations, a method of generating elliptically symmetric densities with a specific marginal, and transformation noise.
 Publication:

ICC 1982  The Digital Revolution, Volume 1
 Pub Date:
 1982
 Bibcode:
 1982icc.....1S...1S
 Keywords:

 Binary Data;
 Data Transmission;
 Multivariate Statistical Analysis;
 Random Noise;
 Signal Detection;
 Transmission Efficiency;
 Communication Theory;
 Mathematical Models;
 Optimization;
 Probability Density Functions;
 Stochastic Processes;
 Transformations (Mathematics);
 Communications and Radar