A software package for the generation of noise with widely divergent spectral properties : The Simulation of Realistic Stationary Detector Noise

Abstract

The computer program discussed is capable of generating stationary noise with optional statistical parameters. Five standard noise types can be generated: white, 1/f, first order, Gaussian, and damped cosine noise. The input to the program can be a measured or arbitrarily chosen autocovariance function (or power spectrum). The autocovariance function (power spectrum) computed from the generated noise will be in perfect agreement with the input autocovariance function, if an infinite number of noise data is generated. Some theoretical work on noise generation on the basis of the autocovariance function is described. The autocovariance function of 1/f noise is theoretically derived from the power spectrum, based on a model described earlier. A few examples of the use of the program and an example of a possible application are given. A peak-finding procedure is tested with a simulated chromatogram contaminated with different types of noise. Applications are possible in data processing, information extraction, simulation and automation.