Synthesis of multivariate Gaussian random processes with a preassigned covariance (Corresp.)

Evaluation of complex systems in a laboratory environment requires the generation of inputs to the system sensors that are representative of the operational environment. It is therefore necessary to synthesize input test signals that reflect the mutual dependencies found in situ. For multivariate Gaussian inputs, algorithms are derived allowing 1) the transformation of dependent Gaussian random variables into independent variables; 2) the generation of jointly Gaussian random variables with a constant covariance matrix; and 3) the synthesis of stationary multivariate Gaussian random processes. These algorithms have simple electronic hardware and computer software implementations that will facilitate the laboratory evaluation and digital computer simulation of complex systems.