Statistical techniques for objective characterization of microwave device statistical data

A comprehensive treatment of statistical metrics for the characterization of microwave device statistical data is presented. The primary aim is to investigate the power of these tests in their ability to faithfully distinguish between like and unlike joint probability density functions. It is shown that the adequate techniques are available to solve this problem, and a novel application of these techniques is illustrated by distinguishing the statistical difference between two GaAs FET databases with identical means standard deviations, kurtosis, skewness, and correlations. The authors verify the proposed characterization approach by design centering a small-signal amplifier, both with and without the use of statistically characterized device data.<>