Process capability index analysis of bit error rate of QPSK with limited observations

This paper studies the effect of a selection of probability density function to model a bit error rate of a QPSK modulation for process capability analysis purposes. Process capability indices are used widely used as quality metrics within the industry. The bit error rate may be modeled on linear and on logarithm scales and the shape of the probability distribution function changes significantly from one scale to another. An accurate modeling based on measurement observations would need infinite observations, but a limited amount of observations define a measurement horizon. The measurement horizon pools observations below the horizon to zero and thus changes the shape of the probability density function of bit error rate forming a bimodal distribution. If zero results are omitted, the best results are screened out and the quality analysis is done with measurement results which are not representative of the true population.