Comparison of jitter decomposition methods for BER analysis of high-speed serial links

We present and compare various jitter decomposition methods for bit error rate (BER) estimation, as required for the diagnosis and test of serial high-speed transceivers. The methods are based on the analysis of jitter distributions using the well known Gaussian Q-normalization principle. The comparison is realized with an efficient optimization scheme, which allows for a combined utilization and performance evaluation of the different methods. Error quantities and test distributions are also provided to support extensive performance simulations. Analyses include the influence of distribution sample size, test distribution shape and the timing resolution.