Constructing confidence intervals for polarization mode dispersion

Polarization mode dispersion (PMD) causes significant impairment in high bit-rate optical telecommunications systems. Knowledge of a fibre’s PMD mean value, and the relevant confidence interval, is essential for determining a fibre’s maximum allowable bit-rate. Various methods of confidence interval construction for time series data were tested in this dissertation using simulation. These included the autocovariance-matrix methods as suggested by Box and Jenkins, as well as the more practical and simpler batch means methods. Some of these methods were shown to be significantly better than the standard method of calculating confidence intervals for non time series data. The best of the tested methods were used on actual PMD data. The effect of using polarization scramblers was also tested