The analysis and performance improvement of optical fibre communication systems at high data rates.

Optical fibres have been increasingly used in communication due to their higher capacity, higher speed and higher reliability over other means of communication. Unfortunately, efforts to maximize their utilization have been hampered by a number of limitations, the most significant among them being dispersion in the fibres. A lot of efforts have therefore been expended in attempting to reduce the effects of dispersion. Among the methods used are optical phase conjugation, pulse shaping, and both electrical and optical equalization. In this research, dispersion-related distortions in pulses were first investigated, and followed later by attempts to reduce the distortions by pulse shaping. A mathematical model for dispersion-related pulse distortions in Gaussian-shaped pulses was developed. This was extended to take into account both the finite line width and the chirp of the laser. The results showed the expected pulse broadening associated with second-order dispersion and the separation of single pulses into multiple pulses associated with third-order dispersion. They also showed the presence of interference pulses produced when any two pulses interfered with one another. For more clarity on pulse distortion, the shapes of the interference pulses rather than their spectral content was analyzed, an approach which differs from work reported elsewhere. The amplitude of the interference pulses was found to be higher in dispersive fibres than in non-dispersive ones, and to decrease with the width of the source spectrum. It was also found to decrease as the distance between the interfering pulses increased. The interference pulses were found to be positioned midway between the interfering pulses when third-order dispersion was negligible, and to be positioned at the later occurring interfering pulses when the third-order dispersion increased. A mathematical model for dispersion-related pulse distortions was developed for square-shaped pulses. The distortions in the pulses were found to follow the same pattern as that observed for the Gaussian-shaped pulses with two exceptions. Firstly, there was less distortion outside the pulse widths of the square-shaped pulses because of their time-limited nature, and no interference pulses were found to be present when the fibre was non-dispersive. Secondly, the interference pulses were found to be positioned at the same positions as the interfering pulses when third-order dispersion was negligible, but they were found to be positioned at the same positions as the later occurring interfering pulses with increased third-order dispersion. The results for the square-shaped pulses are unique because no previous references were found on the analysis of such pulses. Attempts were made to investigate the pulse distortions caused by the focusing elements at the junctions of the communication lines. In this case, a new tool was developed by adapting tools developed for use in general optics to the analysis of pulse distortions in optical fibre communication. By employing this tool, the pulses were found to be distorted when the transmitting or receiving components at the junctions were spaced away from the focal point of the focusing elements, and they were also found to be distorted when the focusing elements were dispersive. The distortions were found to be greatly reduced when an optimized focusing system was used, showing that pulse distortion could be reduced by using well designed focusing elements, and by positioning the receiving components at the focal points of the focusing systems. Attempts were also made to find out if pulse distortions could be reduced by choosing appropriate shapes for the transmitted pulses. In the first attempt, data pulses were constructed from smaller pulses (mini-pulses) whose amplitudes were chosen so as to give the desired shapes to the data pulses. The pulses were then analyzed using the square-shaped pulse model. This attempt, for which the idea of using mini-pulses was new, did not produce improved results. The second attempt was based on a model developed from the laser rate equations, the Gaussian-shaped and the square-shaped pulse models. With this model, it was possible to use the laser rate equations to analyze the effects of second-order dispersion and third-order dispersion separately for the first time. Results from this attempt showed significant improvements when trapezoidal pulses with a pedestal on the leading edge were used. The developed models and tools have been used to propose new methods of combating dispersion-related pulse distortions, and they have also opened room for future work on such distortions.