Mathematical modelling of all-optical buffering for ultrafast optical time division multiplexed networks.

The development of a practical solution to all-optical buffer remains a challenge for high-speed (> 20 Gbit/s) optical networks. Most of the research in the field has concentrated on building test-bed solutions, however the literature review shows little evidence in the use of mathematical models to aid in the design process. This PhD study is an attempt to design and develop a mathematical model of an all-optical buffer suitable for use within optical time division multiplexed systems. The emphasis is placed on recirculating fibre loop buffers because of their inherent storage advantages. The most critical of these advantages is that the storage delay time is independent of the fibre length. While there is a precedent of employing large recirculating fibre loop architectures to simulate ultra-long haul transmission lines in research projects, their use in short length (< 500m) buffering architectures is not prevalent in the literature. This work finds a niche in this domain where the physical effects of the buffer components (e.g. optical switches) have not been previously documented. In order to optimise the bit error rate performance and characterise its dependence on the physical buffer characteristics, the buffer models are designed and simulated in MATLAB and VPI. The associated mathematical models, developed in this work, are validated by the results produced using these simulation packages. The benefit of this research is reflected in the fact that varying the parameters of the mathematical model effectively simulates the changing of physical device characteristics. Consequently, the designing process becomes less arduous, as lengthy simulation times are now reduced. Moreover, as physical implementation can now be delayed until the buffer design is optimised, production cost may be reduced.