Scalable optical switches: Architectures and physical layer characterization

Optical switches are the key components in realizing reconfigurable all optical networks. The deployment of optical switches (replacing the conventional electronic switches) at the network nodes enables increased switching capacity and flexibility to data rate, format, and protocol. The scaling of optical networks, while benefiting from optical transparency is however influenced by the accumulations of certain physical impairments, such as interference (crosstalk) and amplifier noises that cannot be entirely eliminated or compensated without the need for signal regeneration. The objective of this work is to investigate the scalability of various optical switches and networks by using theoretical and experimental models. Integrated directional coupler switches are studied by developing an improved numerical model taking into account of all possible impairments for more accurate representation. New architectural options for large port count wavelength add/drop switches that are modular and cost effective are explored and accessed by making use of the experimental characterization results on the switch modules. Further, the performance aspects of cascading of switches in periodic networks are investigated experimentally by using a re-circulating test bed in the presence of combined physical impairments, and their impact on the tolerance to the switch crosstalk is revealed. Architectural options for the next generation of multi-granularity switching node, consisting of wavelength switches and semiconductor optical amplifier gates, are also explored and their physical layer performance in single and cascaded stages are experimentally characterized for optimum operating conditions. The models, experimental and numerical, presented in this thesis, can be used to understand and predict the limitations due to various impairments in optical switches and networks and to design those systems by comparing different switching configurations and system parameters for an optimized performance, for any given set of requirements