Integrated Benes optical switches: an automated bottom-up design implementation.

The trend in increase in bandwidth consumption, as well as low latency applications, is pushing the network technology in the direction of a completely transparent optical transport layer. Optical switching elements and topologies are fundamental components to enable routing in a flexible, or even software-defined network. Due to their transparency they allow optical routing without requiring a more expensive and power consuming optical-to-electrical conversion of data packets. The advantages are numerous due to their small footprint, low energy-consumption and latency, as well as silicon photonic compatibility, making this class of components cardinal in the implementation of general purpose photonic transmission networks and data centers applications. Their implementation as photonic integrated circuits (PICs), compatible with Silicon Photonics processes, is crucial in making these components affordable and practical for implementation in common structures. The availability of applications for the numerical simulation and the design of PICs has allowed me to analyze, through a bottom-up approach, a class of these switches, namely the Benes multistage crossover switch, following a generalized method, from the physical design of the internal components, up to the production mask, with element-wise and system level performance simulations. The use of the photonic simulation and design suite (Synopsis, Inc.), in conjunction with ad-hoc external scripts I wrote, allowed the generation of a general tool for the design of Benes optical switches. This additional tool offers the possibility of developing and designing custom made optical switches under a single unified workflow, allowing the user to customize the internal components of the PIC, as well as the transmission parameters, such as the number of channels, bandwidth, and central frequency. To assist the user in the evaluation of the performances, I implemented a generalized deterministic algorithm to evaluate routing optimization for any arbitrary sized Benes networks. This algorithm operates on a mathematical abstraction of the switch, and could be applied as a control driver for the switching element. The generalized approach and workflow was finally used to compare and test different implementations of the basic switching elements inside the network, showing the effect on the transmission performances, such as bit error rate, inter-channel crosstalk, as well as power efficiency. In conjunction to the practical results on the performances of this class of switching networks, the generalized bottom-up approach is a testament to the strength of the interaction between PICs simulation suites and external scripts compatibility, allowing for in-depth analysis of different sets of photonic devices. The developed tools offer a reliable and coherent platform to compare different Benes optical switches, but acts also as a case-study of a unified design process, from material and technological specifications, up to system levels parameters.