UC San Diego

Photonic systems have proven to be an integral element in the core areas of next generation communication infrastructures such as Wavelength division multiplexing, Discrete Time Optical Processing, Optical Time Division Multiplexing, and Optical Code Division Multiplexing. An optical setup can either replace, or be integrated into existing architectures to sidestep the inflexibility of conventional electronic circuitry used in telecommunications systems. In comparison to their electrical counterparts, photonic signal processors are able to outperform by providing superior sampling rates and crucial increase in bandwidth. Additionally, photonic devices benefit from very low transmission loss, and are impervious to the electromagnetic sources of interference that plague electronic devices. However, realistic considerations arise when implementing photonic signal processors. Both innate material properties and capabilities of current fabrication technology result in a degradation in the system performance of photonic devices. Such unavoidable behavior prevents photonic devices from reaching their maximum potential, and in turn hinders the widespread distribution of all-optical infrastructures. From a signal processing perspective, the effects can be readily expressed as a power loss coupled with the signal's propagation through the system. The need for a set of DSP design techniques that specifically consider the unique characteristics of photonic devices is therefore immediately evident. This dissertation explores the analysis and approach to a new class of allpass based filter design algorithms specifically targeted for photonic implementation. Allpass filter based systems are ideal for photonic realizations because the behavior is naturally observed in a variety of nanoscale dieletric components, which can be considered as the basic building blocks to complex systems. We examine the waveguide power loss effect at the most fundamental level, and present a set of design algorithms for phase compensators, bandpass filters, and filter banks based on realistic characterizations of photonic allpass elements. To increase the robustness of practical deployment of the designs, stochastic models that can aid the evaluation of the post fabrication filter performances are also demonstrated