Analog optical processor architectures for sinusoidal signals

Analog optical processors are of interest because of their potentially large bandwidth and the ease with which one can implement addition. Processors for sinusoidal signals are a special case due to the advantages provided by representing them as a complex exponentials. The authors address some issues relevant to optical analog signal processors. It is shown that there are fundamental limitations on optical processors due to the quantum mechanical nature of light that links dynamic range, bandwidth and parallelism to the power required to implement the algorithm. Three well-known analog optical representations of signals are analyzed with respect to power requirements and cascadability, and it is concluded that a representation using four light beams provides the most straightforward implementation of complex arithmetic, requires the least power, and has the best S/N ratio for a cascade process.<
>