Signal Processing using Spectrally Phase-Encoded Optical Frequency Combs for High Speed Computing and Pattern Recognition

Abstract

Researchers at the University of Central Florida have invented a system that introduces coherent optical signal processing via spectrally phase-encoded optical frequency combs for real-time optical filters and pattern recognition. As the amount of digital information and communication increases, timely and accurate data collection for software-based evaluation and correlation becomes increasingly more difficult. The increase in difficulty arises from three main issues: the sheer volume of the data being generated; the nature of the desired information relative to the comparison set; and the desired information that lies in the noise of the overall data set being evaluated. Thus, an integration of fundamental advancements in the generation and recognition of optical patterns to improve the speed of high-performance computing and data mining is needed. Existing high performance computer designs face the historical challenges of generating enormous amounts of heat, processing latencies measured in tens of nanoseconds and poor general processing flexibility. One lossless approach to overcoming these limitations is to perform the computing function with light. However, for the last three decades optical computing has been the perennial technology of tomorrow because of the persistent limitations imposed by the lack of programmability and the slow translation of electric data into optical regime.

Technical Details

The UCF invention is a robust, compact, scalable optical signal processing system for enhancing the speed of high-performance computing by more than two orders of magnitude. This is based upon the integration of fundamental advancements in the generation and relational recognition of optical patterns. The invention demonstrates a clear and achievable path towards making programmable optical computing a reality by overcoming historical barriers.

Benefit

Market Application