SBIR-STTR Award

Fast Fourier transformers (FFFs) are used extensively in military and commercial applications. There is a constant demand for faster and more accurate FFTS. RNS (Residue Number System) based FFTs are candidates to provide these improvements, however techniques for efficient implementation of RNS operations must first be developed. RNS operations will be develope in the proposed Phase I research that will lead to RNS-based hardware designs with significan improvements in both speed and accuracy. The Phase I research will provide the conceptual are preliminary designs of the RNS-based FFT. The most promising designs will be selected an compared in terms of estimated performance versus cost. This will be accomplished by developing a computer simulation to evaluate the candidates for selected moduli. In Phase II the preliminary design will be refined and a hardware processor fabricated. The RNS-based FFT will be evaluated in terms of speed and accuracy and compared, based on performance and cost, to current FFT processors used in military and commercial applications. The technology developed in Phase I and implemented in Phase II will lead to enhanced performance of sensors in future weapons systems. Further , commercial applications that require intensive data analysis a processing will benefit from this research. Anticipated

Benefits:
FFTs have a variety of uses in both military and commercial applications which can benefit from RNS-based FFT processors. The most common military application is radar signal processing. Commercial applications which require intensive analysis of test data in the structural, fluid flow, and thermal areas such as aircraft and automobile engine testing will benefit from this technology. Weather forecasting and seismic modeling of earthquakes which record and process large data bases can effectively use this technology

Keywords:
FOURIER TRANSFORMS RESIDUE NUMBER SYSTEMS RADAR SIGNAL PROCESSING

Fast Fourier transforms (FFTs) are used extensively in military and commercial applications. There is a constant demand for faster and more accurate FFTs. RNS (Residue Number System) -based FFTs are candidates to provide these improvements. The Phase I research was performed to identify and develop an FFT algorithm that utilizes RNS-based techniques and to insure a successful hardware design and development. A prime length FFT algorithm which is computed efficiently and accurately was indentified in Phase I. This FFT algorithm utilizes RNS-based number theoretic transforms (NTTs) and is referred to as FFT/NTT. Parameterization of the NTTs provided design solutions for a variety of data lengths. The design of a RNS multiplier for the NTT computation insures high speed and low-cost performance. Preliminary design of the system architecture was performed in Phase I. Implementation of the FFT/NNT will yield improvements in speed and accuracy over current FFT processors. In Phase II the preliminary design will be refined and a hardware processor fabricated. The FFT/NNT processor will be evaluated in terms of speed and accuracy and compared, based on performance and cost, to current FFT processors used in military and commercial applications. The technolgy developed in Phase I and implemented in Phase II will lead to enhanced performance of sensors in future weapon systems. Further, commercial applications that require intensive data analysis and processing will benefit from this research. FFTs have a variety of uses in both military and commercial applications which can benefit from RNS-based FFT processors. The most common military application is radar signal processing. Image processing which requires two-dimensional processing is another important application. Commercial applications which are computation intensive include environmental monitoring, oil and mineral exploration, robotics, manufacturing, medical imaging and telecommunications.

Keywords:
FOURIER TRANSFORMS; RESIDUE NUMBER SYSTEMS; RADAR SIGNAL PROCESSING