Multi-dimensional Decomposition Algorithm and Hardware Implementation for Ultra-long Point FFT

Abstract

Fast Fourier Transform (FFT) plays a key role in digital signal processing. With the rapid development of signal processing technology, high performance ultra-long point FFT is required. This paper analyzes the characteristics of FFT algorithm, generalizes the two-dimensional decomposition algorithm to the multi-dimensional decomposition algorithm, and proposes a high-performance ultra-long point FFT hardware implementation architecture. This architecture implements the three-dimensional transpose operation by using the collision free bank addressing technology based on prime number banks. The synthesis results show that the operating frequency can reach more than 1GHz. The number of points supported and the performance in this paper are significantly improved compared with the existing research results.