An Efficient DFT Implementation Using Modified Group Distributed Arithmetic

Abstract

Many of the modern signal/image processing applications use Discrete Fourier Transform (DFT) as one of the core functional elements to process the input signal/image from one domain to another. Hardware design of the DFT is complex and many researchers have proposed variety of methods to implement it. Computational complexity of 1D, N-point DFT is $O(N^{2})$. Distributed arithmetic is one of the promising and efficient technique to implement any discrete orthogonal transform. This paper proposes an efficient approach to implement DFT using distributed arithmetic. The proposed technique exploits the repetitive pattern of coefficients and stores efficiently in the memory and reduces the storage by 75% compared to the existing group distributed arithmetic for 8-point DFT. The proposed approach uses the property of group distributed arithmetic efficiently in the architecture.