Efficient DFT Architectures Based Upon Symmetries
Tomaso Erseghe, Gianfranco Cariolaro · IEEE Transactions on Signal Processing · 2006
This paper presents an efficient discrete Fourier transform (DFT) approach based upon an eigenvalue decomposition method. The work is based on some recent results on DFT eigenvectors, expressed exactly (not numerically) with simple exponential terms, with a considerable number of elements constrained to 0, and with a high degree of symmetry. The result provides a generalization of known fast Fourier transform (FFT) algorithms based upon a divide-and-conquer approach. Moreover, it can have interesting applications in the context of fractional Fourier transforms, where it provides an efficient implementation