On computing multidimensional discrete Fourier transform

Summary form only given. An algorithm has been developed for computing an N-dimensional discrete Fourier transform of size P^R×P^R×. . .× p^Rr, where P is a prime number and R is a positive integer. The algorithm is based on grouping the input data on hyperplanes orthogonal to some chosen direction and then taking a one-dimensional discrete Fourier transform. An interesting feature of the algorithm is its inherent suitability for parallel/pipelined implementation in hardware