High-speed computation of autocorrelation using rectangular transforms

A technique based on the recently developed rectangular transforms is proposed for computing the autocorrelation function of a data sequence. The refinements suggested with the technique make it computationally more efficient than Rader's algorithm using the fast Fourier transform and the Winograd-Fourier transform algorithm, when the number of desired lag values is smaller than 30. However, the technique as such can be used for higher values.