Spectral estimation using fast Hartley transform

Abstract

In this study the Fast Hartley Transform (FHT) is utilized for power spectral density estimation implementing the correlation technique. The main objective is providing a faster and less expensive approach compared with the Fast Fourier Transform (FFT) for design and implementation of the digital signal processors. In conventional procedures, and for many applications such as Nonparametric Spectral Estimation (NSE), different types of FFT algorithms are utilized. Implementation of FFT for digital signal processors design requires complex computations. Therefore, if the function of interest is real, the FFT algorithm is unnecessary and time consuming. Alternatively, FHT can accomplish the same task of power estimation in a shorter period of time, since algorithm developed for FHT requires only real multiplications and additions. In addition, the FHT will not compromise the information about phase components of the transformed signal, since the phase components can be obtained using the developed FHT procedures. In the present study, algorithms based on FHT concepts have been developed for digital signal processor implementation and for real time applications.