Modified Welch's PSD based on Parallel FFT

Sreelekha S, Sabi S

Abstract: Spectral analysis is the most widely used methods for data analysis in radar, communication, geophysics, oceanography, atmospheric science, astronomy, hydrology and biology. This paper presents a low-complexity modified algorithm and architectural view to compute the power spectral density (PSD) using the Modified Welch Method. Welchs method is asymptotically unbiased estimate of the power spectrum by averaging the subsample periodograms. Welch algorithm also provides an excellent estimate of spectral power. But high computational complexity is the main drawback of Welch method. To reduce the high computational complexity use 50 % overlap by computing N/2-point FFT, where N is the length of the window and is merged with previous N/2-point FFT and then calculate the N-point FFT. Frequency-domain windowing operation is preferred for high performance. In order to improve the computational speed parallel FFT approach is used.

Keywords: power spectral density, low-power, low-complexity, Welch method, windowing, Fast Fourier Transform

How to Cite?: Sreelekha S, Sabi S, "Modified Welch's PSD based on Parallel FFT", Volume 5 Issue 7, July 2016, International Journal of Science and Research (IJSR), Pages: 345-349, https://www.ijsr.net/getabstract.php?paperid=ART2016130, DOI: https://dx.doi.org/10.21275/ART2016130