Frequency-domain based Reed Solomon decoders for GPU

Recent development and popularity of the Graphical Processing Unit (GPU) has attracted researchers to utilize it for error correction codes like Reed Solomon (RS). In this paper, we have proposed an efficient implementation of the RS decoder based on Frequency-Domain analysis. This decoder employs the Finite-field Fast Fourier Transform (FFFT) to convert the received code in Frequency-Domain; the error locator polynomial is derived using the inversion less Berleykamp-Massey algorithm (iBMA), and the error vector is generated in the frequency-domain using the coefficients of the key equation. Finally, using the Inverse Finite-field Fast Fourier Transform (IFFFT), the original message is retrieved in the time domain. This paper has shown that frequency-domain based RS coders have high efficiency and high throughput for correcting errors on flash memory devices, and they are more suitable for multimedia systems which has GPU.