High speed area efficient modified fast fourier transform algorithm for bioinformatics applications

Abstract

The Fast Fourier Transform (FFT) plays a vital role in signal newlineprocessing applications. The signal processing tools using genomic data are newlineknown as genomic signal processing. The input data size as in human newlinegenomics is increased, for bioinformatics applications like the detection of newlineDiabetic Retinopathy (DR), the performance of the FFT needs to be newlineimproved. Thus there is a research gap to improve the speed of the FFT in the newlinescope of genomic signal processing for variable bit length implementation newlinewith the reduced area for Bioinformatics applications. newlineIn conventional 8-point FFT, the first output sample could be newlinecomputed only by receiving the 8th input data. So the computation of the first newlineiteration requires the last sample. Thus a delay is generated in the newlinecomputation. In Phase 1 of this research, a solution is provided, to avoid this, newlineby re-ordering the computation blocks in FFT. The number of multipliers and newlinethe adders for the computation of N point FFT are reduced to (N/2) and (N) newlinefor the proposed algorithm on receiving the last (N-1) th sample. The input newlinenucleotide samples of 1075 in numbers with different length are selected to newlineexploit the reduction in computation complexity of the proposed algorithm. newlineAs the nucleotides are expressed in characters, it has to be converted into newlinenumerical equivalent to identify the coding region. A weighted scheme is newlineproposed to numerically map the genome sequence for genomic signal newlineprocessing which has 92.32% of accuracy in identifying the codon region in newlinecontrast to all other methods in the existing literature. Out of 1075 samples newlinewhich are numerically encoded, 1053 samples have shown an improvement in newlinecomputation speed of the modified FFT scheme newline