Abstract
vioft2nntf2t|tblJournal|Abstract_paper|0xf4ffc08f0a0000005ac2000001000900
In this paper we present alternate form of Radix-4/8 and split radix FFT’s based on DIF (decimation in frequency) version and discuss their implementation issues that further reduces the arithmetic complexity of power-of-two discrete Fourier Transform. This is achieved with circular shift operation on a subset of the output samples resulting from the decomposition in these FFT algorithms and a proposed dynamic scaling. These modifications not only provide saving in the calculation of twiddle factor, but also reduce the total flop count to ˜4Nlog2N almost 6% fewer than the standard Radix-4 FFT algorithm 2113log12NN? , 5% fewer than the standard Radix-8 FFT, and 273log9NN? , 5.5% fewer than the standard split radix FFT. Keywords: DFT (Discrete Fourier Transform), FFT (Fast Fourier Transform), Radix-4(R4), Radix-8(R8) and Split Radix (SR) FFT and Flop Count
Authors
Shaik Qadeer1, Mohammed Zafar Ali Khan2 and Syed Abdul Sattar3
1Muffakham Jah College of Engineering and Technology, India,2Indian Institute of Technology Hyderabad, India,3Royal Institute of Technology and Science, India
Keywords
DFT (Discrete Fourier Transform), FFT (Fast Fourier Transform),
Radix-4(R4), Radix-8(R8) and Split Radix (SR) FFT and Flop Count