Research And Implementation Of Fast Fourier Transform By Using Custom Instruction Based NiosⅡ Processor

Digital signal processing comprises some different branches of science, such asmathematics, physics, electronics and computer science. It has played a veryimportant role in promoting the development of science and technology. DSPs areused in a number of applications from communication and controls to speech andimage processing. The Fast Fourier Transform (FFT) can convert time or spatialinformation into the frequency domain, which is one of the most widely usedalgorithm in the field of digital signal processing. There are three commonly useddigital signal processors: general purpose digital signal processing chip (DSP),application specific integrated circuit (ASIC) and field programmable gate array chip(FPGA), which have their own characteristics and advantages and dominate indifferent areas. Nios II processor system which is a FPGA-based soft-core processorcan provide an efficient and flexible development platform for the realization ofcomplex algorithms. In addition, the unique custom instruction configuration of NiosII processor also can implement part of operations in FPGA hardware, which willcontribute to reduce the processor’s pressure to calculate the entire algorithm bysoftware and then improve the efficiency and speed of operation. This is the jointdevelopment of hardware and software.The goal of this paper is to implement the radix-2decimation in frequency (DIF)FFT by custom instruction in the Nios II processor system. First of all, we built thehardware platform of Nios II processor on Cyclone III FPGA chip, and thenachieved the butterfly operation of radix-2DIF FFT with the hardware languagevhdl. Secondly, we connected the custom instruction logic area of butterfly operationwith the Nios II processor system. Finally, we could come true the FFT operation bycalling custom instruction with the application code. The communication between processor and computer is via a serial communication port whose control interfacewill be completed in the LabVIEW environment.At the end of this paper, we have analyzed the real-time characteristic andaccuracy of the entire system. The results prove that this design is effective.