Research And Realization Of Digital Signal Processing Algorithm Based On FPGA

Modern signal processing is required to have high performance of real time processing. When the fast Digital Signal Processor(DSP) is still unable to meet the speed requirements, the only choice is to increase the number of processors or adopt customer's customized gate array products. With the development of Programmable Logic Device technology, Field Programmable Gate Array (FPGA) with massively parallel processing capability shows superiority in cost, performance, size etc. In this paper, signal processing algorithm, such as fast Fourier transform, digital filtering, and correlation algorithm are researched and efficiently realized.Firstly, according to the real-time requirement of imaging sonar and the resource constraints of the FPGA, FFT processor with cascade and parallel recursive structure is designed. The method to increase calculation speed with pipelining and parallel processing technology is discussed in detail. Subsequently, some optimization and improvement of butterfly arithmetic unit is proposed.Secondly, the characteristics of the digital filtering are analized, so is the correlation calculation with the same structure. The structure with multiplier and the structure with non-multiplier are adopted in multiply-accumulate calculation (MAC). Distribute algorithm (DA) is adopted in the non-multipler structure, which can greatly improve the computing efficiency by converting multiplication to look up table in PFGA. In addition, the correlation algorithm in frequency domain is researched in the paper.Finally, the pretreatment system of image sonar is completed, and the receiving, filter, orthogonal transformation and transmit of 160 channels signal are realized on one piece of EP2S60. According to the testing results, the algorithm runs well in the experiment and it fulfills the predefined requirement.