《计算机应用》唯一官方网站

• •    下一篇

R2^2FFT算法的FPGA硬件结构优化设计与实现

肖海林1,1,杨昱东1,杨紫伊2,3,刘海龙1,王玉1,张中山4,戴晓明5   

  1. 1. 湖北大学
    2. 湖北大学人工智能学院
    3. 湖北省武汉市武昌区湖北大学人工智能学院
    4. 北京理工大学
    5. 北京科技大学
  • 收稿日期:2024-07-17 修回日期:2024-10-21 发布日期:2024-11-19 出版日期:2024-11-19
  • 通讯作者: 杨昱东
  • 基金资助:
    国家自然科学基金资助项目;广西重点研发计划项目;湖北省高等学校优秀中青年科技创新团队计划项目

Design and implementation of FPGA hardware structure optimization for R2^2FFT algorithm

  • Received:2024-07-17 Revised:2024-10-21 Online:2024-11-19 Published:2024-11-19

摘要: 针对快速傅里叶变换(Fast Fourier Transformation, FFT)算法在处理大规模数据时需要消耗大量资源和时间导致运算速度低的问题,提出了一种基22FFT(Radix 22 Fast Fourier Transformation, R22FFT)算法的现场可编程门阵列(Field Programmable Gate Array, FPGA)硬件结构优化设计与实现方法。首先,采用R22FFT算法构建出一种序列转换功能与流水线结构相结合的Y型双并行阵列结构,有效降低硬件乘法器使用数量的同时,增大了硬件结构的吞吐量,从而提高了FFT算法在FPGA上的运算速度。然后,在R22FFT流水线单级运算中利用旋转因子的相关特性,优化片上存储资源消耗,降低了约50.00%的存储空间。最后,在完成N点R22FFT算法结构优化的基础上,进一步提高硬件结构的可扩展性,实现了 点、 点的扩展运算。采用Verilog HDL语言完成硬件设计,并通过Modelsim进行仿真,接着使用Vivado2018.3软件将所提方法综合并布局布线,最后进行性能分析。实验结果表明,所提方法的运算时间与四种改进的FFT硬件实现方法的运算时间相比分别降低了75.10%,95.34%,38.49%,49.20%,显著提高了运算速度。与此同时,所提方法资源消耗适中,消耗占比低,且具有运行功耗低以及可扩展性强的特点。

关键词: Y型双并行阵列结构, 现场可编程门阵列, 基2^2快速傅里叶变换, 多路径延迟交叉结构流水线, 蝶形运算

Abstract: A design and implementation method of Field Programmable Gate Array (FPGA) hardware structure optimization for Radix 22 Fast Fourier Transformation (R22FFT) algorithm is proposed to solve the problem that the Fast Fourier Transformation (FFT) algorithm requires a lot of resources and time to process large-scale data and thus leading to low operation speed. In the method, a Y-shaped dual parallel array structure for R22FFT algorithm is firstly constructed by combining a sequence conversion function and a pipeline structure, which reduces the number of hardware multipliers and increases the throughput of hardware structure effectively to improve the operation speed of FFT algorithm on FPGA. And then, the correlation characteristics of the twiddle factors are adopted in the single-stage operation of the R22FFT pipeline to optimize the on-chip storage resource consumption and reduced storage space by about 50.00%.. Finally, the scalability of hardware structure is further improved in realizing multi-point expansion such as points and points after the optimization of the N-point R22FFT algorithm structure has been completed. Furthermore, Verilog HDL language and Modelsim are utilized to implement hardware design and simulation, respectively. Then, the proposed method was synthesized and routed by using Vivado2018.3 software, and finally the performance analysis was carried out. The experimental results show that, compared with the four improved FFT hardware implementation methods, the operation time of the proposed method is reduced by 75.10%, 95.34%, 38.49% and 49.20%, respectively, which significantly improves the operation speed. At the same time, the resource consumption of the proposed method is reasonable and the consumption proportion is low. And it has the characteristics of low power consumption and strong scalability.

Key words: Y-shaped dual parallel array structure, Field Programmable Gate Array &#40, FPGA&#41, Radix 2^2 Fast Fourier Transformation &#40

中图分类号: