摘要作为进行高速实时处理的专用处理器，数字信号处理器已经成为数字通信、雷达、声纳、电力监控等领域的基础器件，而快速傅里叶算法是数字信号处理中必不可少的工具，所以研究其算法的优化，减少其对内存资源的占用，以期降低运行功耗，则具有重要意义。80589
本论文基于TI公司的多核处理器TMS320C6678，利用CCS开发环境来研究实现一种改进的FFT算法来减少旋转因子的查找次数，消除冗余的内存引用，减少资源占用，并且调用MATLAB中FFT函数，得到结果与CCS环境中所得的结果进行比较，验证算法的正确性，以及对不同点数的FFT运用两种算法的性能比较。然后为了进一步提高DSP的运算效率，研究实现FFT在多核DSP上的并行实现方法，分析不同点数的FFT在不同核个数上进行的并行运算。
毕业论文关键词 DSP 多核并行处理 FFT算法 TMS320C6678
毕业设计说明书外文摘要
Title Realization of FFT algorithm based on multi-core DSP
Abstract As a special processor for high-speed and real-time processing, Digital Signal Processor has become the basis of the device in digital communication, radar, sonar, power monitoring and other fields。 At the same time, Fast Fourier Transform algorithm is a necessary tool for digital signal processing。 So the research on the optimization of the algorithm to reduce the occupation of memory resource, in order to reduce power consumption, has important significance。
This paper is based on TI’s multi-core processor TMS320C6678, using CCS development environment to realize an improved FFT algorithm to minimize memory references due to twiddle factors, reduce the resource occupation。 I have called FFT function in MATLAB to operate, and compare the results of MATLAB with the results obtained in the CCS environment, so the comparison can verify the correctness of the algorithm。 I have also applied the two different algorithms to implement FFT of different points。 Then in order to further improve the operational efficiency of DSP, I have also investigated the realization of the FFT in parallel implementation method based on multi-core DSP , and then analyzed the experimental results of different points FFT using different nuclear numbers of parallel computing。
Keywords DSP Multi-core parallel processing FFT algorithm TMS320C6678
目 次
1 引言 1
1。1 数字信号处理器发展 1
1。2 FFT算法及应用 2
1。3 论文主要内容安排 3
2 快速傅里叶算法原理 5
2。1 离散傅里叶变换 5
2。2 基-2算法 5
2。3 库利图基混合基算法 8
3 基-2FFT算法的优化 10
3。1 蝶形图分组法 10
3。2 简化旋转因子法 12
4 FFT算法在DSP上的实现 16
4。1 硬件平台TMS320C6678 16
4。2 软件平台CCS 18
4。3 基-2FFT优化算法的实现 19
4。3 优化算法实验结果分析 TMS320C6678的FFT算法的多核DSP实现研究:http://www.youerw.com/tongxin/lunwen_93706.html