适用于G.hn标准中电力线信道的实序列FFT处理器的设计与实现

发布时间：2018-06-26 01:43

本文选题：G.hn + 电力线通信　；参考：《华南理工大学》2013年硕士论文

【摘要】：G.hn是国际电信联盟(ITU)为正在高速发展的有线互联的家庭网络制定的标准,该标准同时支持电话线、同轴电缆和电力线三种传输介质,而电力线是其中最为复杂的介质。其物理层的传输速率理论上可达1Gbit/s,预期将成为新时期家庭网络有线互联技术的基础。G.hn标准采用OFDM调制技术,而快速傅里叶变换(FFT)是该技术的核心。因此在本论文中,我们创新设计了一个具有高处理速度、高吞吐率、高集成度、高灵活性、高精度和接口简单标准化等优点的,适用于G.hn标准中电力线信道的实序列FFT处理器。具体地说,本论文主要完成了以下三个方面的创新工作：第一,本论文研究了电力线信道的特性,并搭建了一个具有频率选择性、显著衰减并且随时间变化的电力线信道模型,用于FFT算法仿真与误差分析。第二,本论文研究了FFT的Radix-2/4/8算法,比较了各种算法的优缺点,选择了Radix-4作为FFT处理器的基本算法。利用实序列FFT算法的周期性和对称性,提出了一种创新算法,可以使运算量减少近一半,大大提高运算效率和处理速度。接着,为提高处理器的运算精度和减少硬件面积,采用Stage-based块浮点(BFP)算法用于IFFT/FFT数据的存储和运算。最后,使用Matlab对Radix-4BFP FFT算法进行仿真,对不同的定点方案做误差分析,最终找到了输入输出数据和旋转因子的最佳精度要求。第三,本论文研究了FFT处理器现有的架构,透彻分析了流水线式和存储器式两种架构的优缺点,设计了适用于G.hn标准OFDM系统的,基于乒乓RAM、流水线的实序列块浮点FFT处理器架构,结合了流水线式和存储器式两种架构的优点,使得本论文设计的FFT处理器具有高处理速度和高吞吐率的优点。接着,基于以上算法原理及架构.使用硬件描述语言Verilog,完成了FFT处理器的电路设计,并使用可重用的VMM架构进行了功能验证,提高了验证效率。最后采用TSMC0.18um1P6M工艺进行实现,面积约为6.3mm2,该处理器最高工作频率可达383MHz,即本论文设计的FFT处理器完全满足G.hn系统的要求。
[Abstract]:G. hn is a standard developed by the International Telecommunication Union (ITU) for a rapidly developing wired home network. The standard supports three transmission mediums: telephone line, coaxial cable and power line, among which power line is the most complex medium. The transmission rate of the physical layer can reach 1Gbit / s theoretically, which is expected to become the basis of the wired interconnection technology of the home network in the new era. The OFDM modulation technology is adopted in the G.hn standard, and the fast Fourier transform (FFT) is the core of the technology. Therefore, in this thesis, we have innovatively designed a real sequence FFT processor with high processing speed, high throughput, high integration, high flexibility, high precision and simple standardization of interface, which is suitable for the power line channel in G.hn standard. Specifically, this thesis mainly completes the following three aspects of innovation: first, this paper studies the characteristics of power line channel, and builds a frequency selective one. The power line channel model, which attenuates significantly and varies with time, is used for FFT algorithm simulation and error analysis. Secondly, this paper studies the Radix-2 / 4 / 8 algorithm of FFT, compares the advantages and disadvantages of various algorithms, and selects Radix-4 as the basic algorithm of FFT processor. Based on the periodicity and symmetry of the real sequence FFT algorithm, an innovative algorithm is proposed, which can reduce the computation cost by nearly half and greatly improve the operation efficiency and processing speed. Then, in order to improve the processing accuracy and reduce the hardware area, Stage-based block floating-point (BFP) algorithm is used to store and compute the IFFT / FFT data. Finally, the Radix-4BFP FFT algorithm is simulated with Matlab, and the error of different fixed-point schemes is analyzed. Finally, the best precision requirements of input and output data and rotation factor are found. Thirdly, this paper studies the existing architecture of FFT processor, thoroughly analyzes the advantages and disadvantages of pipeline and memory architecture, and designs an OFDM system suitable for G. hn standard. Based on ping-pong RAM and pipelined real sequence block floating-point FFT processor architecture, which combines the advantages of pipeline and memory architecture, the FFT processor designed in this paper has the advantages of high processing speed and high throughput. Then, based on the above algorithm principle and architecture. The circuit design of FFT processor is completed by using the hardware description language Verilog. and the function verification is carried out by using reusable VMM architecture, which improves the efficiency of verification. Finally, a TSMC 0.18um1P6M process is used, with an area of about 6.3mm ~ 2, and the maximum operating frequency of the processor can reach 383MHz. The FFT processor designed in this paper can fully meet the requirements of the G.hn system.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP332

【参考文献】