多码率码型QC-LDPC编码器的FPGA实现

FPGA Implementation of Multi-Bitrate Pattern QC-LDPC Encoders

摘要: LDPC码作为一种线性分组码，具备优异的纠错性能，被广泛应用在无线通信中的信道编码领域。采用QC-LDPC生成矩阵的编码方法，设计了一个支持多码率准循环低密度校验码的编码器，编码器的硬件结构由生成校验码单元(gccu)、源码暂存单元(ocsu)和控制单元(ctrl)等三个主要部分构成，ocsu通过对存储单元进行切割，采取乒乓操作，在数据对乒路原始数据进行编码的同时，将另一帧数据写入乓路，提高了编码速率；gccu实现编码的运算过程以及对生成矩阵进行预存储，通过分析各种码型生成矩阵的特点，对矩阵进行划分，提取必要生成向量进行缓存，减少资源占用；为了满足多种码型、码率的编码需求，加入配置信号，可以根据不同矩阵列块数量、块大小的不同分别选择相应数量的运算单元以及运算次数。此套编码器共支持四种码率、六种码型的编码模式，并支持后续的快速拓展。在XILINX的XC7K325tffg900 FPGA上，采用verilog语言完成了各种码率的仿真综合，同时搭建MATLAB软件模型进行验证，证明了本方案的可行性与有效性。

Abstract: As a linear packet code, LDPC code has excellent error correction performance and is widely used in the field of channel coding in wireless communications. Using the QC-LDPC generation matrix coding method, designed an encoder that supports multi-bitrate quasi-cyclic low-density check digitization, the hardware structure of the encoder is composed of three main parts such as generating the checksum unit (gccu), the source code staging unit (ocsu) and the control unit (ctrl). ocsu cuts the storage unit and performs ping-pong operation. When the original data is encoded on the ping circuit, another frame of data is written onto the ping circuit to improve the coding rate. gccu implements the operation process of coding and pre-storage of the generation matrix. By analyzing the characteristics of the generation matrix of various code types, the matrix is divided and the necessary generation vector is extracted for caching to reduce the occupation of resources. In order to meet the coding requirements of various code types and code rates, the corresponding number of operation units and operation times can be selected according to the different numbers and sizes of blocks in different matrix columns by adding configuration signals. This set of encoders supports a total of four-bit rates, six code patterns of coding modes, and supports subsequent rapid expansion. On XILINX’s XC7K325tffg900 FPGA, a simulation synthesis of various bitrates was completed using the verilog language. At the same time, the MATLAB software model was built for verification, which proved the feasibility and effectiveness of the scheme.