| 摘要: |
| 在不改变译码性能的条件下，为了加快最大后验概率（Maximum A Posteriori Probability,MAP）译码器状态信息更新的速度和降低算法的复杂度，提出了一种用于Turbo码的MAP译码器的免归一化处理算法。算法采用二进制补码加法器和减法器将MAP译码过程中的状态信息投影到一个归一化圆上，当状态信息更新时所有的状态信息在归一化圆上移动，通过保持归一化圆上状态信息的正确关系来计算似然比。归一化过程中不用搜索或估计状态信息的最大值，通过简化状态信息归一化过程加速了MAP译码器的状态信息更新并降低了复杂度。所提算法在与传统算法译码性能相同的情况下，可以降低36.2%的计算复杂度和17.4%的关键路径延迟，达到MAP译码器实现中的高速、低复杂度目标。 |
| 关键词: Turbo码 MAP译码器 免归一化 归一化圆 log_MAP算法 状态信息 |
| DOI： |
|
| 基金项目:国家级职业教育教师教学创新团队课题（YB2020080102）;江苏省职业教育教师教学创新团队支持项目（BZ150706） |
|
| A normalization-free information updating algorithm for MAP decoder |
| WANG Xuan,DU Jun |
| (1.School of Electronic Information Engineering,Nanjing Vocational College of Information Technology,Nanjing 210023,China;2.ZTE Photonics Technology Co.,Ltd.,Nanjing 210012,China) |
| Abstract: |
| To speed up the update of the decoders state metrics information and reduce the complexity of the maximum a posteriori probability(MAP) algorithm without changing the decoding performance，a normalizationfree information updating algorithm for the MAP decoder of Turbo codes is proposed.The algorithm applies the 2s complement adders and subtractors to project the status metrics data on a normalization circle.The state metrics will move along the normalization circle as updated,which can correct the relationship between the metrics and finally calculate the likelihood ratio.Therefore,the procedure of finding or estimating the maximum state metrics is not required,and the latency and complexity in the MAP decoder can be reduced through simplifying the normalization.Compared with the traditional algorithm with the same decoding performance,the proposed algorithm can reduce 36.2% area and 17.4% latency,thus achieving the goal of high speed and low complexity in the implementation of MAP decoder. |
| Key words: Turbo code MAP decoder normalization free the normalization circle log_MAP algorithm state metrics information |
