Abstract: Hardware complexity of decoders，which is caused by storage and processing of massive information，is the major reason that encumbers LDPC codes from widely application.Reducing the quantization word length of decoding information can effectively decrease the hardware complexity.For the absolute value of information keeps increasing during decoding，short word length with finite quantization ranges will lead to serious saturation errors，which damages decoding performance.By gradually reducing the quantization precision，the quantization range can be expanded to meet the demand of changing of information.A dynamic quantization scheme is proposed to meet different noise environments and reduce the negative influence caused by quantization precision decreasing in a self-adaptive way.Results show that dynamic quantization scheme can greatly simplify the hardware complexity with very little lose of decoding performance.

Key words: Low Density Parity Check Codes（LDPC）, Turbo-Decode Message Passing（TDMP）, quantization, min-sum

摘要： 为存储、处理大量信息而导致译码器硬件复杂度较高是影响LDPC码广泛应用的主要原因。降低译码信息的量化位宽能有效降低译码器硬件复杂度。由于译码信息的绝对值在译码过程中不断增长，短量化位宽带来的有限量化范围会导致严重的饱和量化误差，影响译码性能。在量化位宽不变的情况下，通过逐渐降低信息量化的精度来实现信息量化范围的扩展。这种动态的量化方式能满足译码信息的绝对值不断增长的要求。为进一步适应不同噪声环境并减少量化精度降低带来的负面影响，这种动态量化方式以自适应的方式实现。仿真结果表明，自适应动态量化方式能以很小的译码性能损失，大幅减少译码器所需存储空间，简化硬件复杂度。

关键词: 低密度奇偶校验码, 层调度, 量化, 最小和算法