| Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 95 === The family of low density parity check codes belongs to linear block codes. Due to its sparse parity check matrix, it can be easily decoded by iterative decoding. It has been proved that when the code lengths approach infinite, it provides performance near Shannon limit. But the optimal decoding algorithm, which we call “Sum Product Algorithm”, needs to calculate a logarithmic function. However it’s difficult to implement the logarithmic function in hardware. The sub-optimal algorithm, called “Min-Sum algorithm”, omits the logarithmic function used in the Sum-Product Algorithm. It reduces some complexity in implementing the logarithmic function, but it also results in performance degradation. In recent years, much works are concentrate on improving the Min-Sum Algorithm such that its performance can be more close to the Sum-Product Algorithm. For example: Normalized Min-Sum algorithm (NMS), Offset Min-sum algorithm (OMS) and Adaptive Normalized Min-sum algorithm (ANMS). In this thesis, we propose a scheme called Modified Adaptive offset Min-sum algorithm (MAOMS). Without increasing too much complexity, the MAOMS algorithm outperforms the MS algorithm. Comparing with the above methods, the MAOMS
algorithm performs more closely to the Sum-product algorithm.
|
|---|
