Performance Analysis of Regular Low-Density Generator-Matrix Codes Under Majority-Rule Based Iterative Decoding Algorithm

Low-density generator-matrix (LDGM) codes are of particular interest not only because of their satisfying performance while maintaining low encoding complexity, but also because of promising applications such as network coding in cooperative multiple-access networks and joint source-channel coding where the LDGM codes are suitable. We investigate the performance of regular LDGM codes under the majority-rule based (MB) iterative decoding algorithm. The threshold and error-floor behaviors of the code can be extracted. A lower-bound expression for the error floors is obtained. The optimal majority weight for the MB decoding algorithm to achieve the lowest error floor is found.