Abstract LDPC(Low-Density-Parity-Check) code is a coding method with a performance closed to Shannon limit, which has e the main technology and hot researching in field of channel coding and especially in field of munications in recent years, and currently has been widely used in satellite digital video, munications, audio broadcasting and deep munications. It’s hard to avoid the trapping sets problem in constructing LDPC code, which is the reason of error-floor in high SNR(signal-to-noise ratio) region during decoding, that the BER(bit error rate) decreases very slowly with the increasing of signal-to-noise ratio, and it hinders the further increase of code performance. Now more and municating applications need lower BER to meet the higher requirements of the applications, so it e one of the problem with research value to improve decoding algorithm in order to reduce the error-floor. In order to reduce error-floor and aspire after lower BER, we analyse and improve the decoding algorithm of LDPC code based on t rapping sets to correct trapping sets error during decoding. By analysis of the trapping sets pattern, this paper research and improve a new bits flipping two-phase iterative decoding algorithm, which can correct the trapping sets error and further reduce the error-floor of the LDPC code and approach the code performance limit in an effective way. This paper designs the LDPC code bits flipping two-phase decoding algorithm simulation system by structure several quasi-cyclic LDPC(QC-LDPC) code of different code length and code rate using the quasi-diagonal cyclic of the check matrix and pick out four codes for using. By simulation of the bits flipping two-phase decoding, this paper analysis the performance and the improvement of bits flipping two-phase decoding in different kind of trapping set. The bits flipping two-phase decoding algorithm researched by this paper can achieve higher probability of correcting the trapping sets error and improve the performa