In order to improve the decoding performance of the irregular Quasi Cyclic Low Density Parity Check（QC-LDPC） code and reduce the implementation difficulty of the decoder

this paper proposes a Based-Degree Normalized Min Sum（BD-NMS） algorithm and an improved decoding structure. The algorithm first analyzes the changing rule of the optimal normalization factor

and obtains its distribution characteristics in different degrees. According to the distribution characteristics

the degrees can be layered. Finally

the normalization factors of each layer are obtained by density evolution and weighted vector. In the proposed structure

the verification nodes are grouped with the lifting value as the unit

and each group is divided by the integral power of 2 as the unit molecular layer. The sub-layers are then updated in sequence. Simulation results and complexity analysis show that the proposed decoding structure consumes less resources and has lower complexity. It can be seen that the BD-NMS algorithm has better decoding performance with faster convergence speed. The decoding performance is 0.19 dB higher than that of Density Evolution Min-Sum（DE-MS） when the bit error rate is 10

-6

. It is also concluded that the increased complexity only includes comparison operation

which is easy to implement in hardware.