Improved method for detecting the short cycles of LDPC codes

It is well known that the short cycles in Tanner Graph (TG) greatly impact on the iterative decoding performance of low density parity check (LDPC) codes. In previous literature two methods can be found for detecting such short cycles. This paper presents an improved version of previous Jun Fan´s method, which is based on combinational analysis and enumeration of the check matrix patterns caused by cycles of length of 4, 6, 8 and 10 in the corresponding TG. The improved method can more exactly detect the cycles of length greater than 4 in comparison with Jun Fan´s method. The validation of the proposed method can be verified by the detection results of short cycles for some LPDC codes.