An adaptive EMS algorithm for nonbinary LDPC codes

The extended min-sum (EMS) algorithm for decoding low-density parity-check codes over the finite field with q elements significantly reduces decoding complexity by truncating each message of length q into a message of effective length n m . The number of effectively dominant components in each truncated message may gradually decrease with the number of decoding iterations. Based on this observation, we propose a novel adaptive EMS algorithm, called a two-length EMS (TL-EMS) algorithm. It chooses one of two candidate values as the effective message length n m for each message by reflecting the concept called message separation. Numerical results show that it can significantly reduce the computational complexity with little performance degradation.