Improved min-sum decoding algorithm for moderate length low density parity check codes

Abstract

In this chapter, a new technique to improve the min-sum decoding algorithm for the low density parity check (LDPC) code has been proposed. This technique is based on the magnitude overestimation correction of the variable message by using two normalized factors in all iterations. The variable message is modified with a normalized factor when there is a sign change and with another normalized factor when there is no sign change during any two consecutive iterations. In this way, the algorithm gives a more optimum approximation to the min-sum decoding algorithm. This new technique outperforms for medium and short length codes and for small number of iterations, which make it suitable for practical applications and hardware implementation. © 2012 Springer Science+Business Media B.V.