Reduced-complexity decoding of low-density parity check codes based on adaptive convergence

Low-density parity-check (LDPC) codes have recently been considered as a viable candidate for forward error correction in system level hardware-redundant, fault-tolerant logics. An important factor that influences the choosing of a specific FEC technique in a nano-scale system implementation is its real-time performance, namely its computational complexity. In this paper, we propose a set of rules to decide whether a variable node in a LDPC decoder should update its value in subsequent iterations of the decoding process, or be considered as converged. We show that by carefully choosing the convergence rules for variable nodes, significant reduction of decoding complexity can be achieved with endurable performance loss.