Parallelized Optimization of LDPC Decoding Algorithms

One of the major problems in communications engineering is to efficiently recover errors of transmitted data blocks by forward error correction (FEC). Low Density Parity Check (LDPC) codes are very popular for FEC. They are known to be one of the best coding schemes in terms of error correction performance. Because the decoding of practically used LDPC codes is NP-hard, only suboptimal algorithms can be applied. To optimize decoding algorithms in terms of decoding performance, a high amount of numerical simulations is needed. We present a parallelized method for evaluating and optimizing the performance of LDPC decoding algorithms. Using performance criteria of the decoding algorithms, the optimization is done by applying nonlinear optimization methods to select algorithm parameters.