Prof. Dr.-Ing. Werner Henkel
We propose a linear programming approach to optimize the variable node degree distribution for a Low-Density Parity-Check code that will provide UEP inside a modulation symbol. Based on a rate maximization criterion, the optimization of the degree distribution is designed for a higher order hierarchical constellation and requires splitting the variable node degree distribution into sub-degree distributions corresponding to the different modulation classes. Once the optimized distributions are available, we continue with obtaining the parity-check matrix of the code and with assigning the different bits inside a codeword to the channels with different noise variances. The noise variances are priorly obtained using the different bit-error probabilities of the bits inside a symbol. As a result, we opt for an irregular LDPC code and we add irregularities to the channel by using different modulation classes originating especially from different distances in a hierarchical modulation signal set. Our aim is to assess if the unequal protection modulation supports in any way the resulting variable node degree distribution. Furthermore, we evaluate the performance of the obtained codes for different constellation parameters by comparing the bit-error ratio performance curves.
[Report] Unequal Error Protection Modulation to Support Low-Density Parity-Check Code Design