Design of LDPC Codes for Binary-input Gaussian Interference Channels

Due to the existence of multiuser interference, codes optimized for point-to-point links may be unsuitable for multiuser networks. In this paper, we are concerned with the design of lowdensity parity-check (LDPC) codes for binary-input Gaussian interference channels (GIFC). We utilize a Monte-Carlo approach based on the downhill-simplex algorithm to obtain good LDPC degree distributions for GIFC. Simulation results show that the constructed codes perform better than codes optimized only for point-to-point links. For symmetric GIFC with coefficient h(2) = 0.75, coding gain about 0.6 dB is obtained. These results indicate that, as opposed to using off-the-shell codes designed for point-topoint links, new codes should be designed in multiuser networks.