Research On Network Low-Density Parity-Check Codes

By processing and forwarding the received data from multi-links, network code can effectively improve the throughput of the network, which is widely concerned in recent years. In the wireless relay network, the transmission efficiency of the relay node can be improved effectively without loss of diversity gain via the network coding technique. In order to improve the BER performance of the relay network, joint network coding and channel coding can be used in the relay node, which has been received much attention during the past few years.A new relaying strategy based on the joint network-channel coding is proposed in this paper to approach the capacity of the orthogonal multi-access relay channel. Assume the signal transmitted from the source node can be decoded correctly in the relay node. Based on a new designed LDPC code, the newly designed relaying strategy permits the relay node jointly encodes the message from multi-source nodes and forwards the parity bits to the destination node. Then the destination node jointly decodes the messages from the source and relay nodes. We call such a procedure as a network LDPC code.Under Gaussian assumption, we derive a generalized bilayer Gaussian Approximation algorithm and a generalized bilayer EXIT Charts, based on which the optimization algorithm is proposed to search the degree distribution of the network LDPC code. The degree distributions of the network LDPC codes are optimized by maximizing the relaying efficiency of relay node under a given decoding threshold. Moreover, the upper bound of the sum capacity of the orthogonal multi-access relay channel is analyzed. The simulations show that, in a network with two sources, one relay and one destination, the gap between the threshold and the system performance with the designed network LDPC code is about 0.6dB, and the gap between the upper bound of the sum capacity and the decoding threshold is less than 0.35dB.Under Rayleigh fading assumption, a generalized bilayer Gaussian Approximation algorithm based on the error probability is derived, based on which an optimization algorithm is proposed to search the degree distribution of the network LDPC code. Results show that the gap between the threshold and the system performance with the designed network low density parity check code is about 0.8dB.