Design of LDPC Codes Degree Sequences on Random Particle Swarm Optimization over Mixed Channel

—In optical, magnetic recording and FrequencyHopping communication with narrow band interference, white Gaussian noise and random erasures are existing all at once. We call this mixed channel. To analyze performance of low density parity-check (LDPC) codes over mixed channel, the stability condition for LDPC codes over it is proposed, from which we can draw a conclusion that a good degree sequence of LDPC codes over Gaussian channel is not optimized over mixed channel. It can also be supported by simulation. To search good degree sequences over mixed channel, the random particle swarm optimization and simulated annealing algorithm are combined to find some capacity-approaching degree sequences in different erasure probabilities. Some high-threshold degree sequences are showed. The threshold of signal-to-noise ratio can increase 1.6159dB than that of the classical degree sequences evaluated by Gaussian approximation over mixed channel.