Improved dual adaptive importance sampling method for LDPC codes
Yuan Liu, Weigang Chen, Jinsheng Yang, Tongxin Dong · 2012
An improved dual adaptive importance sampling (DAIS) method is proposed to accelerate the simulation of low density parity check (LDPC) codes at high Signal-to-Noise Ratios (SNRs). First, we observed an interesting phenomenon that the conditional probabilities of the error events, which are caused by different noise realizations in the same control quantity bin at the different SNRs, are nearly identical in DAIS. Then, we obtain the joint probabilities of the unconstrained simulations at the high SNRs by substituting the conditional probabilities at the high SNRs with that at the low SNRs based on the observation. This spares the decoding computations of the unconstrained simulation at the high SNRs, and therefore reduces the simulation time. For the specific LDPC code, simulation results verify that the improved method achieves significant simulation gain compared with traditional DAIS, while maintaining the same accuracy.