Fast BER Estimation of LDPC Codes

Abstract:
In this paper, we propose an estimation method of a very low bit error rate (BER) of low-density parity-check (LDPC) codes. No analytical tool is available to evaluate performance of LDPC codes, and traditional Monte Carlo simulation methods can not estimate low BER of LDPC codes due to the limitation of time. Cole et al. show a fast frame error rate simulation method of LDPC codes by applying importance sampling (IS). At first, we extend Cole et al.’s method to the BER estimation of LDPC codes. Since the simulation time of the BER evaluation version of the Cole et al.’s method is proportional to the number of total dominant trapping sets used in the simulation, we should appropriately choose the dominant trapping sets to reduce the simulation time. To avoid the difficulty, we propose another simulation method, which selects a dominant trapping set as proportional to its upper bound. Additionally, we show some numerical examples to demonstrate the effectiveness of the proposed method. When we evaluate BER of 10(exp ?16), the simulation time of the proposed method is reduced to 1/6.6 of that of the BER evaluation version of the Cole et al.’s method under the condition of the same accuracy of the estimator.