Abstract:

Based on the particle swarm optimization (PSO) algorithm with gradient, a new design is proposed to optimize the degree distribution in the procedure of Luby transform(LT) encoding. The importance sampling approach is employed to construct an objective function. The estimation for the gradient is obtained by the objective function. And then, the degree distribution can be optimized by PSO with gradient. At last, forms with sparse degree distributions and soliton distributions are optimized with the proposed method respectively. Simulations show that the degree distributions obtained by the proposed method are more efficient compared with that of Robust soliton distribution and importance sampling approach, it can reduce the overhead packets in LT codes with short codelength.

Key words: Luby transform codes, degree distribution, importance sampling, gradient, particle swarm optimization algorithm