Asymptotic Analysis and Optimization for Generalized Distributed Fountain Codes

Abstract

A novel generic fountain coding scheme for multiple access networks over erasure channels is proposed and it can be applied to several instances. Moreover, asymptotic analysis using And-Or tree analysis technique is evaluated and an efficient joint optimization method employing linear and nonlinear programs is presented. Furthermore, selective distributed Luby Transform (SDLT) codes and unequal error protection (UEP) codes are evaluated as two special cases. Simulation results reveal that the proposed coding scheme can reduce decoding overhead and provide strong UEP property across sources.