Cross-layer Raptor coding for broadcasting over wireless channels with memory

Abstract

Raptor codes are a class of rateless codes that have been shown to provide promising performance in erasure channels, and more recently, in noisy channels. This paper investigates the performance of application layer Raptor codes for broadcasting services over wireless channels with memory. A hybrid erasure-soft decoding algorithm is proposed as a cross-layer protocol for application layer raptor codes. These protocols relay corrupted packets into the application layer. The resulting hybrid error-erasure channels are modeled by a hierarchical Markov channel model. Capacity evaluation and simulation results show that the proposed cross-layer decoding algorithms outperform existing erasure decoding schemes significantly without any modification to the transmitter. The effects of channel memory and other parameters are also studied by simulation.