摘要

Rateless codes are popular erasure-filling codes, which are capable of providing the potentially infinite amount of redundancy at the packet level. However, the optimal packet size has not been carefully studied in a wireless communication system from both the application layer and the physical layer. In this paper, a cross-layer approach is proposed to find an optimal rateless-coded packet size in order to maximize the goodput of a wireless system delivering fixed-length messages over block-fading channels. We analyze the system performance and identify the impacts of packet size on the reliability and efficiency. Moreover, the goodput can be improved using the proposed approach by balancing reliability and efficiency. We also compare the performance of the rateless coding system with the conventional system utilizing channel codes only. A better performance of the rateless coding system is shown at relative low signal-to-noise ratios and the goodput is enhanced by about 20 % at 0 dB.