Multi-source spinal coding for coded caching multicast transmissions in wireless networks

Abstract

Recently, coded caching has been treated as a promising technique to alleviate the traffic burden in wireless networks. To support high efficient coded caching multicast transmissions, the time-varying heterogeneous channel conditions need to be considered. In this paper, a practical and novel multi-source spinal coding (MSSC) scheme is developed for coded caching multicast transmissions under heterogeneous channel conditions. By exploring joint design of network coding and spinal coding (SC), MSSC can achieve unequal link rates in multicast transmissions for different users. Moreover, by leveraging the rate-less feature of SC in our design, MSSC can well adapt the link rates of all users in multicast transmissions without any feedback of time-varying channel conditions. A maximum likelihood (ML) based decoding process for MSSC is also developed, which can achieve a linear complexity with respect to the user number in the multicast transmission. Simulation results validate the effectiveness of the MSSC scheme. Compared to the existing scheme, the sum rate of MSSC in multicast transmissions is improved by about 20%. When applying MSSC in coded caching systems, the total transmission time can be reduced by up to 48% for time-varying channels.