Coded joint pushing and caching with asynchronous user requests

Pushing and caching are recognized as promising techniques to improve network capacity and handle explosive growth in data traffic. Caching schemes for synchronous requests have been extensively investigated, but little attention has been paid to the asynchronous case. However, asynchronism is an inherent property of user requests. In this paper, request delay information (RDI) is introduced to characterize the asynchronous user requests. Based on RDI, a coded joint pushing and caching (C-JPC) method is proposed to minimize the network traffic by jointly determining when and which data packets are to be pushed and whether they should be cached. Optimal offline and online C-JPC policies for noncausal and causal RDI can be obtained by solving optimization problems, which however are intractable. Fountain coded caching (FCC) and generalized coded caching (GCC) methods are further proposed to give sub-optimal C-JPC policies with low complexity. In addition, lower and upper bounds on the optimal traffic volume are presented. It is shown that FCC and GCC achieve optimal or near-optimal volumes of traffic in some special cases. Simulation results demonstrate that C-JPC brings significant multicasting gains.