Wireless Content Caching in Sliced Cellular Networks with Multicast Beamforming

In this paper, we consider the joint caching and multicast beamforming optimization in cellular networks with caching functions equipped on each base station (BS), to facilitate the network-wide utility improvement by means of content caching among multiple enhanced Mobile BroadBand (eMBB) slices. Firstly, by presenting the two-timescale setup and sample average approximation (SAA) technique, we decouple the intractable and stochastic joint clustering, caching and multicast beamforming optimization problem into two different timescales, the basic principle behind which is to iteratively solve the large- and small-timescale one on the basis of SAA and semidefinite relaxation (SDR), respectively. Then, we leverage the linear programming (LP) theory to solve the formulated joint clustering and caching problem at the large timescale. In particular, the integer linear programming (ILP)-based large-timescale problem is reformulated and relaxed as an LP one, along with a verification that its optimal solution is exactly equivalent to that of its relaxed LP counterpart. Finally, simulations are conducted to demonstrate the effectiveness of proposed algorithm, revealing that the network-wide utility could benefit from both the multicast beamforming and content sharing among multiple slices.