Summary: | To alleviate the backhaul burden and reduce user-perceived latency, content caching at base stations has been identified as a key technology. However, the caching strategy design at the wireless edge is challenging, especially when both wired backhaul condition and wireless channel quality are considered in the optimization. In this paper, taking into account the conditions of the backhaul in terms of delay and wireless channel quality, joint design and optimization of the caching and user association policy to minimize the average download delay is studied in a cache-enabled heterogeneous network. We first prove the joint caching and association optimization problem is NP-hard based on a reduction to the facility location problem. Furthermore, in order to reduce the complexity, a distributed algorithm is developed by decomposing the NP-hard problem into an assignment problem solvable by the Hungarian method and two simple linear integer subproblems, with the aid of McCormick envelopes and the Lagrange partial relaxation method. Simulation results reveal a near-optimal performance that performs up to 22% better in term of delay compared with those in the literatures at a low complexity of O(nm<sup>3</sup>/ε<sup>2</sup>).
|