Inapproximability results and suboptimal algorithms for minimum delay cache placement in campus networks with content-centric network routers

Abstract

We consider the cache placement problem in campus networks where routers have heterogeneous cache capacity and the objective is to minimize the total delay of all requests. We prove that the problem is NP-hard to approximate to within any factor less than $$n/m^{\epsilon }+\hbox{poly}(m)$$ , where n is the number of routers, m is the number of contents, $$\epsilon$$ is any fixed positive constant, and $$\hbox{poly}(m)$$ is any polynomial function of m. We propose (exponential-time) exact algorithms based on integer linear programming and propose techniques to decompose the network and remove redundant variables for speeding up the computation. By relaxing the integer linear programs, we also propose three (polynomial-time) heuristic algorithms. Numerical results show that the proposed algorithms give a shorter delay than existing cache decision algorithms for content-centric networks.