Nearly Optimal Algorithm for Placing Content-Replication in Elastic Optical Networks

Abstract

We propose a network-aware light-weight content-replication placement algorithm (NetLCP) for efficient large-scale content distribution in elastic optical networks. We previously proposed an architecture of information-centric elastic optical transport networks called i-CEON for efficient large-scale content delivery. In i-CEON, we deploy in-network cache functions (ICFs) to reduce redundant use of network resources. The key lies in how to adequately place replicated content in ICFs so as to improve the efficiency of optical paths. In our previous work, we formulated the joint optimization of such a problem as mixed integer linear programming. Though the previous method always finds the optimal solutions, it fails to ensure network scalability and quick responsiveness. To overcome this issue, we develop a nearly optimal heuristic algorithm called NetLCP that improves efficiency of network resources with practical computational overhead. We conducted intensive numerical experiments and demonstrated that NetLCP always found nearly optimal solutions and indicated fewer than 10% of approximation errors compared with the optimal solutions.