Joint optimization for cooperative content delivery network design

Abstract

Content delivery network (CDN) with a high quality-of-service (QoS) requirement deals with three challenging problems: 1) a cache server deployment problem to determine where to locate cache servers, 2) a request routing problem to settle which cache server serves a user request, and 3) a content replication problem to decide which content is stored in a cache server. Cooperation among cache servers significantly improves the user QoS by increasing the cached content diversity of the network. Since the aforementioned problems are tightly coupled, we formulate a new CDN model which jointly considers all the three problems under the cooperative content replication with optimal solutions using integer linear programming. As the joint optimization problem is an NP-hard problem, we propose a greedy heuristic algorithm which is close to the optimal solution. Furthermore, we also suggest the dynamic algorithm determining request routing and stored contents with cache server positions obtained from the optimal solution. In the simulation part, we consider two scenarios for application of the algorithm, where the popularities of contents are identical and are different among nodes which are homogeneous and heterogeneous scenarios, respectively. The three kinds of simulations are conducted. First, the simulation on $2\times3$ grid and tree topologies at the homogeneous scenario is conducted. Second, the simulation on $2\times3$ grid and tree topologies at the heterogeneous scenario is conducted. By conducting four simulations, the performance between the cooperation and the non-cooperation models are compared and the performance of the heuristic algorithm is investigated. Third, the performance of the algorithm is investigated on the actual topology sampled from the Level 3 CDN network. In all three cases, the performance evaluations are conducted between optimal solution and solution from the dynamic algorithm. By increasing the cached content diversity in the network,...