Object Placement for Cooperative Caches with Bandwidth Constraints

Abstract

The projected growth in video traffic delivered to mobile devices is expected to stress the backhaul and core of a broadband wireless network. Caches deployed at the edge elements, such as base stations, are one of alleviating this stress. Limits on the sizes of the base station caches and restrictions on frequent upgrades to the hardware necessitate that techniques that can increase the hit rates with the growing traffic, given the constraints, be explored. In this paper, we consider using cooperative caching schemes for the purpose. The edge elements are connected via bandwidth-constrained links, and hence, the assumption made in most prior work that the cooperating nodes are located on a high-speed network do not apply here. We show that the problem of placing objects to maximize hit rate in such a bandwidth-constrained caching system is NP-hard in the strong sense. We develop an efficient placement algorithm when the caches have identical characteristics and show that its performance is within a constant factor of the optimal under practical conditions. We also discuss how to extend the algorithm for the non-identical case. Our simulation experiments show that in practice, the performance of our algorithm is very close to the optimal and a few tens of cooperating nodes are sufficient to significantly increase the hit rate even with a 1% base cache size.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.