Content Placement in Cache Networks Using Graph-Coloring

TL;DR

This paper introduces a graph-coloring based approach for cache placement in small cell networks, reducing macro base station load by approximately 25% compared to traditional methods.

Contribution

It models cache placement as a graph-coloring problem and proposes a simple, efficient coloring algorithm based on point processes for improved caching performance.

Findings

Load on macro base station reduced by 25%

Proposed method outperforms conventional caching policies

Efficient graph-coloring algorithm simplifies cache placement

Abstract

Small cell densification is one of the effective ideas addressing the demand for higher capacity in cellular networks. The major problem faced in such networks is the wireless backhaul link and its limited capacity. Caching most popular files in the memories of small cells base station (SBSs) is an effective solution to this problem. One of the main challenges in caching is choosing the files that are going to be stored in the memory of SBS. In this paper, we model the described caching problem as a graph. This graph is divided into four sub-graphs including placement, access, SBSs and delivery graphs. By making some modifications to the SBS graph, we convert it to a graph that can be colored. Coloring of the generated graph is NP-hard and we use an algorithm proposed in graph-coloring area to color it. To overcome the complexity of above coloring technique, we then propose a simple graph-coloring method based on two point processes, Matern Core-type I and II. We model our network with a new weighted graph which simply can be colored and after that the files are cached accordingly. We evaluate the performance of our proposed methods through simulations. Our results show that by employing out proposed method in a typical considered SBSs network, the load on the macro base station can be reduced by around $25 \%$, at the distribution parameter of popularity of files equals to $0.6$, compared to conventional policy which caches the most popular content every where.