An Adaptive Per-Application Storage Management Scheme Based on Manifold Learning in Information Centric Networks.

Yuemei Xu,Yang Li,Tao Lin,Zihou Wang,Guoqiang Zhang,Hui Tang,Song Ci
DOI: https://doi.org/10.1016/j.future.2013.09.016
IF: 7.307
2013-01-01
Future Generation Computer Systems
Abstract:Information Centric Network (ICN) is an emerging network paradigm centered around the named contents rather than the host-to-host connectivity. The common characteristic of ICN leverages in-network caching to achieve an efficient and reliable content distribution but also brings challenges. The in-network caching technique equips all ICN routers with cache storage. However, no existing works focus on the cache storage sharing mechanism among different applications to satisfy the line speed requirements and diversity of applications in ICN. In this paper, we formulate the per-application storage management problem into an optimized resource allocation problem and introduce a manifold learning method to classify the priority of applications. Dynamic programming is adopted to solve the formulated problem and an adaptive per-application storage management scheme is proposed on the basis of the optimal solutions. Extensive experiments have been performed to evaluate the proposed scheme and show that our approach is superior to static partitioning and shared storage schemes.
What problem does this paper attempt to address?