Non-uniform EWMA-PCA based cache size allocation scheme in Named Data Networks

Narges Mehran,Naser Movahhedinia
DOI: https://doi.org/10.1007/s11432-016-0501-5
2017-08-08
Abstract:As a data-centric cache-enabled architecture, Named Data Networking (NDN) is considered to be an appropriate alternative to the current host-centric IP-based Internet infrastructure. Leveraging in-network caching, name-based routing, and receiver-driven sessions, NDN can greatly enhance the way Internet resources are being used. A critical issue in NDN is the procedure of cache allocation and management. Our main contribution in this research is the analysis of memory requirements to allocate suitable Content-Store size to NDN routers, with respect to combined impacts of long-term centrality-based metric and Exponential Weighted Moving Average (EWMA) of short-term parameters such as users behaviors and outgoing traffic. To determine correlations in such large data sets, data mining methods can prove valuable to researchers. In this paper, we apply a data-fusion approach, namely Principal Component Analysis (PCA), to discover relations from short- and long-term parameters of the router. The output of PCA, exploited to mine out raw data sets, is used to allocate a proper cache size to the router. Evaluation results show an increase in the hit ratio of Content-Stores in sources, and NDN routers. Moreover, for the proposed cache size allocation scheme, the number of unsatisfied and pending Interests in NDN routers is smaller than the Degree-Centrality cache size scheme.
Networking and Internet Architecture
What problem does this paper attempt to address?