Design and performance evaluation of tag caching router architecture for CGM content

Abstract

In recent years, social interest in viewing consumer-generated media (CGM) contents has been rapidly growing, so their network traffic has been increasing considerably. To cope with this new increase in network traffic, we propose a tag caching router (TCR) architecture that supports folksonomies-based search and content caching for CGM content. Folksonomies are key words or metadata attached to the associated content by CGM-content creators, providers, and viewers to characterize the CGM content. The falksonomies-based search helps a user to find his/her interesting content because it automatically collects from the network the information of candidate contents with the folksonomy specified by the user and presents it as a content list to the user. The TCR architecture caches both content lists for content search and the associated contents for content downloading. We call this enhanced caching method as content caching. The performance of the proposed caching architecture was evaluated by performing a simulation for a wide variety of CGM contents with different traffic characteristics. The simulation results indicate that under a network model with 5 × 5 mesh-topology TCRs, content caching enables the TCR architecture combined with a conventional URI-based search to support a maximum of six-times more CGM access requests due to content caching than the number of accesses that can be supported by a conventional TCP/IP-based end-to-end architecture with a URI-based search. The number of accesses supported by the TCR architecture combined with folksonomies-based search increases to 17-times the number of accesses supported by the conventional TCP/IP-based end-to-end architecture with a URI-based search.