Virtual Content-Centric Networking for Realizing Efficient and Secure Content Retrieval and Distribution
圭一郎 塚本 · OUKA (Osaka University Knowledge Archive) (Osaka University) · 2014
building VCCN slices, each of which is composed of multiple VCCN router instances.Through a preliminary performance evaluation of the VCCN implementation, the author shows that the introduction of VCCN has a positive or negative impact on CCN performance and that CCN router virtualization in VCCN incurs a little overhead to CCN in terms of the content delivery time.However, two issues on efficiency of content retrieval and distribution that result from introduction of VCCN have not been addressed.The first issue is to balance the overall network performance and network performance for each VCCN slice in terms of efficiency of content distribution.The second issue is to quickly locate relevant parts in distributed content in terms of efficiency of content retrieval.For resolving the first issue on efficiency of content distribution in VCCN, the author analytically and quantitatively investigates a trade-off among the network fairness for VCCN slices and overall network performance.The author investigates what resource allocation method provides the best balance between the network fairness for VCCN slices and overall network performance in conceivable three allocation methods (i.e., an exclusive method, a shared method and a hybrid method).Using several numerical examples, the author shows that when content request patterns are heterogeneous, a hybrid resource allocation method will provide the best balance between fairness and overall network performance.For resolving the second issue on efficiency of content retrieval in VCCN, the author proposes an application-level approach for improving the efficiency Web browsing, which is the representative Internet usage in content retrieval of the current Internet.The approach called HypErlink Referring Block estimation (HERB) segments Web pages into blocks and infers the existence and location of all relevant content on hyperlinked Web pages based on a block-to-block similarity.Through experiments simulating ordinary Web browsing, the author shows that HERB can infer blocks relevant to a hyperlink with precision and recall that are as high or higher than those of existing methods on a block-based Web search.Furthermore, the two HERB-enabled implementations, namely, a Web proxy and Web browser, are also designed.