Implementation and performance evaluation of keyword-based content retrieval

Abstract

Information-Centric Networking (ICN) is receiving attention as an emerging network technology. Due to its attractive features, such as accessing content by its name, in-network caching, etc., ICN is expected to apply essentially to the Internet of Things (IoT). IoT applications typically collect a large amount of data widely distributed in various locations. For such usage, we proposed the architecture of Keyword-Based Content Retrieval (KBCR) in a previous paper, as an extension of ICN. In KBCR, a requester sends one request with some keywords, and can receive related content in one response. In this paper, we describe the implementation and performance evaluation of KBCR, and show the feasibility and efficiency of it. For implementation, we implemented KBCR protocol in our testbed system to confirm its functions and developed a demonstration application to show its effectiveness. For performance evaluation, we compared KBCR and NDN by queueing analysis using measured data on the prototype, and showed that KBCR provides better performance both in a requester load and the network delay/throughput.