Improving the freshness of NDN forwarding states

Named Data Networking (NDN) is a new Internet architecture that replaces today's focus on where - addresses and hosts - with what - the content that users and applications care about. A unique advantage of NDN over IP is the adaptive forwarding plane, which, by observing the performance of Interest/Data exchange, can dynamically select the best performing forwarding path, detect and recover from failures, load-balance across multiple paths, and mitigate attacks such as prefix hijacking and DDoS. A key component of adaptive forwarding is interface ranking, namely when and how to update the interfaces' metrics and rank them. As we will point out in this paper, however, the existing interface ranking scheme suffers from the problem of outdated forwarding states. Using two concrete problems, SRTT slow-convergence and probing oscillation, we illustrate how outdated forwarding states can impact the forwarding performance. We propose new forwarding strategies with Adaptive SRTT Update (ASU) and Proactive Probing to achieve up-to-date forwarding states, and evaluate how these strategies are able to address the two problems. Both theoretical analysis and simulation results show that the new strategies can reduce SRTT convergence time by 37.9% and the loss rate by 75% to 94.75%, compared to the existing interface ranking strategies.