A study of user-assisted content distribution performance in named data networking

Abstract

In this paper, we study the impacts of user's content request and upload strategies on the user-assisted content distribution performance in named data networking (NDN). Through extensive simulations based on real a BitTorrent trace, we show that content request strategy affects the aggregate user upload throughput and the efficiency of in-network caching and multicast, while content upload strategy affects traffic localization by influencing the content availability in the same network domain. In other words, user strategies have noticeable impact on the content distribution (e.g., the best case reduces the download completion time by 50.6% compared to the worst case) even though most P2P content distribution functions are supported by NDN. This means that elaborate design for user strategies is also required in addition to current research on NDN architecture itself.