Improved Content Finding in Named Data Networking

Suwannasa, Atthapol and Broadbent, Matthew and Mauthe, Andreas (2021) Improved Content Finding in Named Data Networking. PhD thesis, Lancaster University.

Text (2021AtthapolPhD)
2021AtthapolPhD.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (4MB)

Abstract

In today’s Internet, the current architecture may not be able to support various challenges (e.g., security, mobility, scalability, and quality of service) in a sufficient level. Information-centric communication model is expected to address the bottleneck of the traditional host-centric model. A number of Information Centric Network (ICN) approaches have been proposed by aiming to replace or augment the current host-to-host routing architecture. ICN focuses on finding and transmitting content to end-users and content routing is location-independent, thereby being able to support multi-sourcing for content consumers. Named Data Networking (NDN) is one of the promising ICN proposals that allows users (i.e., consumers) to find content objects by their names. In the default forwarding strategy of NDN, an interest packet is forwarded to locate content. A corresponding data packet will be returned back in the reverse path to its requester and will be replicated along this path (called on-path caching). When a consumer requests a content object, it may be found at an intermediate on-path cache. However, several replicas that are often cached off-path especially in nearby nodes of the consumer’s vicinity could be the better potential source but they are not effectively utilised, causing a worse than necessary delivery efficiency. Therefore, this thesis investigates the potential of off-path content finding in NDN. We examine how we can design a flexible and efficient solution to supplement the existing NDN architecture. We then propose a new design called a Vicinity-based Content Finding scheme (VCoF) to utilise nearby replicas in each vicinity for improving content finding. This includes analysing the efficiency of the proposed scheme in comparison to default NDN. We consider content popularity, which can impact content finding results due to the different number of content replicas (i.e., content availability). We also explore our scheme in supporting mobility, particularly for the issues of missing content because of handover. Through a prototype implementation, we evaluate the delivery efficiency against overhead costs in different scenarios, made possible through effective deployment on real NDN environments.