Name Centric Prefetching in Named-Data Networking
Abstract
This thesis presents Name-Centric Prefetching (NCP) that prefetches data before users request for them to improve user experience. NCP seeks to identify user request patterns solely based on names in prior interests, and without any other knowledge about user applications. As such, NCP can be easily implemented without any modifications for applications. A prototype of NCP has been built within Named-Data Networking (NDN). The implementation includes multiple modules that make it easy to implement and test new prefetching algorithms and to manage the computation, storage, and bandwidth overheads. The utility of NCP is evaluated under two scenarios, one derived from a real-world trace from a Google cluster and the other constructed by mimicking the behaviors of a variety of applications, and three different prefetching algorithms. Testbed emulation results demonstrate that NCP is able to significantly reduce end-to-end latency experienced by users while incurring little additional network traffic.
Citation
Khan, Mohd Faisal (2020). Name Centric Prefetching in Named-Data Networking. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /192537.