Networking at Information Centric Edge
Abstract
More and more heterogeneous devices are getting connected to the Internet, leading us to the era of Internet of Things. However, this trend is hardly sustainable under today's Internet, because of the inefficiency of IP networking for user information and the potential long delay of cloud services. To address these issues, information centric networking and edge computing have been proposed in the literature. These two concepts naturally fit into each other, resulting in Information Centric Edge, an emerging network architecture where heterogeneous devices are connected mostly through wireless to nearby edge servers that have computation resources provisioned, and their inter-networking is centered on user contents. With information centric edge, users can get what they want faster. However, such network architecture also face challenges including limited edge resources, strategic users, implementation and so on.
In this dissertation, we address these challenges of realizing Information Centric Edge by designing and implementing systems and algorithms therein. To cope with the limited capacity of edge servers, we study an dynamic service caching problem and propose an online algorithm with provable worst-case performance guarantee. We also investigate a content caching problem that calls for joint optimization of content placement, cache selection, and video version selection, and present practical distributed algorithms that achieve close to optimal performance in implementation. To handle strategic users, we design a non-monetary mechanism via efficient cost allocation which makes distributed individual optimization converge to the global optimum. The same method is generalized to distribute revenue of a cyber-enabled manufacturing system to its geo-distributed suppliers. Besides, we are building an Information Centric Edge testbed with a programmable network stack for intelligent caching, routing, and scheduling algorithms to support emerging virtual reality applications in the era of Internet of Things.
Subject
Information centric networkingedge computing
online algorithms
named data networking
game theory
Citation
Zhao, Tao (2019). Networking at Information Centric Edge. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /186455.