Recently, edge computing has emerged as a promising computing paradigm to meet stringent quality-of-service requirements of an increasing number of latency-sensitive applications. The core principle of edge computing is to bring the capability of cloud computing in close proximity to mobile devices, sensors, actuators, connected things and end users, thereby supporting various types of services and applications at the network edge. In this article, we design capacity-centric FiWi broadband access networks enhanced with edge computing as well as resulting fiber backhaul sharing and computation offloading capabilities. More specifically, we introduce the concept of FiWi enhanced two-level edge computing at the access edge cloud and metro edge cloud. To guarantee low end-to-end latency, we propose a TDMA based polling scheme for resource management. Furthermore, given the vital importance of experimentally demonstrating the potential and practical limitations of edge computing, we develop an experimental testbed for edge computing across converged FiWi broadband access networks. The proof-of-concept demonstration of the testbed is studied in terms of response time and response time efficiency of both edge clouds, including their respective energy consumption.
Rimal, B.P., Maier, M., Satyanarayanan, M.
IEEE Communications Magazine, 56(160-167), August 2018