Fiber Access network architecture featuring distributed processing and function virtualization

Abstract

The increasing demand for converged media services, massive growth in bandwidth intensive data-centric applications and a desire to roll out new services at an everfaster clip raise great economic challenges to the network and service delivery ecosystem. Currently, access networks rely upon centralized packet multiplexers to perform network and protocol processing functions, manage subscriber state, and aggregate network traffic. On the other hand cloud-based applications are built around public or private clouds, web-based tools and resources executed on centralized infrastructure accessible over core networks and the Internet. However, this monolithic model can lead to capacity and space bottlenecks, compromise network services because of a lack of common feature sets, and limit the revenue that data services can produce for operators. This paper proposes a novel fiber access network infrastructure that can support distributed processing and function virtualization exploiting the HYDRA (HYbriD long-Reach fiber Access network) architecture. The proposed scheme leads to a higher degree of node consolidation and network and service convergence. We discuss how logical network virtualization and efficient resource orchestration for flexible service creation and delivery can be achieved describing indicative future services that could exploit the enhanced functionality and distributed processing paradigm.