User-Centric Quality of Service in Next Generation SD-WAN

点击翻译按钮获取中文摘要

The emergence of Intent-based Networking (IBN) in Software-Defined Networks has promised the realization of  "programmable" networks in which the network administrators are able to deploy new services and network functions through abstract commands (e.g. "Configure protocol xx") or high-level scripting languages, and a network orchestration layer would translate such intents into device-level actions. However so far this new degree of flexibility has been limited to basic network protocol and management configuration. The next generation Software-Defined  Wide-Area networks (SD-WAN) is expected to take this flexibility to a whole new paradigm, where specific network services are defined with specific end-user quality expectation (performance, reliability, security etc), e.g. "Deploy a video stream for N customers in Australia with Excellent quality", in which the excellence is defined through key performance indicators as well as user satisfaction ratios. The Intent-based SD-WAN not only configures and deploys these services, but also monitors the quality requirements, and dynamically and proactively adjusts network resources to maintain the quality level. The objective of this research program is to focus particularly on challenges regarding the definition and maintenance of Quality of Experience (QoE) for SDWAN services, and to design solutions to address these challenges. The design of QoE-aware SDWAN service will include three dimensions: The network, the administrator, and the user Experience. Challenges to address include: dynamic resource allocation,  transport across heterogeneous (and possibly virtualized or cloud-based) network infrastructures, service reliability, translation of service-level intents into the set of network parameters that could potentially affect quality of service, and new and more complex service-level agreements, among others. In order to facilitate service introduction or change to existing services, an intent-based orchestration layer must be designed and mapped to network parameters. This abstraction layer must integrate routing and security policies and automate the creation and installation of network services. Also from a user perspective, user feedback mechanisms must be integrated into this layer to allow real-time quality measurements for adjustment of network service paths in case the quality of experience deteriorate. This research program will attempt to investigate and address the above issues as related to the design of next generation network services, architectures and protocols. It will also  contribute to the training of the next generation of researchers in the field of communication and networking in Canada, which remains particularly in demand for decades; and providing them with the skill set that is most relevant to this field in both academic and industrial research.