CAREER: Meeting the Promise of 5G with Synergistic RAN Slicing

Emerging applications such as augmented reality, telerobotics, and Internet-of-Things (IoT) will revolutionize education, healthcare, training, and agriculture. Such societal applications rely on cellular connectivity, and impose stringent bandwidth, latency, and scalability requirements on the cellular infrastructure. Current network infrastructure’s inability to meet these requirements has been a key impediment for realizing these applications effectively. Next generation (5G/6G) cellular standards promise to meet these requirements, but the mechanisms for delivering this promise are yet to be worked out. The proposed research will develop techniques that will aid in fulfilling this promise, with the end goal of enabling such emerging societal applications. The proposed research will also contribute to education by engaging students at different levels to assist with the research activities, and curriculum development at undergraduate and graduate levels. All code and teaching materials developed as part of this proposal will be made publicly available. The proposal will build upon the notion of radio access network (RAN) slicing, a new feature in 5G cellular standards, defined to meet the above requirements. With RAN slicing, each slice (composed of a group of users) enters into service-level agreements (SLAs) with the network operator who divides radio resources among slices so as to achieve efficient spectrum usage, meet the SLAs, and allow each slice to customize how their share of resources are allocated to their users. Current proposals for RAN slicing adopt a two-staged approach where the network operators first divide their radio resources among slices, and slice operators then divide their share of resources among their users. However, such a two-staged approach ignores the inherent inter-dependencies between the network operator’s inter-slice resource allocation decisions and the slice operator’s intra-slice policies, which results in inefficient spectrum usage and lower user throughput. This proposal puts forward a new approach to RAN slicing that explicitly embraces these inter-dependencies, allowing the inter-slice and intra-slice resource allocation decisions to feed into one another, so as to better meet the network-wide objective of efficient spectrum usage, as well as the customizable objectives of individual slices. To that end, the proposed research will break its efforts along the following dimensions: (1) new network-wide (inter-slice) resource allocation mechanisms that are informed by intra-slice policies, (2) new intra-slice resource allocation mechanisms, driven by the requirements of emerging applications, that inform the network-wide mechanisms, and (3) an end-to-end system that brings together the above mechanisms, tackling practical challenges around timely decision-making at scale. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.