CSR: Small: Toward a Scalable, Multi-Tenant, Edge-Cloud Infrastructure for Real-Time Computation

George Washington University, Washington DC

Investigators

Abstract

Applications are increasingly cloud-connected, and are dependent on fast service that is impossible if their requests must travel to a cloud data center. Edge-computing places a smaller amount of computation on the network very close to the applications themselves, providing fast servicing of requests. This enables new application types such as virtual and augmented reality, which want to display information to users with imperceptible overheads. Unfortunately, current cloud computing systems cannot support the density or diversity of applications found in edge-computing deployments. Each connected device -- be it a cell phone, a smart building sensor, or an autonomous vehicle -- may require different software to run in the edge cloud, and mobile users may constantly be arriving and departing. Since these software components cannot necessarily be trusted, resources must be carefully isolated to provide security and prevent interference. This research creates EdgeOS, which provides strong multi-tenant isolation while ensuring quick response times, and efficient use of computational resources. EdgeOS introduces a number of system abstractions to enable multi-tenant edge-clouds: a new lightweight process abstraction that enables a very large number of clients to be served efficiently, and with isolation; controlled communication to ensure safe, efficient coordination; and system support for controlling the service time of various clients to ensure the timely processing of requests. These new operating system abstractions will help edge clouds realize their potential in offering low-latency services. The EdgeOS research is paired with an educational initiative that will deepen the systems curriculum, engage undergraduate students in research, and involve a diverse body of students. 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.