EdgeFaaS: A Function-based Framework for Edge Computing

This paper introduces EdgeFaaS, a function-based framework for programming across heterogeneous and geographically distributed IoT, edge, and cloud resources. It uses function virtualization and storage virtualization to provide consistent interfaces for function deployment, execution, data storage, and access. The reported implementation supports workflows including video analytics, federated learning, and audio classification. Evaluation is conducted on a real testbed containing more than 100 distributed IoT devices, edge servers, and cloud services, enabling users to vary function placement and federated-learning cluster configurations while studying computation, communication, speed, and accuracy tradeoffs.