EdgeFaaS Framework Unifies IoT, Edge, and Cloud for Heterogeneous Computing

Edge computing faces unique challenges due to the high diversity in capabilities and capacities of distributed resources across IoT devices, edge servers, and cloud. Existing distributed computing frameworks struggle with this heterogeneity and scale. To address this, researchers from Arizona State University (Neha Vadnere, Yu-Ting Wang, Yitao Chen, Sreehari Sadesh, Ming Zhao) introduce EdgeFaaS, a function-based framework that uses function virtualization and storage virtualization to abstract physical resources into consistent virtual interfaces. This allows edge applications to deploy and execute functions and store/access data seamlessly across the entire continuum.

EdgeFaaS was evaluated on a testbed of over 100 geographically distributed IoT devices, edge servers, and cloud services. It supports three representative workflows: video analytics, federated learning, and audio classification. A key feature is user-configurable deployment: for video processing, users can vary function placement across IoT, edge, and cloud to study computation vs. communication tradeoffs; for federated learning, they can adjust cluster count and size to balance training accuracy and speed. The framework provides comprehensive support for diverse edge computing workflows while maintaining flexibility for important tradeoff exploration.