Edge Computing Architecture for applying AI to IoT

Abstract	The proliferation of connected IoT devices creates a big data problem for AI based approaches. The response time required by such devices necessitates IoT data to be processed at the edge, but the edge typically lacks the resources to learn the AI models. We present an architecture which preserves the advantages of both edge processing and server-based/cloud-centric computing for AI algorithms. We discuss how policy management can be used to improve and support this architecture.
Authors	Seraphin Calo (IBM US) Maroun Touma (IBM US) Dinesh Verma (IBM US) Alan Cullen (BAE)
Date	Dec-2017
Venue	1st IEEE Big Data International Workshop on Policy-based Autonomic Data Governance (PADG)