A Capability for Autonomous IoT System Security: Pushing IoT Assurance to the Edge

Abstract

Complex systems of IoT devices (SIoTD) are systems that have a single purpose but are made up of multiple IoT devices. These systems are becoming ubiquitous, have complex security requirements, and face a diverse and ever-changing array of cyber threats. Issues of privacy and bandwidth will preclude sending all the data from these systems to a central place, and so these systems cannot totally rely on a centralized cloud-based service for their security. The security of these systems must be provided locally and in an autonomous fashion. In this paper, we describe a capability to address this problem, explain specifications for the system, present our work on SIoTD assurance, and show initial results of a novel edge-based application of machine learning to build this capability.