Downloaded: 962 times
© 2020 Gianmarco Baldini | Piotr Fröhlich | Erol Gelenbe | Jose Luis Hernandez-Ramos | Mateusz Nowak | Slawek Nowak | Stavros Papadopoulos | Anastasis Drosou | Dimitrios Tzovaras
Cyber attacks on the Internet of Things (IoT) can be the source of major economic damage. They can disrupt production lines, manufacturing processes, and supply chains. They can adversely impact the physical safety of vehicles and transportation systems, and damage the health of living beings both through supply chains for food, medicines, and other vital items, as well as through direct attacks on sensors and actuators that may be connected to vital functions. Thus, securing the IoT is of primary importance to our societies. This paper describes the technical approach that we adopt for IoT security in the SerIoT Research and Innovation Project that is funded by the European Commission. We first discuss the risk scenario for the IoT and briefly review approaches that have been developed to mitigate such risks. Then, we discuss a policy-based lightweight approach that mitigates risks at the level of the attachment of IoT devices to a network. We follow this with a detailed proposal based on using a distributed Machine Learning approach to risk and attack detection in real time, as well as suggestions for future work.