Security for Cyber-Physical Systems

One of the most important technological developments is the growing importance of cyberphysical systems (CPS). CPS are connected networks of computational nodes that interact closely with their physical environment. Unlike general purpose computers of the type connected to the Internet, the devices comprising CPS are more specialized. These networks can range from being loosely coordinated to tightly federated. CPS systems are also generally designed as platforms on which third parties can operate devices and software. Applications in CPS tend to be more safety-critical. Prominent examples of CPS include self-driving cars and medical device platforms.

One of the challenges is that CPS were not designed with any security.  This roundtable is part of an NSF project focused on rectifying that shortcoming. Attacks on CPS can come both in the digital form and via the physical environment. Typically, each individual device is subject to greater resource constraints than a monolithic server environment. At the same time, each device must be more self-reliant in storing data and performing security, because other nodes in the network may be compromised or otherwise untrustworthy.

The NSF project takes a novel approach to security. The major design shift is that security failures are acknowledged to be inevitable. The NSF project combines multiple layers of protection, including (1) prevention methods such as encryption, (2) fast detection of and recovery from failures, (3) fusion of diverse sensor technologies for robustness, and (4) tamperproof data logging for forensic purposes.

This work is supported in part by NSF CNS-1505799 and the Intel-NSF Partnership for Cyber-Physical Systems Security and Privacy.