Trusted Control MachinesLed by Professor Julie McCann, Department of Computing

A control system uses sensor measurements to monitor the behaviour of a system to generate appropriate control actions to achieve the desired system performance, and as such control underpins many engineering and physical sciences systems. The core components in this are the plant (the things that are controlled), the sensors, the controller (decision maker) and the actuators and the communicator. The latter is the glue that binds those components and in effect implements the feedback loop. A lot is known about control systems with well understood control hardware, much less is known when we put low-cost, low-powered (energy harvested), wireless hardware to the controller task. Yet there is an increasing demand for autonomy in infrastructures and systems hitherto manual (for unmanning/driverless etc.) where traditional control approaches are no longer practical.

The vision of this work is to embrace the non-determinism that comes with the aforementioned computer systems to build a control system that is control-friendly in terms of robustness, safety, resilience guarantees, etc., and hopes to answer the question:

"Can we build a Trusted Control Machine? That is, a set design patterns that can implement trusted control systems for given classes of application?"

Such patterns involve everything from the translation of the physicality of the plant to the hardware, software and requirements, in effect what has become known as a cyber-physical system.