Complex autonomous real-time and embedded systems (RTES), e.g. self-driving cars, will increasingly operate unattended over extended periods of time. The limitations of continuous supervision given their real-time nature, and the threat vectors originating from combined cyber and physical attacks, make these systems vulnerable and call for improved protection, such as tolerance to intrusions, in particular in face of time-domain attacks, to achieve safety and security in such critical systems.In this project, we investigate the interplay of intrusion tolerance techniques, such as Byzantine fault-tolerant consensus (BFT), and real-time communication and processing paradigms, e.g. event- and time-triggered abstractions. To date, the examples of research in this area are confined to very specialised and constrained RTES (e.g. flight control), thus not generalizable, and providing limited safety-security balance. Hence the hardness of the problem of reconciling security paradigms, of asynchronous (non-timed) nature, with real-time paradigms, of non-secure nature, to achieve secure and timely operation and cooperation of autonomous systems, over general networks.ByzRT will bridge this gap, by investigating on two essential fronts:-devising innovative BFT algorithms exhibiting real-time capability for single- and inter-RTES operation; and-extending formal methods for proving such protocols correct, by incorporating notions of time.If successful, ByzRT will significantly contribute to the design of safer and more secure critical cyber-physical systems.