Resource efficient dynamic agreement and replication

• Applicant: Professor Dr.-Ing. Rüdiger Kapitza
• Subject Area: Security and Dependability, Operating-, Communication- and Distributed Systems
• Term: from 2015 to 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 206480214

There is an ongoing trend towards cyber-physical systems, which raises many questions and concerns prominently including safety and security matters. The research unit Controlling Concurrent Change (CCC) aims at answering a specific subset of these open issues, namely how the flexibility of software and its opportunity for reconfiguration and updates can be utilized, while still preserving essential mission-critical non-functional properties. The subproject A3 "Resource efficient dynamic agreement and replication" addresses the challenge to provide a basis for joint and reliable reconfiguration of CCC-controlled domains and offers applications that are dispersed over multiple CCC-domains support for fault-tolerant coordination and replication. Thereby, faults are not limited to crashes but depending on the scenario range from hardware induced malfunctioning (e.g. bit flips) up to the extreme of intentional and coordinated malicious behavior. At the technical level the A3 subproject aims at providing novel protocols for resource efficient Byzantine agreement that are tailored towards resource constrained embedded systems. Depending on the use case wireless and mobile settings are targeted. While the first poses the challenge of being less reliable and more constrained compared to the typical deployment environment of agreement protocols the intended smart use of multi-homing might mitigate these effects. Mobile settings furthermore typically demand to address frequent membership changes, which poses a challenge for existing Byzantine agreement protocols and so far has been neglected. As one of the core parts of the CCC infrastructure, the online validation of an update requires for all non-trivial cases a resource-intensive planning process. Naively replication of this functionality for coordinated reconfiguration of multiple domains would waste vast amounts of resources. Accordingly tailored replication schemes for this core task will be devised and evaluated. The latter will be performed in combination with the developed agreement support in scope of simulated environments as well as a physical deployment consisting of a set of embedded systems.

• DFG Programme: Research Units
• Subproject of: FOR 1800: Controlling Concurrent Change (CCC)