A Logical Framework for Self-Optimizing Networked Cyber-Physical Systems

The objective of this research is to develop foundations for the newly emerging generation of Networked Cyber-Physical Systems (NCPS). In par- ticular, the project is investigating declarative approaches to distributed control of NCPS. Networked cyber-physical systems present many intellectual challenges not suitably addressed by existing computing paradigms. They must achieve system-wide objectives through local, asynchronous ac- tions, using distributed control loops through the environment. A key challenge is to develop a robust computational foundation that supports a wide spectrum of system operation between autonomy and cooperation to adapt to uncertainties, changes, failures, and resource constraints, in particular to limitations of computational, energy, and networking resources.

As a distributed computing foundation we have developped a model that is based on partially-ordered knowledge sharing and avoids any transactional semantics that could limit its implementability and scalability. Based on this loosely-coupled partially-ordered knowledge sharing model we have developed a sound and complete distributed logic for NCPS. Its novelity is that unlike traditional logics it can interact with the environment during the synthesis of a proof and the notion of proof is inherently distributed and cooperative. In parallel, we have investigated distributed optimization and control on top of the same foundation and recently developed a new executable model for workflows of cyber-physical ensembles. Similar to the distributed logic, we are using optimization strategies to enable local actions to maintain or improve the satisfaction of system goals. The framework has been implemented and demonstrated in the context of networked mobile robotic teams by interfacing it the the Player-Stage multi-robot simulator. The framework also supports deployment on real as opposed to simulated mobile ad hoc networks. Our target platform for the next set of experiments is a networked quadcopter testbed based on modified Parrot AR Drones that we currently building at SRI International.

The results of this project have a variety of applications including dis- tributed surveillance, instrumented pervasive spaces, crisis response, medical systems, green buildings, self-assembling structures, networked space/satellite missions, and distributed critical infrastructure monitoring and control. All papers and developments of this project with tutorial-style examples are publicly available from our project web site http://ncps.csl.sri.com.

Award ID: 0932397