Typical cyber physical systems (CPSs), such as smart grid, unmanned aerial vehicles (UAVs) and robotic networks, consist of physical dynamics, sensors, communication network and controllers. The Communication network, which conveys system measurements from sensors to controllers, plays a key role in CPSs, similarly to the nerve system in human beings. Traditional data communication networks (e.g., cellular networks or WiFi) focus only the delivery of data packets, while the ultimate goal of a CPS is to control the physical dynamics (e.g., stabilizing the voltages and frequencies in power networks). Hence, the traditional design of communication network may not be optimal in the context of CPSs, due to the mismatched goals of designs. This results in a pressing need to study the design of communication networks in CPSs, which helps to enhance the agility, robustness and efficiency of CPSs. This project studies how to efficiently design system-dynamics-aware communication networks for CPS, which integrates the areas of communications, networking, control, and dynamical systems, and has applications to the important and growing field of CPS in critical infrastructures.
The project specifically addresses the following research tasks: (a) Joint Design as Hybrid Systems: The theory of hybrid systems is used to model CPS, in which the operation mode of communication network is modeled as the discrete state of a hybrid system, while the physical dynamics are modeled using the continuous state. The communication and control sub-systems are designed jointly by optimizing the hybrid system dynamics; (b) Separate Design via Information Interface: The communication and control sub-systems are designed separately and are bridged via designated interfaces, such as communication quality of service (QoS) or virtual queue mapping; (c) Coexistence with Elastic Data Traffics: The realtime data traffic may share the same communication resource with elastic data traffics such as Internet data. The queuing dynamics of elastic data traffic and the physical dynamics of CPS are integrated in the same framework, and the scheduling for the two types of traffics (elastic and realtime) is studied; (d) Application and Implementation in Smart Grid: The principles, algorithms, and protocols obtained in the previous two tasks are applied and substantiated in smart grids as a case study for CPS. The voltage control in microgrid, both centralized and decentralized, is particularly studied. They are implemented in both software simulation testbed, in which the communication and control sub-systems are co-simulated, and hardware testbed using USRP boards and the microgrid testbed.
