The industrial internet of things (IIoT) is leading a new industrial revolution, powered by the emerging technologies of wireless sensor-actuator networks, edge computing, and machine learning. In traditional industrial control, plant control and network management operate in isolation, ignoring the significant dynamic interactions between networks and physical plants. In sharp contrast, a cyber-physical systems (CPS) approach closely integrates cyber (i.e., computation and communication) and physical processes. To enhance IoT-driven plant control performance, and network efficiency, and to improve the resiliency of the system to cyber and physical interferences, we propose holistic control -- a CPS approach. The holistic control framework closes the loop of plant control, networks, and computation at run-time.

In this talk, I will present three aspects of my recent work on CPS holistic control: holistic wireless control, holistic edge control, and hybrid simulation tools for holistic control. First, I will describe the design and evaluation of holistic control that exploits a range of wireless network reconfiguration strategies, and will introduce novel network reconfiguration protocols tailored for holistic control. Next, I will explore holistic edge control for multi-tier CPS, using a learning-based approach to illustrate the trade-offs between computation and communication in terms of control performance. Finally, I will discuss hybrid simulation tools for holistic control. These real-time wireless cyber-physical simulators integrate real wireless sensor-actuator networks (WCPS-RT), edge computing platforms (WCPS-EC), and simulated physical plants.