Model-based compositional design of networked control systems

The rapid advancement in digital technology over the past few decades has fueled the progress in computation and communication technologies. Enabled by this progress, complex engineered systems commonly referred to as Cyber-Physical Systems (CPS), resulting from the integration of computing, communications and control, and in direct interaction with the physical world, are becoming ubiquitous in our daily lives. Examples of these systems include process control, automotive systems, networked robotics, medical systems, electrical power grids and environmental monitoring systems among others. These real world CPS are increasingly being monitored and controlled by networked control systems (NCS) and are often employed in critical settings, therefore the assurance of properties such as stability, performance, safety and security are essential. As a result, the analysis and design of NCS architectures have recently gained increasing attention. This dissertation addresses several fundamental challenges in the modeling, design, analysis and evaluation of dependable networked control systems. First, a domain specific modeling language (DSML), Passive Networked Control Systems (PaNeCS), is presented. PaNeCS raises the level of abstraction of NCS design and allows automated analysis, code generation, system configuration, deployment, and testing. PaNeCS is based on passivity and ensures the “correct-by-construction” design of NCS by enforcing passivity constraints on the components of the NCS as well as their interconnections. Simulation and experimental models generated from the tool are presented to demonstrate the robustness of NCS designs using the tool. Second, an integrated passivity-based adaptive sampling control (PBASC) architecture is presented. PBASC architecture addresses the challenges due to the limited network resources as well as the presence of network uncertainties. The underlying idea of PBASC architecture is to simultaneously allow the variability of sampling intervals as well as ensure stability. Hence, in the proposed framework, while passivity ensures the robustness of the NCS in the presence of uncertainties, adaptive sampling ensures the efficient utilization of network resources. Third, an integrated modeling and simulation tool, Networked Control Systems WindTunnel (NCSWT), based on High Level Architecture (HLA), is introduced. NCSWT integrates Matlab/Simulink and ns-2 for the accurate and efficient evaluation of NCS. Finally, an energy-ba