Control of Cyber-Physical Systems

An important paradigm in the design of modern engineering systems are the so-called "Cyber-Physical Systems" (CPS). CPS are complex, heterogeneous, spatially distributed systems where physical processes interact with distributed computing units through non-ideal communication networks. Our research includes the following three topics:

1. A key feature of CPS is the heterogeneity of their subsystems: while physical processes are usually modeled by differential equations, hardware and software in the computing units are modeled by finite state machines. Symbolic models are general enough to describe each component of CPS. A symbolic model is a finite state machine in which an aggregate of continuous states corresponds to a symbolic state. Using symbolic models, it is possible to solve in a systematic way control problems in which hardware and software interact with physical processes through non-ideal communication networks.

2. Wireless communication networks for control systems are of great interest because they offer advantages such as reduction of cost of installation, flexibility, robustness to failures, ease of maintenance and diagnostics. The formal analysis of these networks has the power to quantify robustness and performance of the closed loop system, allowing the use of methods for the design of control systems that are less conservative. The leit motiv of this research line is the definition of a mathematical model that formalizes the interconnection of a wireless network with an embedded control system, and the development of systematic methods for the design and testing of control systems on wireless networks.

3. In sensor-actuator networks it is important to establish existence, convergence and robustness of decentralized algorithms able to replace centralized methodology for the analysis and control of dynamic systems. The framework of the study is the distribution of control tasks among agents, none of which being able in isolation to satisfy the design requirements of the overall  system. These agents can be connected by a communication network allowing them to act in a distributed and coopearative fashion. Once a general model is formalized, the research concerns the selection of parameters that optimize the performance of the system within constraints imposed by the network and the computing units.

These research areas are applied to domains of interest such as air traffic management systems, traffic control, smart buildings, automotive.