Optimization Congestion Control for Networked Control Systems

Technological advances in embedded systems and communication networks have given birth to devices with sensing, processing, actuating, and communication capabilities. When these sensing, processing, and actuating units are networked together, they can be used to monitor and control a remote physical environment. Sensors gather data and forward it to processing units, called controllers. Controllers process data and send decisions to actuators. Networked control finds applications in industrial automation, distributed instrumentations, unmanned vehicles, and home robotics. In accord to [1] and all the resources contained therein, when these sensors, controllers, and actuators exist in a distributed fashion and communicate over IP networks, they are referred to by the term Networked Control Systems (or NCSs), see Fig. 1. NCSs have several challenges, which are due to the nondeterministic nature of communication networks. For example, communication delays, delay jitter, and packet losses adversely affect the performance of the system, and may even cause instability. Related research in this field has emerged into two distinct directions. The first direction aims at a control theoretical analysis while considering the network as a fixed parameter. The second direction targets the design of new communication network infrastructures, algorithms or protocols. Examples of the first approach include designing special controllers, and varying the sampling rate; and examples of the second include designing static and dynamic message scheduling algorithms. In this document, we present our approach that combines both directions and relies on the well-established results in both communication networks and control theory. We informally state our objective as simple as: