Dynamic Rate and Control Adaptation in Networked Control Systems

Modern embedded control systems are asked to exhibit higher distribution, higher degree of integration and to support operational flexibility. Conventionally the planning of distributed control systems was done either by considering worst-case requirements, leading to expensive and inefficient designs, or by considering average requirements, potentially leading to occasional run-time overloads. The dynamic rate adaptation technique was developed to support operational flexibility and a higher degree of integration in distributed control systems. This technique adapts the communication requirements of distributed feedback control loops by increasing the sampling periods of the messages associated to the control action in order to free network bandwidth when a possible network overload condition is detected. The distributed controllers have to be able to deal with the change of the sampling period and the variable sampling to actuation delay. This work reports different control approaches used in the implementation of the dynamic rate adaptation method, namely a static version based on a set of controllers with parameters pre-computed off-line, a dynamic version where the system parameters are identified online according to the dynamic behaviour of the distributed system and a hybrid version using an adaptive controller with a set of online identifiers.