Convergence Rate Abstractions for Weakly-Hard Real-Time Control

Abstract: Classically, a control loop is designed to be executed strictly periodically. This is, however, difficult to achieve in many scenarios, for example, when overload or packet loss cannot be entirely avoided. Here, weakly-hard real-time control systems are a common approach which relaxes timing constraints and leverages the inherent robustness of controllers. Yet, their analysis is often hampered by the complexity arising from the system dimension and the vast number of possible timing sequences. In this paper, we present the novel concept of convergence rate abstractions that provide a sound yet simple one-dimensional system description. This approach simplifies the stability analysis of weakly-hard real-time control systems. At the same time, our abstractions facilitate efficient computation of bounds on the worst-case system state at run-time and thus the implementation of adaptation mechanisms.