Study on the Performance of an Event-Based Cloud Multiple NMPC
Alvin Immanuel Surjana, Elmar Ahle, Dirk Söffker · 2025
Interest in cloud control systems (CCS) has increased since their introduction, notably in the context of industrial cyber-physical systems due to decentralized control strategies. A key factor is the removal of restrictions on local computing units, as the computational implementation of the control system is shifted to the (theoretically unlimited) cloud. This expands the computing capacity of the control system to carry out complex optimization processes such as nonlinear model predictive control (NMPC). Although the use of cloud computing can reduce the computation time, the computation delay can still be longer than one time step. The computation delay and communication latency due to the decoupling of components in the control loop should be compensated to take full advantage of cloud computing resources. In a previous study, an event-based cloud NMPC algorithm with time-delay compensation was developed for compensation. In the event-based strategy, the optimal input sequence is used until a new sequence is available. In this paper, the event-based cloud NMPC framework is extended with the ability to integrate multiple identical controller instances into the control loop by using a centralized scheduling strategy. In addition, a special rule is introduced in which the first computed optimal sequence without compensation of the delay time is used.