Reducing Information Exchange in Distributed Control of Multiagent Systems: A Norm-Free and Adaptive Event-Triggering Approach⋆
Deniz Kurtoglu, Tansel Yucelen, Stefan Ristevski, Jonathan A. Muse · 2022 American Control Conference (ACC) · 2022
For reducing information exchange in distributed control of multiagent systems, we propose a norm-free and adaptive event-triggering rule, which is decentralized and predicated on the solution-predictor curve method. Here, the decentralized feature implies that the proposed event-triggering rule depends on own error signals of an agent. Moreover, the norm-free feature implies that the left side of the proposed event-triggering rule inequality does not depend on distances such as absolute values of error signals to yield better agent-to-agent information exchange reduction. To achieve decentralized and norm-free features at the same time, an adaptive term is also utilized in the event-triggering rule for each agent in order to estimate unknown variable unavailable to an agent. The proposed event-triggering rule works both for the sampled data exchange case as well as for the data exchange case predicated on the solution-predictor curve method. In contrast to standard sampled data exchange, the solution-predictor curve method has the ability to further reduce agent-to-agent information exchange, where each agent stores this curve and exchanges its parameters when an event occurs in a distributed manner for approximating the solution trajectory of each agent.