15 July 2024 in IEEE Transactions on Automatic Control
We present a framework for the design of time- and event-triggered communication schemes for a broad class of multi-agent systems (MAS). The framework is general in the sense that we consider a class of heterogeneous nonlinear MAS with directed communication graphs subject to disturbances and the inevitable imperfections induced by packet-based networked communication, such as time-varying transmission delays and intervals. Under suitable conditions, the designed distributed time- or event-triggered communication schemes lead to a MAS that is dissipative with respect to a desired supply rate, and thus can be used to study various properties such as $\mathcal {L}_{p}$ -gain stability, input-to-state stability and passivity. Strictly positive lower bounds on the inter-event times are ensured and robustness to non-uniform time-varying delays in terms of maximum allowable delays is guaranteed. In addition, the framework allows for the consideration of destination protocols which determine, at each transmission instant, to which (groups of) agent(s) a local output measurement is being transmitted. These destination protocols can be used to include packet loss and denial-of-service, as will be shown in the companion paper [1].
17 June 2024 in IEEE Control Systems Letters
In this paper, we present a unified general non-strict Finsler lemma. This result is general in the sense that it does not impose any restrictions on the involved matrices and, thereby, it encompasses all existing non-strict versions of Finsler’s lemma that do impose such restrictions. To further illustrate its usefulness, we showcase applications of the non-strict Finsler’s lemma in deriving a structured solution to a special case of the non-strict projection lemma, and we use the unified non-strict Finsler’s lemma to prove a more general version of the matrix Finsler’s lemma.
15 July 2024 in IEEE Transactions on Automatic Control
We consider the design of event-triggered distributed controllers for multi-agent systems that are digitally implemented on local computation platforms and communicate over a packet-based network. Each agent is equipped with a local triggering mechanism that is only evaluated at the local sampling instants, thereby taking a periodic event-triggered approach in which the sampling intervals are allowed to vary (jitter). Moreover, the locally triggered transmissions are subject to unknown, bounded delays, and a destination protocol is locally implemented to only send the packet to a selection of the neighboring agents at each triggering instant. Building upon the framework of Part I, we present an emulation-based design of the local periodic event-triggering rules, including the maximum allowable sampling period (MASP), so that, under appropriate conditions, a general dissipativity property holds for the overall system. Interestingly, the presented digital implementation requires only minor modifications to the conditions presented in Part I. Additionally, we show how to exploit the destination protocols to ensure resilience to information loss issues such as packet losses and denial-of-service. We conclude this paper with case studies on the consensus of single integrator agents and a nonlinear stabilization problem.
1 January 2026 in IEEE Transactions on Automatic Control