This talk will provide a review of the state of the art on distributed control on one side, and on security and resiliency on the other. The main goal is to provide the participants with the fundamental concepts and definitions that will later be reprised by other talks, ultimately leading to present the big picture underlying the workshop. We will start by reviewing classical approaches to decentralized and distributed control, following in the footsteps of the works by D.D. Siljak, moving then to modern techniques such as consensus-based, leader-follower and coalitional control. The aspect of security will be addressed by initially presenting the well-known Confidentiality-Integrity-Availability paradigm which is central to Information Technology security, and then declining it to the special case of Operation Technology, which is of interest for the control systems community. In particular the role of attacker models and of confidentiality-preserving techniques such as privacy and encryption will be highlighted. Finally the concept of resiliency for autonomous, distributed systems will be introduced, linking it to well-established literature in the field of fault-tolerant control.

The transition to sustainable energy is one of the major challenges of the 21st century, with offshore wind playing a critical role due to the reliability and energy density of sea-based wind. However, the remote and harsh operating conditions expose them to significant risks, particularly cyberattacks aiming to disrupt energy production or shorten their operational lifetime. This talk highlights the security challenges that offshore wind farms face, and presents ongoing efforts in the EU Horizon projects SUDOCO and TWAIN, which develop control-oriented approaches to enhance resilience. We will conclude with recent advancements and current research developed within the projects.

Networked dynamical systems, such as infrastructure networks, supply chains, biological networks, social networks, and so on, are central to our lives. Yet, they often fail catastrophically when faced with large disturbances arising from extreme events. A wide range of tools from systems theory, such as adaptive control, as well as from AI/machine learning, have been developed to guarantee robustness in these settings. However, they do not scale well, lack guarantees on resulting solutions, and cannot deal with large disturbances that push systems into far-from-equilibrium regimes in which they are not typically designed to operate. In this context, this talk will address the problem of ensuring operational resilience of networked dynamical systems to extreme events/shocks. The key idea is that algorithms that capture and utilize control-relevant physical properties or constraints — such as dissipativity, monotonicity, conservation laws, or symmetries — can lead to scalable designs that can adapt to large disturbances and achieve operational resilience. Specifically, we will discuss (i) data-driven approaches to learn models of these systems capturing control-relevant properties like dissipativity in nonlinear and non-equilibrium settings, and (ii) scalable and compositional learning-based control designs that leverage these properties to provably guarantee safety and operational resilience.

Recent technological advances have enabled the deployment of learning-based multi-agent systems. Many of the potential applications, however, have practical constraints related to limited and uncertain computational power, battery resources, connectivity, and communication bandwidth, not addressed by classical learning approaches. In this talk, we will discuss how to design multi-agent learning algorithms based on Alternating Direction Method of Multipliers and other techniques in the presence of such practical constraints. We will present novel mathematical tools based on open networks and show how convergence analysis and guaranteed resilience can be established. Illustrative numerical results will be described together with directions for future research. The presentation is based on joint work with Nicola Bastianello, Diego Deplano, and Mauro Franceschelli.

Distributed optimization is a fundamental tool for multi-agent systems, allowing them to cooperatively perform tasks such as learning, coordination, exploration. Solving distributed optimization problems has thus received wide attention in the literature, with gradient-based methods emerging as the state of the art. These algorithms integrate the agents’ local gradient computations with peer-to-peer communications that implement cooperation. However, in practical scenarios some of the agents might be malicious, injecting arbitrary communications with the goal of degrading the performance of the distributed algorithm. Therefore, we focus on designing detection techniques, deployed by the agents directly, which can locate malicious agents and mitigate their impact. In particular, we propose a security metric which quantifies the maximum performance degradation caused by attackers, as a function of the monitoring agents’ locations. We discuss how to apply this approach to Distributed Gradient Descent (DGD) and Gradient Tracking (GT) algorithms, corroborating our theoretical analysis with numerical experiments.

Multi-robot systems are increasingly integrated into real-world applications, from autonomous vehicle fleets to search-and-rescue teams. Ensuring their coordination algorithms remain robust against unreliable communication, security threats, and corrupted data is essential. In this talk we will discuss multiagent consensus as a core primitive to many coordination tasks. We will study the Friedkin Johnson dynamic as a method for dealing with unreliable agents or data in the system. We will discuss recent advances in understanding the role of a time-varying competition parameter, along with comparisons to distributed optimization in multiagent systems.

Sophisticated distributed control schemes and the growing interconnectivity of control systems pave the way for exciting new opportunities. At the same time, connected systems are susceptible to cyberattacks and data leakages. Against this background, encrypted control aims to increase the security and safety of cyber-physical systems. A central goal is to ensure the confidentiality of process data during networked controller evaluations, which is enabled by, e.g., homomorphic encryption (HE) or secure multi-party computation (SMPC). However, the integration of advanced cryptographic systems renders the design of encrypted controllers an interdisciplinary challenge and profound knowledge of suitable cryptosystems is crucial. The talk aims to facilitate the access to encrypted control for multi-agent systems by providing illustrative insights on pioneering realizations as well as modern implementations.

In this talk, we provide an overview on the recent advances in the research of multi-agent systems operating in hostile environments. We will focus on the influence of misbehaving agents in a network capable to inject false data in their transmissions and how to mitigate such attacks by approaches based on fault-tolerant distributed algorithms from computer science for Byzantine consensus. Agents equipped with such algorithms will seek safe state values  and ignore their neighbors taking extreme state values.  We will see that characterizations on the properties necessary for network topologies have been established, and moreover that network resiliency can be enhanced when more communication and computational resources are available. In particular, we will discuss methods based on centerpoint computation and multi-hop relaying in communication.

We will analyze the cybersecurity challenges and solutions applicable to distributed renewable energy generation systems, interconnected through communication networks that enable the monitoring, supervision, and centralized control of geographically dispersed generating units. The increasing integration of typical Industrial Control System (ICS) components—such as SCADA, DCS, and PLCs—into critical energy infrastructures expands the cyberattack surface and requires the adoption of specific practices and standards to ensure the principles of confidentiality, integrity, and availability. In this context, we explore key regulatory references and international frameworks, such as the ISA/IEC 62443 series, NIST SP 800-82, NISTIR 7628, and the Distributed Energy Resource Cybersecurity Framework (DER-CF), which guide the implementation of technical and organizational measures for the protection of assets in energy automation and control environments. Recent initiatives are also highlighted, promoting the concept of ``secure by design'' and the secure update of critical components such as inverters and microgrid controllers.