CAREER: Timely, Efficient, and Risk-Aware Control and Communication Policies for Networked Multi-Agent Systems

About This Grant

Networked Multi-agent Systems (NMS), such as fleets of drones, connected autonomous vehicles, or smart power-grids, consist of multiple plants, controllers, and sensors exchanging data over a shared communication network managed by a network manager. This network manager allocates communication services (CS), such as bandwidth, reliability, and latency, to each agent, enabling sensors to transmit data to their respective controllers. For optimal NMS performance, the problem must be jointly studied at the agent level and the network level. Agents need to design communication-aware controllers, while the network manager must allocate communication services in a control-aware manner. Specifically, agents must develop controllers that proactively incorporate allocated communication services, analyzing their impact on sensor data quality, timing, and resolution. Simultaneously, the network manager must dynamically allocate communication resources to meet agents' evolving needs while ensuring fairness in allocations across all agents. At its core, this problem requires developing an optimal control-communication theory to guide decision-making for both agents and the network manager. The proposal envisions enabling optimal decision-making for NMS across various domains: from enhancing coordination and cooperation in multi-robot systems to optimizing information exchange among connected and autonomous vehicles to voltage and frequency control in power-grids. With advancements in communication and networking technologies, communication-aware controllers are especially important for real-world applications, offering improved performance, reliability, and resource utilization. The educational objectives of this proposal aim to ignite interest in NMS and control systems in general, while the outreach activities are designed to inspire K-12 students to pursue STEM education. This proposal systematically investigates NMS operations at both agent and network levels. The first contribution is a principled control-communication theory that addresses the challenges of designing CS-aware controllers and control-aware CS allocations. This theory enables tradeoff and sensitivity analyses within the CS parameter space, answering questions such as whether one parameter (e.g., latency) can compensate for another (e.g., bandwidth). It provides insights into which parameter has the most significant impact under varying conditions, allowing the network manager to allocate resources effectively. The research extends beyond agent and network levels by examining environmental impacts on CS allocations. A risk-aware controller synthesis framework is proposed to address uncertainties that traditional robust control techniques cannot adequately handle. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.