Distributed Control for Cyber-Physical Systems

Networked Cyber-Physical Systems (CPS) are fundamentally constrained by the tight coupling and closedloop control and actuation of physical processes. To address actuation in such closed-loop wireless control systems there is a strong need to re-think the communication architectures and protocols for maintaining stability and performance in the presence of disturbances to the network, environment and overall system objectives. We review the current state of network control efforts for CPS and present two complementary approaches for robust, optimal and composable control over networks. We first introduce a computer systems approach with Embedded Virtual Machines (EVM), a programming abstraction where controller tasks, with their control and timing properties, are maintained across physical node boundaries. Controller functionality is decoupled from the physical substrate and is capable of runtime migration to the most competent set of physical controllers to maintain stability in the presence of changes to nodes, links and network topology. We then view the problem from a control theoretic perspective to deliver fully distributed control over networks with Wireless Control Networks (WCN). As opposed to traditional networked control schemes where the nodes simply route information to and from a dedicated controller, our approach treats the network itself as the controller. In other words, the computation of the control law is done in a fully distributed way inside the network. In this approach, at each time-step, each node updates its internal state to be a linear combination of the states of the nodes in its neighborhood. This causes the entire network to behave as a linear dynamical system, with sparsity constraints imposed by the network topology. This eliminates the need for routing between “sensor → channel → dedicated controller/estimator → channel → actuator”, allows for simple transmission scheduling, is operational on resource constrained low-power nodes and allows for composition of additional control loops and plants. We demonstrate the potential of such distributed controllers to be robust to a high degree of link failures and to maintain stability even in cases of node failures.