Dynamic Wireless Network Reconfiguration for Control System applied to a Nuclear Reactor Case Study

Control systems using sensors and wireless networks are becoming more prevalent, due to its ease of deployment (no wires and no electricity) and maintenance (longer lifetime). However, interference and noise can cause delays and packet losses that can degrade control system performance, which leads us to find the optimal network configuration to minimize that impact. Another problem of wireless networks for control systems is caused by time-varying fault patterns, which motivates us to do network reconfiguration at runtime. We focus on the online wireless network reconfiguration in an embedded networked and cyber-physical system (CPS), with four main contributions: (1) the design and implementation of a novel network reconfiguration framework with offline and online components that consider time-correlated link failures; (2) a network imperfection model; (3) six online reconfiguration algorithms for wireless control system; (4) a case study with 12 hops and up to 50 nodes that controls a nonlinear small modular nuclear reactor. Our network imperfection model is accurate (with 0.993 Pearson correlation) and our online reconfiguration algorithms have smaller error and longer network lifetime than a state-of-the-art static scheme.