Channel Switching Algorithms for a Robust Networked Control System with a Delay and Packet Errors

Abstract

Redundancies in modern systems, including multiple channels, processes, and storages, are often exploited to ensure robust operation. Similarly, a Networked Control System (NCS) may utilize multiple channels to facilitate reliable information transfer in case of channel failure. To enhance the performance of Linear Quadratic Gaussian (LQG) control in environments with multiple channels, delays, and packet errors, we propose channel-switching algorithms. Leveraging the encoder and decoder structure for channel modeling, we derive the decoder estimation error covariance matrix, characterizing LQG control performance with respect to delay. Based on this insight, we develop two threshold-based channel-switching algorithms, proven to ensure finite total decoder estimation error variance under certain conditions. Specific conditions are also identified where the proposed algorithms offer improved probabilistic stability. Numerical simulations confirm the superior performance of the proposed algorithms compared to conventional methods across diverse channel environments. Notably, the proposed algorithms demonstrate near-optimal performance in a practical operational scenario involving multiple channels, specifically 5G cellular link and Starlink.