Device Collaboration for Stability Assurance in Distributed Cyber-Physical Systems

Abstract

Distributed Cyber-Physical Systems (DCPS) are special control systems because of involving distributed systems characteristics. To enable reliable DCPS, stability assurance is of utmost importance. But due to the system's distributed nature, network delay is inevitable and can affect stability adversely. Existing work to assure stability in DCPS has to rely on either a fixed and accurate model of network delay, or the scheduling of messages in the network. However, in reality, it is difficult to obtain an accurate network delay model to support the former approach, because the network can exhibit very complex behaviors. The latter approach also suffers from a notable problem, that is, message scheduling is often not allowed or supported by the network. In this paper, we propose a novel approach to overcome these drawbacks. We augment DCPS devices, including sensor, actuator and controller, with certain distributed intelligence. Thus, they can collaborate to understand the characteristics of network delay at runtime, and then adapt their behaviors accordingly to achieve stability. In this way, we avoid the reliance on network delay modeling and message scheduling, and also make DCPS adaptive to the dynamic network environment. Furthermore, we conduct theoretical analysis, and derive some stability criteria to guide the distributed collaboration and adaptation. The effectiveness of our approach has been validated in a simulated green building application.