Stability analysis of networked control systems using three-layer delay modelling
Networked Control Systems (NCS) play a vital role in industrial process automation by enabling communication among sensors, controllers, and actuators through shared communication networks. However, delays caused by field devices and multi-layer protocol stacks significantly influence closed-loop stability and transient performance. This paper investigates the dynamic effects of delay in a three-layer network structure within industrial process control. A systematic delay assessment model is developed to evaluate the significance of delay within each layer and to include these delays in closed-loop stability analysis. Unlike existing lumped-delay models, the proposed methodology explicitly decouples deterministic intra-device processing delays from multi-layer protocol communication delays. The paper derives the exact Maximum Allowable Communication Delay (MACD) using frequency-domain Phase Margin criteria and validates closed-loop stability margins. Furthermore, an industrial layer-aware Smith Predictor architecture is developed to compensate for network transport delays, demonstrating that while communication latencies can be effectively eliminated from the closed-loop characteristic equation, uncompensated internal device execution delays establish an ultimate physical performance boundary. Benchmark simulations across Open-Loop, Closed-Loop, and Cascade-Loop configurations validate that the proposed co-design framework enhances stability margins, mitigates phase erosion, and minimizes transient lag in industrial automation.
Authors
- Kannan Srinivasan (ORCID: https://orcid.org/0000-0003-3626-1723)
- Guduru Jahnavi Chowdary
- S. Priyanka
Institutions
- VIT-AP University (IN)
- Vellore Institute of Technology University (IN)
Publication Details
- Journal
- Scientific Reports
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1038/s41598-026-80538-6
- Primary Topic
- Stability and Control of Uncertain Systems
- Type
- article
- Field-Weighted Citation Impact
- 0.00