Abstract
Wireless networked control systems (WNCS) are pivotal for the Industrial Internet of Things (IIoT), yet they face inherent tradeoffs between latency, data rate, and reliability. This paper investigates these tradeoffs using a joint framework of Control Theory and Finite-Blocklength Information Theory. We identify a "latency paradox" where stability can be maintained at high latencies given sufficient signal quality, though performance costs rise. Our analysis provides rigorous stability conditions and identifies optimal design parameters for practical IIoT applications.
Introduction
In the Industrial Internet of Things (IIoT), the transition from wired to wireless communication for sensors and actuators introduces significant challenges. Wireless channels are characterized by latency (delays), finite data rates, and imperfect reliability (packet drops). Traditional research often isolates these factors or utilizes simplified models that ignore the physical constraints of radio systems. For instance, increasing reliability typically necessitates longer codes (higher latency) or reduced data rates. This paper bridges the gap by jointly modeling these parameters to determine the stability and performance boundaries of wireless control loops.
System Modeling and Methodology
The study employs a cross-disciplinary approach, merging Control Theory with Finite-Blocklength Information Theory to capture real-world communication constraints.
System Architecture
The researchers model a scalar linear time-invariant (LTI) plant. A remote controller transmits quantized and encoded commands to an actuator over an Additive White Gaussian Noise (AWGN) channel.
Analytical Framework
- Finite-Blocklength Theory: Unlike asymptotic models that assume infinite-length codes, this framework captures the hard tradeoffs between blocklength ($n$), data rate ($R$), and error probability ($\epsilon$).
- Zooming Quantizer: A value-independent adaptive quantizer is implemented. It expands its range during communication failures to capture drifting states and shrinks upon success to enhance precision.
- Performance Metrics: Rigorous upper and lower bounds for the "average cost function" (mean-square error of the plant state) were derived to quantify system performance.
As a theoretical study, the methodology relies on mathematical proofs and numerical simulations with synthetic parameters (SNR: 3–30 dB; plant instability factor $a$: 1.01–2.1) rather than external datasets.
Key Insights: The Latency Paradox
A primary finding is the "latency paradox": a plant can technically remain stable with an arbitrarily large blocklength (latency), provided the channel's Signal-to-Noise Ratio (SNR) exceeds a specific threshold. However, stability does not equate to quality. While the system may not "blow up" at high latencies, the average cost—representing the deviation from the target state—grows unbounded. This underscores that low latency is a requirement for operational performance rather than just basic stability in high-SNR environments.
Results and Stability Analysis
The authors derived a necessary and sufficient condition for mean-square stability: $a^{2n} + (1-\epsilon)\frac{a^{2n}}{2^{2nR}} < 1$. If this inequality holds, the system is stable.
Numerical analysis for a plant with $a=2.1$ revealed that at an SNR of 10 dB, the stability region is strictly bounded by latency. Conversely, at 11 dB and 20 dB, the region becomes "n-unbounded," allowing for stability at any blocklength. Optimization of the cost-bound analysis identified "sweet spots" for engineering design; for example, a blocklength of $n=120$ and a data rate of $R=0.7$ minimized the average cost in a specific test scenario.
Conclusion and Future Work
While this work provides a robust foundation for scalar systems, real-world industrial applications often involve multi-variable (vector) systems with higher complexity. Furthermore, the current model assumes static AWGN channels; future research must incorporate fading channels where signal strength fluctuates. Finally, integrating energy constraints into a Pareto-optimal design framework remains a critical step for battery-powered IoT deployments.
Summary updated