Stability of Linear Systems Under Extended Weakly-Hard Constraints

Nils Vreman, Paolo Pazzaglia, Victor Magron, Jie Wang, Martina MaggioIEEE Control. Syst. Lett 2022 · OpenReview

TL;DR

This paper introduces a novel framework for analyzing the stability of linear control systems implemented on embedded platforms that may experience occasional deadline misses. By extending the "weakly-hard" real-time task model to explicitly account for different strategies of handling these misses, the authors provide a more realistic and generalizable approach to ensuring system reliability. This work bridges the gap between control theory and real-time systems engineering, offering a powerful tool for designing robust embedded controllers.

What Problem Does This Paper Address?

Modern control systems are increasingly deployed on resource-constrained embedded platforms. While these systems are designed to be robust to external disturbances and model inaccuracies, a critical challenge arises from computational limitations. Specifically, control tasks, implemented as periodic software routines, may not always complete within their designated deadlines. This phenomenon, known as a "deadline miss," can occur due to various factors like unexpected computational overloads or system interrupts.

Traditional control theory often assumes perfect execution, while real-time systems analysis has developed models like the "weakly-hard" model to handle sporadic deadline misses. However, existing weakly-hard models often make simplifying assumptions about how deadline misses are handled (e.g., discarding the missed computation) and may not be general enough to cover diverse real-world scenarios. This paper addresses this gap by proposing a comprehensive stability analysis that explicitly incorporates different deadline miss handling strategies (like "kill" or "skip-next") and their impact on the control system's stability. This allows for a more accurate assessment of robustness in practical embedded control applications.

Methods

The core contribution of this paper lies in extending the weakly-hard real-time task model to explicitly incorporate deadline miss handling strategies. This is achieved through several key steps:

  • Extended Weakly-Hard Model: The authors define an "extended weakly-hard model" that considers not only the pattern of deadline hits and misses but also the specific strategy employed when a miss occurs. This involves defining new alphabets of task outcomes (e.g., "hit," "miss," "recovery") depending on the handling strategy.
  • Automaton Representation: Each extended weakly-hard constraint is systematically represented using a finite automaton. These automata capture all feasible sequences of task outcomes that satisfy the given constraints.
  • Kronecker Lifting: To analyze the stability of the resulting switching system (where the system dynamics change based on the task outcomes), the authors employ a technique called Kronecker lifting. This method transforms the system dynamics and the automaton transitions into a single, larger system with a set of matrices that can be analyzed using arbitrary switching system techniques.
  • Joint Spectral Radius (JSR): The stability of the lifted system is then determined by computing its Joint Spectral Radius (JSR). The JSR is a powerful tool for analyzing the stability of systems that switch between different dynamics. The paper leverages existing JSR computation tools and introduces a sparse variant for efficiency.
  • Constraint Domination: The paper also formalizes the concept of "constraint domination," showing how stricter weakly-hard constraints lead to more stable systems, and how this translates to JSR properties.

Datasets

Not discussed in the paper. The evaluation uses a representative process industry plant controlled by a PI-controller.

Key Insights and Ideas

  • Handling Strategy Matters: The paper highlights that the specific strategy used to handle deadline misses (e.g., discarding the computation vs. delaying it) has a significant impact on the overall system stability. This is a crucial insight for practical controller design.
  • Unified Framework: By extending the weakly-hard model and using Kronecker lifting, the authors provide a unified framework that can analyze a broad range of weakly-hard constraints and handling strategies. This simplifies the analysis process and makes it more generalizable.
  • Bridging Control and Real-Time: The work effectively bridges the gap between control theory (stability analysis) and real-time systems engineering (deadline miss analysis). This allows control engineers to better understand the implications of real-time execution constraints on their designs.
  • Computational Robustness: The analysis directly addresses the robustness of control systems to computational imperfections, moving beyond traditional robustness to external disturbances.
  • Efficiency through Sparsity: The use of sparse SOS (Sum of Squares) relaxations for JSR computation significantly speeds up the analysis, especially for systems with sparse matrix structures, making the approach more practical for complex scenarios.

Key Results

The paper presents a theoretical framework for analyzing the stability of linear control systems under extended weakly-hard constraints. The key results are:

  • Formalization of Extended Weakly-Hard Model: A generalized model that accounts for various deadline miss handling strategies.
  • Analytic Stability Bounds: The framework provides a method to derive analytic bounds on the stability of the closed-loop system, expressed through the Joint Spectral Radius (JSR).
  • Strategy-Independent Analysis: The core analysis is strategy-independent, meaning the same framework can be applied to different handling strategies.
  • Empirical Validation: An evaluation on a representative process industry plant demonstrates the effectiveness of the approach. The results show that for certain constraints, the system is stable under various strategies and actuator modes. Notably, the sparse SOS method for JSR computation offers a significant speedup compared to the dense method. For example, in Table I, speedup factors ("x") for the sparse method over the dense method can be as high as 77.8.

Limitations and Future Work

  • Linear Systems: The current analysis is limited to linear time-invariant (LTI) systems. Extending the framework to nonlinear systems would be a significant advancement.
  • Performance Analysis: While stability is addressed, the paper focuses on stability and does not delve deeply into performance degradation due to deadline misses. Future work could explore quantifying performance loss.
  • Computational Complexity: While sparsity is leveraged for efficiency, computing JSR can still be computationally intensive for very large systems or complex constraint sets.
  • Future Directions: The authors suggest future work on performance loss analysis and extending the framework to nonlinear systems. They also mention that the current SDP solver in SparseJSR might be improved with a different solver for memory efficiency.

Summary updated

Abstract

Control systems can show robustness to many events, like disturbances and model inaccuracies. It is natural to speculate that they are also robust to sporadic deadline misses when implemented as digital tasks on an embedded platform. This letter proposes a comprehensive stability analysis for control systems subject to deadline misses, leveraging a new formulation to describe the patterns experienced by the control task under different handling strategies. Such analysis brings the assessment of control systems robustness to computational problems one step closer to the controller implementation.
BibTeX
@inproceedings{vreman2022stability,
  author = {Nils Vreman and Paolo Pazzaglia and Victor Magron and Jie Wang and Martina Maggio},
  title = {{Stability of Linear Systems Under Extended Weakly-Hard Constraints}},
  booktitle = {IEEE Control. Syst. Lett},
  year = {2022},
  url = {https://openreview.net/forum?id=QpGHX8cx19},
}
Figure