Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Architectural Transformations and Emerging Verification Demands in AI-Enabled Cyber-Physical Systems

Hadiza Umar Yusuf, Khouloud Gaaloul

TL;DR

The paper investigates how AI integration reshapes the architecture and verification of CPS by comparing AI-driven DRL controllers against traditional MPC/PID controllers implemented in Simulink. It introduces a multi-method framework with four phases to collect models, analyze structural composition, examine dynamic flow, and assess verification implications using S-TaLiRo falsification across eight benchmark systems. Key findings include a shift from continuous to discrete, logic-driven components, higher block/connection counts, and deeper hierarchies in AI-enabled models, which correlate with greater verification challenges and longer fault-detection times for AI configurations. The work highlights the need for adapted verification strategies for AI-CPS and provides a replication package to support reproducibility and extended studies in this domain.

Abstract

In the world of Cyber-Physical Systems (CPS), a captivating real-time fusion occurs where digital technology meets the physical world. This synergy has been significantly transformed by the integration of artificial intelligence (AI), a move that dramatically enhances system adaptability and introduces a layer of complexity that impacts CPS control optimization and reliability. Despite advancements in AI integration, a significant gap remains in understanding how this shift affects CPS architecture, operational complexity, and verification practices. The extended abstract addresses this gap by investigating architectural distinctions between AI-driven and traditional control models designed in Simulink and their respective implications for system verification.

Architectural Transformations and Emerging Verification Demands in AI-Enabled Cyber-Physical Systems

TL;DR

The paper investigates how AI integration reshapes the architecture and verification of CPS by comparing AI-driven DRL controllers against traditional MPC/PID controllers implemented in Simulink. It introduces a multi-method framework with four phases to collect models, analyze structural composition, examine dynamic flow, and assess verification implications using S-TaLiRo falsification across eight benchmark systems. Key findings include a shift from continuous to discrete, logic-driven components, higher block/connection counts, and deeper hierarchies in AI-enabled models, which correlate with greater verification challenges and longer fault-detection times for AI configurations. The work highlights the need for adapted verification strategies for AI-CPS and provides a replication package to support reproducibility and extended studies in this domain.

Abstract

In the world of Cyber-Physical Systems (CPS), a captivating real-time fusion occurs where digital technology meets the physical world. This synergy has been significantly transformed by the integration of artificial intelligence (AI), a move that dramatically enhances system adaptability and introduces a layer of complexity that impacts CPS control optimization and reliability. Despite advancements in AI integration, a significant gap remains in understanding how this shift affects CPS architecture, operational complexity, and verification practices. The extended abstract addresses this gap by investigating architectural distinctions between AI-driven and traditional control models designed in Simulink and their respective implications for system verification.

Paper Structure

This paper contains 16 sections, 4 equations, 6 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Approach
  4. Phase-A: Model Collection and Filtering
  5. Phase-B: Structural Composition Analysis
  6. Phase-C: Dynamic Flow Analysis
  7. Phase-D: Implications on CPS verification
  8. Evaluation
  9. Experimental Settings
  10. RQ1
  11. RQ2
  12. RQ3
  13. Threats to Validity
  14. Related Works
  15. Conclusion
  16. ...and 1 more sections

Figures (6)

  • Figure 1: Control Framework of Adaptive Cruise Control System.
  • Figure 2: CPS workflow with AI vs. Traditional Controller
  • Figure 3: Overview of our Multi-Method Approach.
  • Figure 4: A Simple Flow Graph form the ACC Model
  • Figure 5: Category-Wise Atomic Block Differences between AI-Driven and Traditional CPS models.
  • ...and 1 more figures