Reconstruction-Based Adaptive Scheduling Using AI Inferences in Safety-Critical Systems

2025-09-24

AI Summary

To address safety-critical challenges—including message conflicts, deadlock risks, and scheduling infeasibility—in adaptive scheduling of time-triggered systems (TTS) under dynamic environments, this paper proposes a reconstruction-based, AI-driven scheduling framework. The framework integrates a lightweight AI inference model with a priority transformation algorithm, enabling online scheduling reconstruction to enforce precedence constraints and ensure conflict-free communication. It further incorporates formal constraint checking, dynamic resource allocation, and exception recovery modules to support fault-tolerant recovery and multi-objective optimization—namely, makespan minimization, load balancing, and energy efficiency—across operational mode switches. Experimental evaluation demonstrates that the approach guarantees 100% scheduling feasibility and safety while reducing average makespan by 18.7%, improving scheduling success rate by 9.3%, and lowering computational overhead by 42% compared to conventional search-based methods—thereby significantly enhancing the real-time performance and robustness of TTS in safety-critical dynamic scenarios.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories
• Humans and AI: Planning and Decision Support for Human-Machine Teams

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Adaptive scheduling is crucial for ensuring the reliability and safety of time-triggered systems (TTS) in dynamic operational environments. Scheduling frameworks face significant challenges, including message collisions, locked loops from incorrect precedence handling, and the generation of incomplete or invalid schedules, which can compromise system safety and performance. To address these challenges, this paper presents a novel reconstruction framework designed to dynamically validate and assemble schedules. The proposed reconstruction models operate by systematically transforming AI-generated or heuristically derived scheduling priorities into fully executable schedules, ensuring adherence to critical system constraints such as precedence rules and collision-free communication. It incorporates robust safety checks, efficient allocation algorithms, and recovery mechanisms to handle unexpected context events, including hardware failures and mode transitions. Comprehensive experiments were conducted across multiple performance profiles, including makespan minimisation, workload balancing, and energy efficiency, to validate the operational effectiveness of the reconstruction models. Results demonstrate that the proposed framework significantly enhances system adaptability, operational integrity, and runtime performance while maintaining computational efficiency. Overall, this work contributes a practical and scalable solution to the problem of safe schedule generation in safety-critical TTS, enabling reliable and flexible real-time scheduling even under highly dynamic and uncertain operational conditions.

Problem

• Dynamically validate and assemble schedules for time-triggered safety-critical systems
• Transform AI-generated priorities into executable schedules with safety constraints
• Handle message collisions and incomplete schedules in dynamic operational environments

Innovation

• Reconstruction framework dynamically validates and assembles schedules
• Transforms AI-generated priorities into executable schedules with constraints
• Incorporates safety checks and recovery for unexpected events