A Behavior-Centric Hardware-in-the-Loop Validation Framework for Cross-Domain Interaction Analysis in Automotive Chassis and Powertrain Systems

Abstract

Modern automotive embedded control systems are increasingly characterized by multi-rate execution environments, distributed electronic control unit (ECU) architectures, and tightly coupled interactions among chassis, powertrain, braking, and vehicle stability subsystems. As vehicles evolve toward software-defined and highly automated platforms, ensuring system reliability requires validation approaches capable of capturing complex cross-domain interactions rather than evaluating individual components in isolation. Conventional Hardware-in-the-Loop (HIL) validation methodologies predominantly employ signal-centric verification strategies that assess input-output relationships of individual ECUs. Although effective for component-level testing, these approaches are insufficient for detecting emergent system behaviors arising from interactions among distributed controllers, communication networks, and shared control objectives. Such behaviors may include torque coordination anomalies, delayed stability responses, oscillatory control effects, and fault propagation across interconnected subsystems. This paper proposes a behavior-centric HIL validation framework that elevates system behavior to a primary validation artifact through formal representations based on temporal constraints, causal state transitions, and Signal Temporal Logic (STL) specifications. The framework integrates Functional Mock-up Interface (FMI)-based multi-domain co-simulation, observer-based residual generation, structured fault injection mechanisms, and real-time behavioral monitoring to systematically identify, quantify, and diagnose emergent behaviors during validation. Experimental evaluation was conducted using a dSPACE SCALEXIO HIL platform across 240 cross-domain test scenarios involving chassis and powertrain interactions under normal and fault conditions. The results demonstrate significant improvements over conventional signal-centric validation approaches, achieving a fault-detection coverage of 93.7% compared with 61.2% for the baseline method, a mean fault-detection latency of only 38 ms, and a temporal-constraint satisfaction rate of 91.4%. These findings indicate that behavior-centric validation provides a more comprehensive assessment of system-level dynamics and safety-critical interactions. The proposed framework transforms HIL from a component-verification environment into a behavioral-intelligence platform, offering automotive validation engineers a rigorous and scalable methodology for validating next-generation connected, autonomous, and software-defined vehicle architectures.