Dependency-Aware Over-the-Air Framework for Reliable Software-Defined Vehicle Updates

A Framework for Cross-Domain Integration and Validation in Software-Defined Vehicle Systems

The transition from isolated domain-specific electronic control unit (ECU) architectures to fully integrated software-defined vehicle (SDV) platforms has introduced a new class of engineering challenge: cross-domain integration complexity. As vehicle software increasingly emerges from the dynamic interaction of subsystems spanning body control, powertrain coordination, thermal management, active safety, and connectivity, validation strategies designed for component-level verification are fundamentally insufficient. Integration defects   arising from timing misalignment, inconsistent signal interpretation, and conflicting control arbitration across domains surface late in development when remediation costs are highest. This paper presents a structured framework for cross-domain integration and system-level validation in SDV environments. The framework unifies four foundational engineering capabilities: explicit modeling of inter-domain dependencies, standardized signal interface definitions aligned with AUTOSAR Adaptive specifications, cross-domain temporal synchronization mechanisms, and coordinated scenario-based validation orchestration. Empirical evaluation against component-only validation baselines, using a reference defect set of confirmed cross-domain integration defects drawn from multi-domain automotive program integration records, demonstrates that the framework increases cross-domain defect detection rate by 64%, reduces integration cycle time by 58%, and expands cross-domain validation coverage from 43% to 91%. By elevating cross-domain interactions from emergent side effects to first-class engineering artifacts, the framework materially advances the state of practice in SDV system integration and validation

From Timing Budgets to WCETs: Robust SIL-and BSW-Aware Clustering and Allocation for Iterative Automotive Software Development

Towards a Unified Engineering Approach for Variability and Modular Architecture Management in Automotive Systems

Automotive systems are experiencing a rapid increase in complexity driven by the transition towards software-defined vehicles, autonomous functionalities and increasingly interconnected E/E architectures. This transformation intensifies variability across hardware and software domains and challenges established engineering approaches. Traditional modular product development (MPD) provides structural mechanisms to manage hardware complexity, while systems and software product line engineering (SPLE) offers methods for managing software variability. However, these paradigms are typically applied in isolation and lack an integrated methodology capable of addressing cross-domain variability and architectural synchronization in automotive systems. This paper investigates how SPLE and MPD can be systematically integrated to manage variability and architectural complexity in automotive systems. Following a design-oriented research approach, industry requirements are derived from an automotive case study at an OEM. Existing SPLE and modularization approaches are analyzed against these requirements, revealing gaps in cross-domain traceability, synchronization mechanisms, and lifecycle coordination. Based on this analysis, we propose an integrated methodology that combines variability modeling principles from SPLE with architectural modularization concepts. The approach enables management of module structures, supporting system-level consistency in automotive environments. The main contribution is a model-based-integration framework that bridges variability management and modular architecture design to address increasing system complexity in the automotive industry.

A Fully Automated, Deployment-Aware Testing Pipeline for IoT-Based Automotive Applications

This work presents an end-to-end, deployment-aware testing pipeline for IoT-based automotive applications that combines requirement-driven test and code generation with large language model (LLM) and vision-language model (VLM) assistance, and human-in-the-loop curation to reduce manual effort and improve consistency.

From Industrial Information Integration to Closed-Loop Operations Synchronization: An Evidence-Based Review of Data-Driven Smart Manufacturing

Smart Factory programs increasingly connect shop-floor, quality, asset and planning data, but integrated data infrastructures do not necessarily align operational decisions. This paper reviews how smart manufacturing literature explains the transition from industrial data integration to closed-loop operations synchronization and value capture in high-throughput manufacturing contexts. Using the Systematic Search Flow method, 1949 records were screened and reduced to a final portfolio of 73 studies. The papers were coded by thematic cluster, dominant technology, research method, primary theme, value-stream coverage and operations-synchronization relevance. The coding shows that roadmaps, interoperability architectures, analytics applications and digital-twin models dominate the portfolio. Explicit operations-synchronization mechanisms are addressed in 16 of the 73 studies, mainly through planning-execution coupling and digital-twin-based decision support. Coverage across value streams is uneven, with stronger evidence for Strategy, Make and Plan than for Quality and Assets. Based on this evidence map, the paper proposes a Data-Driven Operations Synchronization Stack that links operational data capture, semantic and IT/OT interoperability, analytics-supported decision-making, closed-loop synchronization and operational or financial value capture.

Mapping C3P Control Objectives to ISO/IEC 42001 Requirements for AI Lifecycle Governance

This article examines how the four control objectives of the Continuous Compliance Control Protocol align with the requirements of ISO/IEC 42001 for managing an artificial intelligence system across its lifecycle, and how they generalize to autonomous agents. Organizations adopt artificial intelligence in regulated settings faster than they build the mechanisms that continuously prove governance, and a management system standard states what must be governed, while leaving open how durable evidence is produced as systems run. A baseline mapping establishes that the four control points meet the requirements of the standard for a conventional, human-in-the-loop deployment. The central finding is that the same four points generalize without modification to AI coding agents and to general agentic workflows, where a defined purpose, an auditable trail, verification, and safe delivery produce tamper-evident evidence at every stage. That evidence assembles into a chain of custody for the agent that supplies exactly the documentation an ISO/IEC 42001 audit expects. The work serves compliance leaders, audit professionals, and engineering teams who govern autonomous artificial intelligence.