Simulation Validation of a Next-Generation Security Architecture for the Adaptive AutoSAR Platform Integrating an AI-Based Intrusion Detection System(AI-IDPS) and Hybrid Access Control(FA-RBAC)

Abstract

The rapid evolution of Software-Defined Vehicles (SDVs) and autonomous driving technologies has significantly increased in-vehicle communication complexity and cybersecurity vulnerabilities. While traditional static Role-Based Access Control (RBAC) is incapable of reflecting dynamic driving contexts, independent Attribute-Based Access Control (ABAC) introduces unsustainable computational overhead within AutoSAR environments. To address international regulations such as UNECE WP. 29 R155/156, ISO/SAE 21434, and ISO 24089, this paper proposes a threestage hybrid access control architecture. This framework integrates Flexible Attribute-based RBAC (FA-RBAC) and an AI-based Intrusion Detection System (AI-IDPS) into the Identity and Access Management (IAM) of the AutoSAR Adaptive Platform. Simulations verifying core SDV scenarios—including driving mode transitions, Over-The-Air (OTA) updates, V2X, and diagnostics—were conducted using a SOME/IP bridge between ROS2 and Adaptive AUTOSAR. The proposed model satisfied real-time constraints with a maximum latency of 7.2ms. Quantitative results further demonstrate a 98.5% detection rate for contextual zero-day attacks and a 60% reduction in the number of written policy rules (WPN).