Towards Mixed-Criticality Software Architectures for Centralized HPC Platforms in Software-Defined Vehicles: A Systematic Literature Review

TL;DR

This systematic literature review explores how mixed-criticality software can be integrated into centralized HPC architectures for software-defined vehicles, providing guidelines and architectural insights for industry and research.

Contribution

It introduces a rigorous, systematic methodology for analyzing and designing mixed-criticality automotive software architectures based on empirical literature review.

Findings

Identification of key functional domains and constraints

Techniques and architectural patterns for mixed-criticality integration

A prototype architecture demonstrating the approach

Abstract

Centralized electrical/electronic architectures and High-Performance Computers (HPCs) are redefining automotive software development, challenging traditional microcontroller-based approaches. Ensuring real-time, safety, and scalability in software-defined vehicles necessitates reevaluating how mixed-criticality software is integrated into centralized architectures. While existing research on automotive SoftWare Architectures (SWAs) is relevant to the industry, it often lacks validation through systematic, empirical methods. To address this gap, we conduct a systematic literature review focusing on automotive mixed-criticality SWAs. Our goal is to provide practitioner-oriented guidelines that assist automotive software architects and developers design centralized, mixed-criticality SWAs based on a rigorous and transparent methodology. First, we set up a systematic review protocol grounded in established guidelines. Second, we apply this protocol to identify relevant studies. Third, we extract key functional domains, constraints, and enabling technologies that drive changes in automotive SWAs, thereby assessing the protocol's effectiveness. Additionally, we extract techniques, architectural patterns, and design practices for integrating mixed-criticality requirements into HPC-based SWAs, further demonstrating the protocol's applicability. Based on these insights, we propose an exemplary SWA for a microprocessor-based system-on-chip. In conclusion, this study provides a structured approach to explore and realize mixed-criticality software integration for next-generation automotive SWAs, offering valuable insights for industry and research applications.