Modern Middlewares for Automated Vehicles: A Tutorial

Klüner, David Philipp; Molz, Marius; Kampmann, Alexandru; Kowalewski, Stefan; Alrifaee, Bassam · 2026 · Crossref

DOI: 10.1109/mits.2026.3654011

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This tutorial addresses the critical transition in automotive software architectures driven by the rise of Automated Vehicles (AVs). The authors identify that traditional domain-based Electrical/Electronic (E/E) architectures, which rely on dedicated, resource-constrained Electronic Control Units (ECUs) and static signal-oriented communication, are insufficient for the complex perception, decision-making, and continuous update requirements of modern AVs. These legacy systems suffer from bandwidth bottlenecks, wiring complexity, and poor software reusability. Consequently, the paper aims to provide a comprehensive overview of modern middlewares that facilitate the shift toward centralized, zone-based E/E architectures, where high-performance compute platforms handle multiple functions. The paper employs a tutorial-style review methodology, structuring its analysis around fundamental concepts, comparative evaluation, and architectural implications. It first defines the evolution from domain-based to zone-based E/E architectures, highlighting the role of Zone Control Units (ZCUs) and automotive Ethernet. It then distinguishes between communication middlewares and architecture platforms to clarify terminology. The core of the study involves a detailed examination and comparison of five state-of-the-art middlewares: FastDDS, SOME/IP, Zenoh, ROS 2, and AUTOSAR Adaptive Platform (AP). The authors analyze these tools based on their capabilities, supported features, software architecture integration, and performance characteristics. Additionally, the paper uses a case study derived from the UNICARagil project to illustrate how different software architectures impact the maintainability and expansion of automated driving functions, specifically contrasting the difficulties of updating legacy signal-oriented systems with the flexibility offered by middleware-based approaches. The findings highlight that modern middlewares serve as foundational software layers that abstract the operating system, enabling developers to focus on application-specific tasks while providing crucial services such as communication, security, and over-the-air updates. The comparison reveals significant variations among the five selected middlewares in terms of licensing, feature sets, and suitability for automotive constraints. The case study demonstrates that while legacy architectures offer predictable behavior and cost optimization for specific tasks, they are prohibitively expensive and complex to modify post-manufacture. In contrast, middleware-enabled architectures support dynamic reconfiguration and software reuse, which are essential for Software Defined Vehicles (SDVs). The paper identifies that while middlewares like ROS 2 are ubiquitous in robotics, they have not yet achieved market dominance in the automotive sector, which remains under active development. The significance of this work lies in its guidance for manufacturers, developers, and researchers navigating the selection of middleware for future vehicle software. By clarifying the distinctions between communication middlewares and architecture platforms, and by evaluating current state-of-the-art options, the paper provides a framework for designing scalable, flexible, and maintainable automotive software systems. It underscores the necessity of adopting modern middlewares to overcome the limitations of traditional E/E architectures, thereby enabling the continuous evolution of AV capabilities throughout the vehicle lifecycle. The discussion of open research challenges further directs future efforts toward optimizing these middlewares for the specific deterministic and safety requirements of the automotive domain.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• driverless ads