[Paper Review] A QoS Aware Approach to Service-Oriented Communication in Future Automotive Networks

Mehmet Çakır, Timo Häckel+4 more|arXiv (Cornell University)|Nov 5, 2019

Service-Oriented Architecture and Web Services20 references4 citations

TL;DR

This paper proposes a QoS-aware service-oriented middleware for future automotive networks using a dynamic QoS negotiation protocol and a multi-protocol stack to support heterogeneous real-time requirements. Evaluation in an OMNeT++-based simulation shows that setup times remain under 2 ms with up to 70% cross-traffic, meeting automotive QoS constraints for non-safety-critical services.

ABSTRACT

Service-Oriented Architecture (SOA) is about to enter automotive networks based on the SOME/IP middleware and an Ethernet high-bandwidth communication layer. It promises to meet the growing demands on connectivity and flexibility for software components in modern cars. Largely heterogeneous service requirements and time-sensitive network functions make Quality-of-Service (QoS) agreements a vital building block within future automobiles. Existing middleware solutions, however, do not allow for a dynamic selection of QoS. This paper presents a service-oriented middleware for QoS aware communication in future cars. We contribute a protocol for dynamic QoS negotiation along with a multi-protocol stack, which supports the different communication classes as derived from a thorough requirements analysis. We validate the feasibility of our approach in a case study and evaluate its performance in a simulation model of a realistic in-car network. Our findings indicate that QoS aware communication can indeed meet the requirements, while the impact of the service negotiations and setup times of the network remain acceptable provided the cross-traffic during negotiations stays below 70% of the available bandwidth.

Motivation & Objective

• To address the lack of dynamic QoS support in existing automotive middleware for service-oriented communication.
• To enable QoS-aware communication in future automotive networks with heterogeneous real-time requirements.
• To design a protocol stack that supports multiple communication classes based on rigorous requirements analysis.
• To evaluate the performance impact of QoS negotiation on setup times in realistic in-car network scenarios.
• To validate the feasibility of dynamic QoS negotiation under realistic cross-traffic conditions.

Proposed method

• Proposes a QoS Negotiation Protocol (QoSNP) for dynamic, service-level QoS agreement negotiation.
• Designs a multi-protocol stack supporting four distinct QoS classes: Best-Effort, Rate-Constrained, Static Real-Time, and Audio-Video Bridging.
• Integrates the QoSNP with the SOME/IP middleware and TSN-based Ethernet to enable QoS-aware service communication.
• Uses a case study with a realistic automotive network topology derived from a production car’s legacy CAN-based architecture.
• Simulates the network in OMNeT++ with real in-car communication traces to measure setup times and protocol overhead.
• Evaluates performance under varying cross-traffic levels (0–1000 Mbit/s) to assess scalability and delay impact.

Experimental results

Research questions

• RQ1 Can a dynamic QoS negotiation protocol effectively support heterogeneous QoS requirements in automotive service-oriented communication?
• RQ2 What is the impact of cross-traffic on QoS negotiation setup times in a realistic in-vehicle network?
• RQ3 Does the proposed middleware maintain acceptable setup delays under realistic load conditions?
• RQ4 How do different QoS classes (e.g., RT, BE) coexist and interoperate in a multi-protocol stack?
• RQ5 Can the system meet automotive real-time requirements (e.g., 150–200 ms) for non-safety-critical services?

Key findings

• The proposed QoS negotiation protocol enables dynamic QoS agreement between services, supporting four distinct QoS classes.
• Setup times remain below 2 ms for up to 70% of available bandwidth under cross-traffic, which is acceptable for most automotive applications.
• At approximately 60–70% cross-traffic, setup times begin to increase drastically, indicating a performance threshold.
• The minimum setup time remains nearly constant, while average and maximum times grow almost exponentially with increasing cross-traffic.
• The system complies with automotive QoS requirements of 150–200 ms for non-safety-critical traffic, with protocol delays in the millisecond range.
• For safety-critical traffic, static TDMA-based classes are recommended to avoid negotiation-induced jitter.