Franz Korf

Tomorrow's In-Car Interconnect? A Competitive Evaluation of IEEE 802.1 AVB and Time-Triggered Ethernet (AS6802)

Ethernet-based in-car communication is currently a hot topic in the automotive industry. Soon Ethernet will start to oust MOST bus in its domain of info- and entertainment applications. However, the full benefit of a technologically integrated in-car network will only become rewarding with the deployment of an Ethernet-based backbone that integrates all automotive domains on a single layer at increased bandwidth, reduced complexity and cost, while opening car intelligence for future innovations. Such backbone must transport critical control data in real-time. Standard Ethernet requires extensions to comply with the strict timing requirements of driver assistance and safety applications while simultaneously supporting broadband multimedia traffic. In this paper, we compare IEEE 802.1 AVB and Time-triggered Ethernet, two competing real-time approaches. While the first fosters over- provisioning and prioritisation, the second is based on a coordinated time-division-multiple-access (TDMA) policy for media access. By simulating a realistic in-car backbone design and traffic model, we reveal the strengths and weaknesses of both protocols and point to the diverging characteristics of event- and time-triggered policies. Our results show that in this in-car network scenario both protocols are able to meet the rigid timing requirements, while each has its unique benefits and disadvantages.

Comparing time-triggered Ethernet with FlexRay: An evaluation of competing approaches to real-time for in-vehicle networks

FlexRay is considered the next generation state-of-the-art technology for in-car networks, while time-triggered Ethernet emerges with the promise to integrate real-time and best-effort traffic into one homogeneous backbone. This paper contributes a competitive analysis of FlexRay and time-triggered Ethernet. By showing that it is possible to transfer a fully utilized FlexRay system to a system based on time-triggered Ethernet, it is demonstrated that time-triggered Ethernet is a suitable replacement of current in-vehicle bus-systems. Further it is shown that a switched system has advantages in bandwidth utilization over a shared bus, when using group communication.

Extending IEEE 802.1 AVB with time-triggered scheduling: A simulation study of the coexistence of synchronous and asynchronous traffic

In-car networks based on Ethernet are expected to be the first choice for future applications in the domain of info-and entertainment. However, the full benefit of a technologically integrated in-car network will only become rewarding with an Ethernet-based backbone, unifying several automotive domains in a single infrastructure. Today, there is remarkable interest in the IEEE 802.1 Audio/Video Bridging (AVB) protocol suite, that provides end-to-end performance guarantees in Ethernet networks. But for the strict timing requirements of automotive control-traffic, these guarantees are too weak. An extension of Ethernet AVB with synchronous time-triggered traffic can overcome these limitations. In this paper, we investigate the coexistence of synchronous and asynchronous traffic by experimentally adding time-triggered messages to the credit-based shaper of AVB in a straightforward way. Based on simulations and analytical evaluations, we quantify the impact of such integration concepts for a reasonable design range. Our results demonstrate the feasibility of a shaping strategy with concurrent AVB and time-triggered message, but show a significant impact of the schedule design on the asynchronous AVB streams. Based on our findings, we provide recommendations for configurations that can improve end-to-end network performance for in-car applications by over 100%.

Performance Analysis of Time-Triggered Ether-Networks Using Off-the-Shelf-Components

The performance analysis and validation of distributed real-time systems poses significant challenges due to high accuracy requirements at the measurement tools. A fully synchronized time-scale at ultrafine granularity is not easy to generate. Even though there are several analyzer tools for standard switched Ethernet, these tools cannot be applied in time-triggered networks, since they do not meet the requirements of synchronized packet generation. This paper introduces a low cost and lightweight approach to measure end-to-end latency of time-triggered Ethernet traffic with off-the-shelf components. By using standard computer hardware and a real-time Linux Kernel, it is shown that measurement can be achieved in a resolution of microseconds. Furthermore, a validation with an Ethernet performance analyzer and a mathematical framework is presented to check the given results.

Demo: Real-time Ethernet in-car backbones: First insights into an automotive prototype

The communication infrastructure of todays automobiles forms a complex composition of heterogeneously interconnected components. At the same time, demands for higher bandwidth and low-latency communication are emerging from chassis control, camera based driver assistance and infotainment that cannot be accommodated by established technologies. A new approach towards a flexible highly scalable network is real-time Ethernet. The RECBAR research project develops and evaluates concepts and technologies for next-generation in-car backbones. In this demo we show a prototype based on a 2014 Volkswagen Golf 7 series car additionally equipped with high-bandwidth sensors, such as HD cameras and 3D laser scanners. The car uses a backbone network utilising time-triggered real-time Ethernet for the deterministic transmission of messages with hard realtime as well as rate-limiting and best-effort frames for messages with relaxed timing requirements. On the physical layer the setup utilises the OPEN Alliance 100Mbit/s BroadR-Reach (OABR or 100 BASE-T1) in addition to 100BASE-TX.

Object Tracking and Dynamic Estimation on Evidential Grids

Autonomous driving is one of the most challenging tasks of the automotive industry. As a subtask, the estimation of driveable and non driveable space is often solved by applying occupancy grids. The information about non driveable space can be used to improve object tracking. This paper presents an approach for object tracking and modelling in an occupancy grid map. Tracking objects on grid cells yields the advantage of a consistent environmental model on the occupancy grid map. We introduce the occupancy grid map as the only information source for the object tracking module. Taking advantage of the Dempster Shafer theory, a dynamic belief of conflicting cells can be estimated. This dynamic belief is then accumulated in a tracked object model. This is a grid based free form object model that uses detached grid cells to model vehicles in urban environment. We reduce false positives and initialization time by maintaining a dynamic belief for each object.

Software stacks for mixed-critical applications: Consolidating IEEE 802.1 AVB and time-triggered ethernet in next-generation automotive electronics

Real-time Ethernet variants are expected to build the future communication infrastructure in cars. First camera based driver assistance functions will communicate using IEEE 802.1 AVBs credit-based shaping. But for the strict timing requirements of automotive control-traffic, AVBs current timing guarantees are insufficient. The upcoming IEEE 802.1Qbv standard proposes synchronous time-triggered traffic to overcome these limitations. This paper presents a low footprint microcontroller based communication architecture, that supports both traffic classes in parallel while using standard hardware components. It allows first realistic performance analyses of coexistent traffic shaping strategies in a software based implementation.

Efficient automotive grid maps using a sensor ray based refinement process

The occupancy grid mapping technique is widely used for environmental mapping of moving vehicles. Occupancy grid maps with fixed cell size have been extended using the quadtree implementation with adaptive cell size. Adaptive grid maps have proven to be more resource efficient than fixed cell size grid maps. Dynamic cell sizes introduce the necessity of a split and merge process to trigger the refinement of grid cells. This paper presents a novel ray-based refinement process in order to choose the appropriate resolution for the sensor observation. Based on measurement conflicts some approaches use an iterative refinement process until all conflicts are solved. In contrast this paper presents an non-iterative approach based on the sensor resolution. Using the measurement data efficiently we propose an algorithm, which solves the problem of partially free cells in an adaptive grid map. The proposed algorithm is compared against other widely used algorithms and methodologies.

Design of TDMA-based in-car networks: Applying multiprocessor scheduling strategies on time-triggered switched ethernet communication

Real-time Ethernet variants gain importance for communication infrastructure of various time-critical domains, such as in-car networks. Synchronous time-triggered traffic guarantees strict timing but requires a detailed schedule for all participants. Designing these schedules by hand is extensive work and with increasing network size almost impossible. In this paper, we contribute a mapping of the time-triggered network scheduling problem into the domain of multiprocessor scheduling. This set of transformation rules allows us to apply established scheduling algorithms as well as new strategies to organise time-triggered switched networks. Experimental results from a prototype implementation of a scheduling framework based on this mapping show the feasibility of our concept. The framework demonstrates a multiple solver approach that uses algorithms with different optimality criteria in parallel.

Beware of the hidden! How cross-traffic affects quality assurances of competing real-time Ethernet standards for in-car communication

Real-time Ethernet is expected to become the core technology of future in-car communication networks. Following its current adoption in subsystems for info- and entertainment, broadband Ethernet promises new features in the core of upcoming car series. Its full potential will enfold when deploying Ethernet-based backbones that consolidate all automotive domains on a single physical layer at increased bandwidth but reduced complexity and cost. In such a backbone, traffic with a variety of real-time requirements and best-effort characteristics will share the same physical infrastructure. However, certain applications like online diagnosis, data- or firmware updates, and access to off-board backends will introduce bursty high traffic loads to the sensitive core of the cars communication network. In this work, we analyze the robustness against cross-traffic of real-time Ethernet protocols. Based on a realistic in-car scenario, we demonstrate that background cross-traffic can have significant impact on in-car backbone networks-even for real-time protocols with strict prioritization. By comparing the real-time approaches Ethernet AVBs asynchronous credit based shaping with the time-triggered and rate-constrained traffic classes of Time-triggered Ethernet (AS6802) we quantify how different media access policies suffer from low priority bursts of applications such as diagnosis, online updates or backend-based services. Our simulation study of a realistic in-car backbone design and traffic model reveals that in a realistic in-car network design, cross-traffic may increase end-to-end latency by more than 500% while the jitter can become 14 times higher than for a network without background tasks. We discuss ways to mitigate these degrading effects.