MENU

MOST and AVB: Two Candidates for Next-Gen Automotive Infotainment Networks

MOST and AVB: Two Candidates for Next-Gen Automotive Infotainment Networks

Technology News |
By eeNews Europe



The MOST Technology framework is the dominant technology in automotive telematics and infotainment networks, and has been smoothly brought to the road in its third generation with MOST150 in 2012. The MOST Cooperation has proven to be an excellent body, permitting very direct specification and implementation processes for the benefit of both OEMs and suppliers. The close and open cooperation of OEMs and suppliers is the key factor of this success story. MOST150 systems will continue to boom in car models as the dominant infotainment network system until at least 2020.

A succeeding “Next Generation” network system for the time after MOST150 needs to meet high expectations, considering the dramatic increases in bandwidth demands driven by camera, display link and consumer electronics (CE) device based use cases. Flexible topology options and the seamless interconnection of previously separated car domains are further key topics. Aside from that, general requirements apply, such as cost, scalability, future-proof design, efficient supply chains, automotive maturity, tools ecosystem and IT security.

With IEEE802.1 Audio-/Video-Bridging standards [1-5], packet-based synchronized streaming has become a competitive alternative in media streaming applications. Ethernet Network Interfaces supporting AVB have just been launched in consumer products [6] and AVB will conceivably become a common feature in CE devices which commonly use IEEE802.3 Ethernet and IEEE802.11 WiFi LAN technologies.

Here, another trend has to be taken into account, as Ethernet is also currently evaluated in the automotive industry as a future technology option in automotive domains adjacent to the telematics domain.

Due to these facts and driven by new use cases, cross-domain interoperability is rapidly gaining importance and the applicability of AVB concepts and their interoperability with MOST need to be evaluated.

MOST and AVB

Both the AVB and the MOST approaches encounter critiques on various technical aspects. These can be answered more or less satisfyingly, as we are talking about systems-to-be and that kind of critique is mostly based on experience gained from current systems. Concept proposals, on the other hand, are the technical basis for the time being. They are a flexible matter and therefore often deliver only weak arguments for hard technical decisions.

MOST as a synchronous system has the advantage of inherently providing maximum Quality of Service, since collisions or priority conflicts principally do not occur. Delay is minimal and, moreover, it is deterministic and predictable, thus enabling application-level synchronization with simple fixed delay elements. On top of this, MOST150 offers isochronous and packet channels, including a transparently usable Ethernet channel. In the short term, the MOST Multiport INIC concept as presented earlier [7] already answers the most frequently raised objections against classic MOST systems by enabling flexible, mixed topologies and different speed grades on the same net. A MOST successor, preliminarily called “MOSTnG”, will further improve on this, key points being error detection/correction and bridging between network clusters, which may run at different speed grades over various, even mixed physical layers.

With the introduction of AVB standards, Ethernet as a packet-switched networking technology responds to the lack of guarantees and determinism with a synchronized network layer, enabling time-synchronized low-latency streaming services. The standards cover precise clock synchronization, traffic classes with bounded latency, partitioning of bandwidth, stream reservation, traffic prioritization and data link layer transport protocol. Layering of this synchronized network then looks very similar to synchronous MOST (Figure 1). However, application level synchronization is a question not in the scope of AVB standards so far and thus must be taken into account by application developers.

Figure 1: ISO/OSI model layering for synchronized audio/video streaming on MOST and on AVB

Regarding automotive infotainment systems, a crucial feature is the seamless integration of audio/video (A/V) interfaces in the network interface controllers. A/V streams should be directly transferred to corresponding A/V interfaces, thus unloading the host processor from handling A/V tasks. The MOST150 INIC already integrates various interfaces such as MediaLB, TSI, SPI, I2C and I2S, benefitting from its synchronous network layer transport. AVB-enabled Ethernet NICs can, in principle, provide similar interfaces by separating AVB streams from other traffic on the Data Link Layer, taking care that the AV streams do not burden the host processor. However, such functions are not found in today’s NIC implementations (Figure 2).

Figure 2: Integration of host and physical layer hardware interfaces in MOST and Ethernet AVB.

Both technologies are essentially independent of baseband physical layers. MOST150 INIC interconnects the FOT using LVDS interfaces, whereas Ethernet MACs connect to Ethernet PHY units with Media Independent Interface variants (also Figure 2). Otherwise, choice of a physical layer, whether on optical media, electrical over twisted pairs or coax, is to a large extent independent of the approach taken on the data link layer.

Whereas MOST has been designed and developed for automotive use and automotive qualification, for AVB a lot of experience needs to be gained regarding automotive use of the standards and realization of automotive requirements. This goes from the API level to service-discovery and control mechanisms, to wake-up-strategies and -times, to power consumption, just to name a few. AVB generation 2, which is only in standardization now, will address some crucial issues, including the handling of high frequency/low volume data such as sensor data and further minimization of latency by preemption of packets [8-10]. AVB standards were written to flexibly support comparatively dynamic networks and to provide interoperability between brands of different manufacturers. The situation in-car is a lot more static, with plenty of opportunity to keep oversizing, redundancy, start-up times and thus power consumption and costs down. Pre-configuration for the startup phase and beginning to reduce complexity right at design time are examples for that. A strip-down of AVB protocols and techniques is imperative for automotive usage. Yet an appropriate form of cooperation for standardization of such a tailored AVB automotive profile needs to be found that fits into the landscape of the IEEE and other groups concerned.

Both MOSTnG and AVB can be considered technologically feasible concepts, sufficient to meet the demands of next generation infotainment networks. MOST has a big bonus in confidence and maturity and is better adapted for automotive use. Development of MOSTnG could further on be directly influenced and tailored via the MOST Cooperation, in line with OEM demands, quickly and pragmatically.

There is a great deal of work to do for automotive use of AVB. Although its standardization enables flexible usage that results in a prolific supplier and market situation, the standardization processes are challenging, with lots of voices from non-automotive interest groups in manifold international committees and working groups.

With these points in mind, at the moment it is very difficult to identify clear decisive factors on a technical level for one approach or the other. Regarding market size and heterogeneity on each side, especially in the long term, AVB is an option that needs to be seriously considered, at least. As for automotive usage, primarily in the infotainment domain, AVB needs to be evaluated and its performance needs to be compared to the MOST system.

Evaluating AVB and Interoperability

For these reasons, we decided to build a system to evaluate both technologies on one single platform. We combine MOST and AVB networking domains by integrating multiple MOST and AVB adapters in standard PCs (Figure 3). The idea is to make use of a unified hardware, operating system and system clock environment to be able to correlate timestamps for both networking domains as a base method for evaluation on higher and application level. This will allow us to further explore interoperability scenarios between MOST and AVB, taking into account that Ethernet systems may be deployed in other time-sensitive automotive domains interacting with MOST Technology deployed in the infotainment domain.

Figure 3: Overview of the MOST and AVB evaluation platform. Standard PCs with NICs of both network domains (blue/red points). Note: MOST domain topology depicted as ring just for illustration purposes.

We aim to evaluate the full scope of AVB standards and will include further AVB building blocks as soon as they are available. Based on the results of these investigations, we then plan to narrow down what will apply to automotive profiles of AVB and interoperation possibilities with other systems. A first step is to create exemplary implementations of scenarios that are typical for automotive infotainment, though yet not common in the consumer market. One scenario for AVB evaluation is application-level synchronization of multiple media streams on different sinks, including CE devices (Figure 4). With AVB providing synchronization on a network level, we want to investigate what additional measures have to be implemented to achieve application-level synchronization and which preconditions have to be met.

Figure 4: Use case A/V-streaming from a source in the car-domain (head unit, tuner, Blu-ray player, etc.) to two tablet computers with user-domain media player, lip-sync with each other and cabin speakers.

As a second step, mapping of a present MOST150 system to AVB on the evaluation system will provide hands-on experience with AVB and allow direct comparison to the functionality and performance that MOST150 delivers. Driving the evaluation system to its limit, we will analyze its behavior regarding factors such as synchronization accuracy or impact on best effort traffic. Additionally, this will allow us to identify optimization possibilities, for example by using static stream classification, reservation or traffic shaping.

We also intend to use the evaluation system for research on how to integrate wireless CE devices. Applicability of AVB concepts for WiFi is therefore another topic we will investigate, though “WiFi AVB” is anything but definite so far.

The main focus of our investigation will be the interoperation of MOST150 and Ethernet/AVB. Synchronization of time and clocks is the basic property of both, and hence the premise for interoperation. Establishing a cross-system time base reference, therefore, is a key topic to build on for a gateway scenario. Another idea is to investigate possibilities to use or enhance the Ethernet channel of MOST to transparently support AVB. Finally, incorporating the steps mentioned before, our target scenario is cross-system, synchronized video-streaming, integrating wireless CE.

Literature

[1] IEEE, Inc. Audio Video Bridging Task Group. IEEE Standard 802.1.

[2] IEEE, Inc. Timing and Synchronization for Time-Sensitive Applications in Bridged Local Area Networks. IEEE802.1AS.

[3] IEEE, Inc. Forwarding and Queuing Enhancements for Time-Sensitive Streams. IEEE802.1Qav.

[4] IEEE, Inc. Stream Reservation Protocol.IEEE 802.1Qat.

[5] IEEE, Inc. Layer 2 Transport Protocol Working Group for Time-Sensitive Streams. IEEE 1722.

[6] https://communities.intel.com/community/wired/blog/2012/10/18/i210-launch-announcement

[7] Schöpp, H.: MOST – The Versatile Automotive Backbone on its Way into the Future. MOST-Forum 2012 Presentation and Special Issue MOST 2012, page 6 to 9.

[8] IEEE, Inc. Timing and Synchronisation: Enhancements and Performance Improvements. IEEE802.1ASbt

[9] IEEE, Inc. Frame Preemption. IEEE802.1Qbu

[10] IEEE, Inc. Enhancements for Scheduled Traffic 802.1Qbv

Authors

Günter Dannhäuser
works as scientific employee in car pre-development on telematics and infotainment networks at Daimler AG in Ulm. He represents the Daimler AG in the MOST multimedia streaming working group.
guenter.dannhaeuser@daimler.com

Dr. Walter Franz
leads the team Telematic Networks at Daimler AG car pre-development in Ulm. He represents the Daimler AG in the MOST Cooperation Steering Committee.
walter.franz@daimler.com

Prof. Dr. Wolfgang Rosenstiel
is professor of the Wilhelm Schickard Institute at the University of Tuebingen and director at the FZI (Forschungszentrum Informatik) in Karlsruhe (Germany).
rosenstiel@fzi.de

Article previously published in Elektronik automotive, MOST Special Edition, April 2013,www.elektroniknet.de

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News

Share:

Linked Articles
10s