In the first of three articles on Audio Video Bridging or AVB I’m going to try to explain a technical solution to a technical problem without getting too technical… Hmm. Wish me luck!
The first thing to say about AVB is that, for reasons examined in part 3 of this series, AVB is going through a name change. At the moment AVB/TSN is probably the most appropriate name but in the future it is likely to just be known as TSN which is short for Time-Sensitive Networking..
AVB/TSN is an open technology, like RAVENNA, However in spite of this AVB/TSN is different from RAVENNA in a very significant way. In the first of this Spotlight series I explained that one of the principles which guided the development of RAVENNA was that it must be based on existing standards and must work on existing networks. It must take networks as they exist in the world today and meet the basic requirements of a professional Audio over IP network of deterministic, low latency delivery of production quality audio. While computer networks are highly routeable and have the advantage of networked distribution as opposed to traditional point to point wiring, they do have some fundamental issues when used to replace traditional point to point wiring. The most significant of these is that because of the way IP/Layer 3 networks operate, it is impossible to accurately predict transit time across the network because while there are mechanisms in place which prioritise traffic on the network to some extent, it isn’t as fast as it could be because these Quality Of Service (QoS) measures can only react to traffic as it arrives - i.e. it is non-deterministic.
Best Fit Or Change The Rules?
The distinction being drawn here is right at the centre of the process we are going through as AoIP standardises and presumably, eventually converges - Do you create a technology which works within the limitations of current technology, and is therefore compatible with current technology, or do you change current technology to address those limitations? AVB/TSN seeks to address the Quality of Service issues in IP/Ethernet by changing Ethernet itself.
New Ethernet Standards
The IEEE are the people who look after Ethernet standards, their AVB Task Group aimed to achieve this by writing specifications which allow time-synchronised, low latency streaming through 802 (Ethernet) networks. AVB/TSN isn’t a protocol, it’s an evolution of standard Ethernet to support production quality audio and video, and control data. With Time Sensitive Networking (TSN) it provides three major enhancements:
- Precise, low jitter clocking.
- A method for an application or endpoint device to reserve the resources necessary to support a particular AV media stream.
- Queuing and forwarding rules (i.e. traffic management) which guarantees transit across the network within the time specified by the reservation.
As this is a description of the standards Ethernet is based on, it’s inevitably going to get technical but I’m not going to go too far into the discussion of standards here. For those who are curious there are detailed descriptions of the new elements introduced to the existing IEEE 802 standards available online. Wikipedia is an excellent place to start. For those of us who just want to know what it means there are four really important differences between Ethernet and AVB/TSN Ethernet. AVB/TSN supports:
- Precise Synchronisation/Clocking - Based on a revised version of Precision Time Protocol.
- Traffic Shaping for AV media streams - Priority is given to AV data and devices are prevented from dumping large bursts of data onto AVB/TSN reserved bandwidth.
- Admission Controls - A stream can only be created or subscribed to if the required bandwidth is available at every stage of its transit across the network.
- Identification of non-participating devices - For a stream to be successfully initiated, every part of the network through which that stream will be routed must meet specific hardware requirements (i.e. be AVB/TSN compliant).
So far so good, look out for part 2, where I look at how AVB/TSN manages traffic on the network to keep latency under control.