During the transition to digital at your facility, you may have discovered that some of the most vexing challenges of transporting audio, video or both over a LAN are latency and lip sync issues.
Most IT-designed systems are built to move raw data through a network as quickly as possible at the highest reliability and lowest cost. The traditional IT network infrastructure doesn't care about timing synchronization, as demonstrated by the liberal use of buffering in most networks. In the IT world, the delivery of raw data reliably is far more significant than delivering data at a specific time.
In the professional audio and video world, priorities are a little different. Our transport of choice is SDI. However, we professionals know that it is not unusual for consumer-based technology, such as IEEE 1394 FireWire, to find its way into professional facilities and applications. This time, the new standard is IEEE 802.1, the IEEE Standard for Local and Metropolitan Area Network Audio Video Bridging (AVB) Systems. IEEE 802.1 is not a single standard but a set of standards designed to redefine how IEEE 802-based A/V networks are built, by addressing issues we broadcasters are quite familiar with.
Broadcast engineers have traditionally worked in a universe of single proprietary daisy-chain-style solutions. This would include simple back-of-the-rack devices such as audio bridges, which have been an important cornerstone in facility infrastructures for decades. When most old-school broadcast engineers think of a bridge, they think of a bridge rectifier or a constant impedance audio bridge. Some might think of it as a lattice. Throughout the history of analog production and broadcasting, engineers have worked, adding bridges, wire and gear in the most fundamental layer of today's IT structure, Layer 1.
Let's take a moment to review a network's open systems interconnection model, or OSI as it is defined by the ISO. OSI consists of seven layers. Layer 1 is known as the Physical Layer and it includes the elements necessary to move bits of data from point to point using cables and connections. Copper wire, fiber and IEEE 802.11 Wi-Fi are examples of Physical Layer 1 components.
AVB is operates in OSI Layer 2, the Data Link Layer. It moves bits organized into frames over the Layer 1 infrastructure, for physical addressing in point-to-point and point-to-multipoint audio and video communications. It also facilitates the interactions of multiple devices with a shared source.
OSI Layer 3 is the Network Layer, which determines the path and logical addressing. Layers 1, 2 and 3 are known as the Media Layers. The remaining four layers are called Host Layers. OSI Layer 4 is the Transport Layer. Layer 5 is the Session Layer, which controls communications between computers. The Presentation Layer is Layer 6. It maps, decrypts and converts context between Application Layer 7, translating between application and network formats.
Physical Layer 7 interacts with software applications that require communication. FTP and HTTP are good examples.
A bridge is typically recognized as the fastest path from one point to another when a signal needs to be routed or switched. As the seriousness of timing and synchronization perception began increasing at professional and consumer levels alike, the IEEE assigned a task group in 2006 to establish a standard to address A/V timing issues.
According to the IEEE, "The purpose of this standard is to specify defaults and profiles that manufacturers of LAN equipment can use to develop AVB-compatible LAN components, and to enable a person not skilled in networking to build a network, using those components, that does not require configuration to provide working Audio and/or Video services." The latest version of this standard, IEEE 802.1BA Draft 2.5, was submitted to the IEEE on July 19, 2011.
In 2009, several manufacturers formed an association called the AVnu Alliance, (www.avnu.org), to promote the adoption of the IEEE 802.1 Audio Video Bridging (AVB), and the related IEEE 1722 and IEEE 1733, standards over various networking link layers. The primary focus of this growing association is on AVB devices and applications in the automotive, professional and consumer electronics markets. In terms of markets and marketing, AVnu appears to be following a path similar to HDMI.org and its parent HDMI Licensing LLC took to define, test, control and promote HDMI interfaces. Specifically, the mission of the AVnu Alliance is to create AVB compliance test procedures and processes to help ensure interoperability between devices that implement the AVB standards.
AVB-compliant networks use a set of protocols as described by IEEE 802.1 that address precise synchronization, traffic shaping for media streams to evenly distribute packets in time, admission controls and identification of nonparticipating devices.
It's about time
IT systems are often based on Internet-calibrated system clocks, but nothing in typical network infrastructure is concerned with precise timing of data delivery. Networks don't address network congestion. They instead depend on higher level protocols to handle TCP issues by controlling transmission bandwidth and retransmitting lost or dropped packets. That works well with text and data, but can create bottlenecks streaming audio and video.
Three key factors make AVB-capable systems work. One is the Grand Master Clock. Every AVB–capable device can be a Grand Master Clock, but only one can be designated the Grand Master Clock within a network. The Grand Master Clock can free-run or be locked to another system clock. A low-jitter Grand Master Clock ensures accurate synchronization of multiple streams.
The second key is a simple reservation protocol allowing an application on an endpoint device to notify networked elements in a path so they can reserve resources necessary to support a particular stream. This is accomplished by inserting specific tags in AVB packets by end points in a signal flow.
Let's say, for example, an AVB-capable system in the newsroom needs to communicate with an AVB-capable system in master control. As the connection is made, the system in the newsroom, known as the "talker," will send out a request known as "talker advertise" message. This message includes the MAC address of the stream source, a talker-specific 16-bit unique ID and the MAC address of the stream destination. The "talker advertise" message also includes QoS requirements and accumulated worse-case latency data. Worse case latency is also recalculated at each AVB bridge so the "listener" in Master Control can pass this information to higher layers for media synchronization if necessary.
When an AVB intermediate bridge receives a "talker advertise" message, it checks bandwidth on its output ports. If the bridge has sufficient resources, it passes the "talker advertise" on to the next station on the network. If resources are not available, it sends a "talker failed" message, with a failure code and bridge identification so a higher layer can provide error checking or notification. If an AVB intermediate bridge receives a "talker failed" message, it passes the message on to the "listener" device. If the "listener" accepts the latency and resources in the "talker advertise" message, it will respond with a "listener ready" message forwarded back to the "talker" and the guaranteed streaming connection between the newsroom and Master Control is established.
The third and arguably equally important key is a set of queuing and forwarding rules that ensure defined streams will pass through a network within the delay specified by the reservation.
What makes AVB different from traditional non-AVB switching is that AVB streamlines the relay of Layer 2 Ethernet frames. AVB is fast because it gets all the information it needs in the tag at the head of each frame, instead of going through the usual process of storing, reading the destination address and searching a table to connect that address to a specified output port.
The AVB target delay, as specified by AVnu, should not to exceed 1ms. With specifications this tight, AVB has the potential to provide a wide spectrum of digital timing solutions for broadcast and AV professionals.