AES67-2013 Looks at Audio Over IP

Audio system design has been an ever-evolving process. As new technologies emerge they add to and complement what’s come before, or replace existing structures.

Analog transitioned to digital. Single signal per cable wiring shifted to multiple multiplexed signals such as AES3 and MADI. Mixers with dedicated inputs and outputs have morphed into mixing control surfaces that act on sources wired to a central electronics hub. Distribution amplifiers are being replaced by Ethernet switches to allow multiple control surfaces and destinations access to any signals within a closed network.

The next step in this evolution looks to be audio-over-IP, or rather more of it. With the publication of AES67-2013, “AES standard for audio applications of networks— High-performance streaming audio- over-IP interoperability,” in September 2013, greater opportunities for integrating IP technology are forecast.

Such was the sentiment expressed at the DTV Audio Group Forum held in New York in December. The forum presented an overview of AoIP, how it compares to other interconnection technologies, and application potential.

“Looking forward, AES67 is enabling where the future is going,” said Greg Shay, chief science officer, Telos Alliance, and one of the DTV Audio Forum’s presenters. “It’s not just an interface technology, but how you interconnect and define a facility. We want the studio to fit into the whole world.”

While high-performance AoIP technology and equipment have been around for at least a decade, my impression is that these systems have been more prevalent in radio than TV installations, and as a way of distributing audio within closed systems.

A stumbling block for more extensive applications was that systems made by different manufacturers weren’t compatible with each other. Similar, yes; IP-based, yes; but not close enough. AES67 should change that, and if the standard is followed, then different manufacturers’ AoIP gear should be able to communicate with each other and pass and interpret AoIP packets from one system to another, both inside a facility, and outside to wide area networks.

Andreas Hildebrand This interoperability with its potential for outside-world connectivity is the key to expanding AoIP opportunities for broadcast, TV, audio and video post and music recording facilities, TV trucks, stadiums and other venues. Facilities can be geographically apart, yet bound together virtually. Or be closer, such as a mobile production truck parked outside of a stadium.

AES67 was developed for what’s considered high-performance audio, that is, at least 16 bit/44.1 kHz full-bandwidth digital audio with a low latency, less than 10 msec. The standard covers synchronization, media clocks, transport, quality of service, encoding and streaming and session description.

“The interoperability of AES67 is based on 10 years of industry experience,” Shay said. For AES67, no new protocols were devised. It is based on existing protocols and standards from organizations such as AES and IEEE, and on the Open Systems Interconnect (OSI) model of network architecture using abstract or virtual layers.

The layer structure needs some explanation to understand how AES67 AoIP compares with Ethernet-based technologies. Andreas Hildebrand, senior product manager at ALC NetworX GmbH, provided a brief overview at the DTV Audio Forum.

Layer 1, the lowest layer, is the physical layer, which can be a copper or fiber connection over which data is transferred between a transmitting device and receiving device. (Radio-link connections are covered by this layer.)

Layer 2 is the data link layer, the layer used by Ethernet-based technologies like AVB (Audio Video Bridging) and TDM (time division multiplexing).

AES67 uses Layer 3 for AoIP. This is the network layer, with an IP-based suite of protocols for packet forwarding between networks (including the internet). According to the standard, Layer 3 “is responsible for packet forwarding and routing of variable-length data sequences from a source to a destination.”

AES67 also uses Layer 4, the transport layer, which, according to the standard, “provides end-to-end communications between devices on a network. The layer handles issues of packet loss and reordering and implements multiplexing so that a single network connection can serve multiple applications on the end station.” The standard calls for realtime transport protocol (RTP) for AoIP.

There are three other OSI Layers—5, 6 and 7—but they aren’t relevant here.

Each layer is “stacked” on top of the next lower layer, with data passing down in a specific way from one layer to the next lower one. But devices on a given layer can virtually communicate with each other.

As Hildebrand explained, Layer 1 systems tend to be proprietary, mostly based on fiber or copper connections, using point-to-point or chain topologies. Switches tend to be custom- built, devices tend to be fixed, with limited channel capacity and support for selected media formats. Because of their proprietary nature, these systems are typically based on a single manufacturer’s products, but for the same reason are generally quite rugged and have very low latency.

Layer 2 systems also tend to be proprietary. These are Ethernet-based, with channel capacity determined by the bandwidth of the Ethernet channel. Devices are connected in a local area network, but “you can’t cross out of the LAN,” Hildebrand said. AVB systems fall into this category. “You need to have an AVB switch,” Hildebrand said. “If you stay inside the AVB cloud, all data is preserved by built-in quality of service mechanisms. [But] none of the AVB streams can pass across a non-AVB bridge to another cloud.”

For Layer 3 systems, Hildebrand said that although there are proprietary systems, most are based on standardized Internet protocols supported by standard switches. This allows networks to connect to other networks. In this layer, “the size of the network is not limited,” Hildebrand said. It is “flexible, scalable, with a flexible choice of media format. Latency varies depending on the size of the network and the utlized payload format.”

As mentioned, AES67 was developed to have low latency, less than 10 msec. Clocking and synchronization, packet size and other encoding and streaming criteria all play a role. These will be discussed next time.

Mary C. Gruszka is a systems design engineer and consultant based in New York. She can be reached via TV Technology.