Greg Shay, chief science officer at
The Telos Alliance, speaks at the
2015 NAB Show. Photo by Davide Moro
LAS VEGAS — Audio over IP orAoIP is a hot topic in the radio broadcasting world. At the NAB Show various conferences delved into this subject. One such paper was that of Greg Shay, chief science officer at The Telos Alliance, who gave a presentation entitled “Moving Audio Processing from Boxes to the Cloud: AES67 & Audio-over-IP Applications.” Another was that of Kevin Campbell, APT sales director, who presented “The Audio Cloud: Decreasing Cost and Improving Reliability of Audio Links for Broadcast.”
AoIP dates back to the last decades of the last century. Radio, like any form of remote communication, began as an analog medium. Before being aired, signals flowed on solid copper wires in the shape of (standardized) waveforms. Beginning in the 1980s, technology advancement allowed digital electronic devices to become more affordable and reliable at the same time, thus opening the era of digital (mass) telecommunications.
It was the time of the so-called “synchronous” technologies. Actions or events that are measured against a time reference, or a clock signal, are referred to as synchronous.
Kevin Campbell, APT sales director,
discusses the audio cloud during the
2015 NAB Show. Photo by Davide Moro
Within any synchronous standard, the information is flowing at the same constant pace (bitrate), and all the involved devices must obey the same time constraints. Under that establishment, all the information elements reach the receiver in the same order the sender sends them, and there is no appreciable difference in the overall transit time scored by each of the information elements. They all also follow the same path.
The step ahead in digital telecommunications came with a wider availability of the so-called “asynchronous” technologies, in which actions are prompted as a response to another signal, typically not governed by a clock signal. Asynchronous devices use specific control signals to notify intent in an information exchange, and then all the devices that need to exchange information can do so at their own natural rate —fast or slow.
A common clock signal is no longer necessary; instead, the involved devices have to generate, detect and manage additional control signals, usually referred to as “handshaking” signals because of the similarity with two people approaching each other and shaking hands before they start talking.
Handshaking signals are generated by the devices themselves and can occur as needed. Furthermore, they do not require an outside supervisory controller such as a clock circuit. Within asynchronous standards, information is flowing across the concerned path with no mandatory time constraints.
Synchronous protocols can usually transfer information faster (per unit time) than asynchronous protocols, because synchronous signals do not require any extra negotiation as a prerequisite to data exchange. On the other hand, asynchronous protocols are generally much more flexible.
Packet switching is an asynchronous digital communication method that divides the information to be transmitted into suitably sized packets and routes the packets through a physical medium that is usually shared with other communication sessions. The information element within each packet is called payload. Each packet includes additional information (header) about the sender and the expected recipient.
The packets are routed individually, sometimes resulting in different paths and out-of-order delivery. Information in the header is used by networking hardware to direct the packet to its destination, where the payload is extracted from multiple packets, grouped according to the original sequence and delivered to the end user. The method of nesting the data payload in a packet with a header is called encapsulation.
Encapsulation of application data descending through the layers.
Image by Wikipedia user Cburnett, CC BY-SA 3.0, GFDL
Packet switching features delivery of variable bitrate data streams (sequences of packets) over a network, which allocates transmission resources as needed, on a best-effort basis. When traversing any network node, packets are buffered and queued, resulting in variable delay and throughput depending on the network’s capacity and the traffic load on the network. IP is a widely adopted asynchronous protocol based on the packet switching method.
Radio is basically a live, real time form of broadcasting. Broadcasters therefore expect that any device along the signal chain can introduce no error and no delay. When a delay is not avoidable (due to the laws of physics), the magnitude of this delay has to be both predictable and steady.
IP lays at the opposite of these expectations. According to its essence, IP is a best-effort standard, it introduces some extra delays (with respect to the bare laws of physics) and these delays vary (also) according to a number of unpredictable conditions.
At the same time however, the advantages to IP are numerous. IP features the utmost flexibility in any communication application; it is cost-effective and is considered the natural converging point of all the digital technologies and standards. The negative aspects of IP can be managed through specific communication architecture and network design.
Imagine one has to send a 10-page paper by the post and he or she fits the entire document into a single envelope and mails it. It’s simple and cheap, but there is a risk that the one envelope gets lost.
As an alternative, one could put each page of the document into a single envelope (packet) and post 10 separate envelopes. The various envelopes might not arrive in the correct order but if only a single envelope gets lost, it would be possible to nearly rebuild the entire document.
The latest developments in technology and network architecture are based on redundant streaming. If one prints a second copy of the document, and fits each page of that document into a dedicated envelope and proceeds to post the envelopes using a different service provider per envelope, it is unlikely that the different service providers will lose envelopes containing the same page.
Under the same error ratio (one envelope lost per carrier/service provider) it is possible to rebuild the original document at the final end of the chain at a price — extra bandwidth (extra paper, extra envelopes, extra postage).
At present, this extra bandwidth is no longer a major concern under currently available bandwidth in IP networks, both public and private ones.
IP audio applications today range from STL applications to remotes/outside broadcast, audio contribution, audio distribution and confidence monitoring. Major broadcast equipment manufacturers developed their proprietary systems and techniques for broadcast-grade IP audio delivery. The result was often a lack of interoperability between network sub-areas featuring different vendor technologies.
In 2013, the Audio Engineering Society released the AES67-2013 standard for “audio applications of networks — high-performance streaming audio-over-IP interoperability.”
AES67 aims at an effective interoperability between previously competing AoIP systems and long-term network interoperation between systems, in a way that various networked audio components from different manufacturers work together in a single system. The Media Networking Alliance was formed in October 2014 to promote adoption of AES67.
Davide Moro reports on the industry for Radio World from Bergamo, Italy.