Video over IP is the delivery of video and audio (usually compressed) over packetized networks using Internet Protocol (IP). (See Figure 1.) Video over IP is directly tied to the history of the delivery of video and audio over wide area networks (WAN). While it may seem to be a new development, it really is a logical extension of what has come before.
ATM over SONET networks
Years ago, AT&T was the telephone company. It provided both local and long-distance voice service. In the 1950s, AT&T built special, nationwide networks for delivery of programming from the big three networks (ABC, CBS and NBC) to affiliates all over the country. This distribution network was a combination of balanced coaxial cable and point-to-point microwave, and it incorporated a lot of purpose-built video equipment.
For many years, this network operated alongside the national voice network. All major cities (and some smaller cities) had dedicated video departments.
In the 1970s, AT&T broke up. The Baby Bells were created to handle local and regional telecommunications, while AT&T handled long-haul communications. In the '80s, phone companies began to switch from wire to fiber. Fiber allowed the phone companies to increase capacity, while reducing costs. They deployed synchronous optical network (SONET) technology as the base layer for their voice and data network, and this technology is still in use today.
During the same period, analog voice circuits were replaced with digital voice transmission, which allowed the telephone companies to multiplex many voice circuits onto a single copper pair and, later, onto a single fiber-optic cable. In the space of a few years, the core telecommunications infrastructure changed from copper/analog to fiber/digital.
The telephone companies decided to use a technology known as asynchronous transfer mode (ATM) to package digital voice data for transmission and switching. Among other things, ATM defines the rules about how data packets are constructed and how connections are established.
As the computer came of age, telecommunications companies began to move significant amounts of data. Data fit well into the ATM over SONET protocol stack. And as computing technology evolved, IP became the dominant networking technology. The telecom industry dealt with this by encapsulating IP into ATM and then transmitting it over SONET networks.
In the meantime, video also transitioned to ATM over SONET for most long-haul applications, but the last-mile delivery of video from the telco central office to the studio could be anything — ATM, dedicated fiber-optic cables or old balanced coax.
IP over SONET networks
About five years ago, the telecom industry made the momentous decision to change its core networks from ATM over SONET to IP over SONET. The telcos realized that virtually all of their data traffic was IP-based and that there were significant savings to be realized in converting voice traffic to IP. Once this decision was made, the transition from ATM to IP seemed to happen overnight.
This rapid transition left video in a predicament. Many large broadcasters had national ATM networks in place, connecting key cities such as Atlanta, Los Angeles and New York.
Broadcasters knew that it was not an option to keep their ATM over SONET networks forever. Once the telcos made the decision to go with IP, it was only a matter of time before broadcasters had to follow. Broadcasters faced the difficult and costly question of when to replace their ATM over SONET networks with IP over SONET networks.
Over the years, many of the video specialists retired from the telephone companies — and they were not replaced. With a few notable exceptions, it was almost impossible for a broadcaster to reach someone at the telephone company who knew much about video. As the telecoms moved to IP over SONET, they spent a great deal of effort training technical people to manage these networks. Their research laboratories have conducted some of the most cutting-edge analyses of delivery of video, voice and data over IP networks to the home. If broadcasters wanted to take advantage of this knowledge, they would have to convert to IP over SONET — and sooner rather than later.
It is only natural that video over IP is becoming a major transport technology in our industry. History and changes in technology have conspired to make it the natural choice for delivery of video, whether for contribution or delivery to the home.
Video over IP can be split into two domains: contribution and delivery to the home. In the contribution domain, delivery of video over packetized networks is typically referred to as video over IP, while video to the home is typically referred to as IPTV. By contribution, I mean video and audio that is sent from a remote venue back to a television studio or post-production facility. Delivery of video to the home typically is part of the video, voice and data triple-play service.
User requirements for contribution video are quite strict. For example, users may require low latency for live interviews (100ms), high uptime requirements (99.999 percent uptime is not unusual), and video and audio quality requirements that essentially require bit-perfect delivery of content from one end to the other.
Another requirement is that the cost of this new digital IP-based service must be in line with traditional video delivery, which has been provided to broadcasters since the 1950s.
While the user requirements seem reasonable from a broadcaster's point of view, they pose some real challenges for video service providers. An example of a conflict in user requirements is a broadcaster's requirement for high uptimes. To achieve anything near 99.999 percent availability, service providers have to address two core issues: maintenance and momentary outages.
Moving to IP allows broadcasters to stay in the well-maintained core of the telecom infrastructure. However, the concept of what constitutes an outage in the data world is significantly different from that of the broadcaster's.
Given that most data customers are not significantly affected by occasional outages, telecom maintenance personnel and IP network equipment vendors design maintenance plans and equipment specifications so that occasional disruptions occur. As an example, if a core router fails, the network will reconverge on another switch in a matter of just a few seconds. This is a good thing for broadcasters — it means that the network is self-healing. However, when the failed router is repaired and returned to service, frequently the network takes another hit as the network reconverges on the repaired router, causing two outages instead of one. This is just one example of how maintenance on data networks can cause problems for a broadcaster.
To deal with momentary outages caused by dirt in connectors or impulse noise, most video service providers use forward error correction (FEC). This scheme adds additional bits of information in the transmitted stream. If data is lost, the extra FEC information allows the receiver to reconstruct the transmitted data.
However, FEC adds overhead to the stream — perhaps as much as 20 percent in a typical application. It also adds latency. The exact amount of overhead and latency is determined by the severity of errors the FEC is designed to handle. Equipment designers work carefully to balance the typical distribution of errors on a network with various parameters in a given FEC scheme to come up with a well-designed system that can recover from most network errors, while incurring a minimum of overhead and latency.
Video over IP is destined to become the predominant technology for the transport of professional video over WANs. Broadcasters are taking advantage of core telecom IP networks for the transport of video over IP. However, challenges exist when broadcasters' user requirements produce conflicting design criteria.
Brad Gilmer is president of Gilmer & Associates, executive director of the Video Services Forum and executive director of the AAF Association.
Send questions and comments to:firstname.lastname@example.org