SONET

In the November column, we gave an overview of ATM and briefly described how ATM uses synchronous optical network (SONET) as its transport layer. This month, we take a closer look at SONET and some of the telephone-industry nomenclature behind the technology.

You might wonder why broadcasters would care about or need to know a core telecommunications technology such as SONET. The answer is simple: convergence. We are now at the point where video, data and voice are commonly being carried over the same infrastructure. And SONET is an ubiquitous, standardized, high-speed network capable of delivering information at very high speeds.

Evolving jargon

Broadcasters frequently run into terms such as T-1, DS-3 and OC-48. Let's see how these terms evolved. When the telecommunications industry transitioned from analog to digital, it settled on a speed of 64kb/s for one voice channel. This defined the lowest level or base signal for digital transport. A trunk line (T-1 line) carries 24 voice channels and runs at 1.544Mb/s. (If you take 64 and multiply it by 24, you come up with 1.536, not 1.544. Extra bits are added to each channel for synchronization.)

Table 1. Various telecommunications circuits and associated bit rates.

AT&T long ago referred to the T-1 datastream as simply a datastream (DS). The resulting nomenclature is DS-n, where “n” indicates the number of datastreams. A DS-0 is equivalent to a 64kb/s voice channel. A DS-1 is the same as a T-1 and has a bit rate of 1.544Mb/s. SONET's base level, optical channel 1 (OC-1), is 51.84Mb/s and is equivalent to one DS-3. OC-n stands for optical channel “n.” STS-n stands for synchronous transport signal “n.” OC-n is the optical equivalent of the electrical transport provided by STS. Table 1 shows the various telecommunications circuits and their associated bit rates.

Currently, OC-192s is the highest speed of commonly deployed SONET networks. At 9.9Gb/s, an OC-192 can carry 129,024 voice circuits or more than 3000 3Mb/s, MPEG-2-compressed television signals. Only connections between the largest cities require this much bandwidth.

SONET is designed to be payload unaware — as an underlying transport technology it has no knowledge of what it is carrying. And any higher protocol layers such as ATM or IP cannot have any dependency on SONET technology per se. This is a classic implementation of the ISO 7 layer data model, where you may substitute layers without disrupting layers below.

Timing is everything

As you can see from Table 1, DS-n channels are asynchronous. DS-n data rates can vary up to 20 parts per million (a DS-3 at 44.736Mb/s can vary by as much as 1.7kb/s). This has several impacts. First, if you want to stack multiple DS-ns together, you have to add extra space for stuffing bits to deal with the variation in timing of each individual DS-n. Since telecom infrastructure is built upon combining tributary datastreams into larger transport pipes, this compounds and becomes significant wasted space.

SONET networks are synchronous and they control timing very tightly. SONET switches use a Stratum 1 atomic clock for their reference, and SONET includes a hierarchy that allows lower nodes to derive timing references from these highly accurate and stable clocks. This tight timing greatly reduces the need for stuffing bits (SONET uses pointers instead). Since DS-3s are commonly used as tributaries to SONET streams, SONET includes the concept of the virtual tributary (VT). VTs are synchronous versions of DS-3s that run at a constant 1.728Mb/s. In this way, asynchronous tributaries can contribute to synchronous SONET networks.

Additionally, tight timing allows switching in the optical domain. This is important because optical switching is faster, more reliable, less costly and results in less signal degradation than switching in the electrical domain. Without this capability, signals would have to go through an optical-to-electrical (O/E) conversion, be switched as required, and then go through an electrical-to-optical (E/O) conversion for onward transmission.

Like TV but different

Understanding how data are loaded into a SONET frame is easy for a television engineer. In a television analogy, each SONET STS-1 frame consists of a television “line” that is 90 bytes wide. Each television “field” comprises nine television lines. The computer scans the line from right to left, jumps down to the next line and does it again. At the beginning of each line is a “horizontal sync pulse” (transport overhead). STS-1 frames are sent at a rate of 8000fps. The math to get to 51.840Mb/s per SONET STS-1 frame is dead simple: 90 bytes per line × 9 lines × 8 bits per byte × 8000fps = 51.840Mb/s

Topologies and features

As you might expect, SONET supports a wide number of networking topologies, including point-to-point, star and ring. It also supports add-drop, a cable-television-industry term with which you may be familiar. Remember, SONET is doing all of this in the optical domain. This is important. Until the advent of SONET, it was not possible for telecommunications companies to switch or to provide handoffs to other carriers in the optical domain. An O/E and subsequent E/O conversion was required to go between boxes from different manufacturers. SONET enabled interoperation of vendors' equipment in the optical domain, allowing carriers to interconnect at lower cost and higher speeds.

SONET's add-drop feature is critical to telecommunications providers and may ultimately prove critical to broadcasters as well. Because SONET is a core transport technology, carriers must be able to add or delete even a single 1.5Mb/s VT-1 circuit from a 2.4Gb/s OC-48 feed (or any other combination, for that matter). Users want to be able to add and remove payload from the system at any point. The process by which SONET enables this is fairly simple, but it involves a whole lot of terminology that we cannot wade through here. Suffice it to say that SONET supports the ability to add or delete tributaries from a larger pipe.

SONET was designed from the ground up to provide “carrier-quality” transport, meaning that it must have very low error rates and elaborate alarming and restoration features. Broadcasters understand the concept of low error rate. But the concept of alarming and restoration may have slightly different connotation in the telecommunications world. Carriers have developed elaborate automated error-monitoring, alarming and restoration features on their networks. These systems constantly monitor transport circuits for errors. When errors on one circuit exceed a preset threshold, the system switches to a backup path and generates an alarm. When the equipment is repaired, the system automatically switches it back into service. Note that it is difficult to equate errors in a SONET network (typically expressed as bit errors) to video error measurements such as EDH errors. Also, because MPEG has varying responses to errors depending on where in the MPEG bitstream the error occurs, it is difficult to correlate errors in an underlying SONET network with outages experienced on a video feed. But broadcasters should know that video ultimately transmitted over SONET is carefully monitored and that elaborate alarm technology is present on all major SONET circuits.

As convergence becomes a reality, it is important for broadcasters to acquaint themselves with a growing body of knowledge and terminology from the telecommunications and IT world. SONET is a key underpinning of almost all terrestrial WAN video transport.

Brad Gilmer is president of Gilmer & Associates, executive director of the Video Services Forum, and executive director of the AAF Association.