Telephony: Time to Synch Up

Independent thinking is a big part of the cable telecom mentality. For the most part, the pioneers who started our business didn’t fit into corporate molds. Most technicians I know would die in a cookie-cutter job. Even our technology is our own special breed, from the way we modified Ethernet to fit our distribution plant through telephony implemented via PacketCable. But in the world of "any application, any device," there’s one dimension where we need to look like everyone else, and that dimension is time. Telecom timing The rest of the telecom world runs on a tightly synchronized time standard, specifically Universal Coordinated Time (UTC). Time implies clocks, and the clocks that determine UTC are devices that monitor the rate that specific radioactive isotopes decay (spit off particles) under controlled environmental conditions. The simplest way to explain UTC is that it is a running average of the most accurate atomic clocks in the world, adjusted with leap seconds to align with one rotation of the Earth, within one second per day. Most atomic clocks are built with isotopes of the element Cesium.

In telecommunications, synchronization is the operation of digital switching and transmission systems at a common clock rate. It comes in two flavors: frequency synchronization and phase synchronization. Frequency synchronization refers to operating two network elements at the same bit rate or frequency. Phase synchronization is the alignment of bits (pulses) such that the beginning and end of a byte (set of bits) can be identified. Without synchronization, repetition or deletion of blocks of bits may occur as information moves between systems, creating what is known as "slips." Depending upon the application and severity of the slip, the result may range from an annoying click in a voice conversation to the "blue screen of death" in a video transmission.

UTC with its underlying atomic clocks provides a basis for the most accurate synchronization among devices in a telecommunications system. The tie between UTC and telecom began, like most digital concepts, with voice telephony. In the digital circuit-switched hierarchy, physically separated Class 5 end office switches needed to be synchronized with each other and with Class 4 toll (long distance) digital switches. UTC did not yet exist, so the Bell System synchronized all its switches to its own reference atomic clock, called the Bell System Reference Frequency (BSRF), in Hillsboro, MO. The Hillsboro clock became known as the Stratum One source.

When the Bell System began offering Digital Data Service (DDS), the network elements within DDS also connected to the BSRF, but via paths that were different from those used by digital telephony switches. As data and enterprise networks evolved, they as well linked to the BSRF, resulting in a complex and difficult-to-administer synchronization network. Standards saved the day, when Bellcore defined a synchronization network that did not require physical connectivity to a single Stratum One source, but rather traceability to a clock that operated within the precision of the original BSRF. Today, there are several sources of Stratum One, including the satellites comprising the Global Positioning System (GPS). These Stratum One clocks are the devices mentioned earlier that determine UTC. They are also known as Primary Reference Sources (PRSs). Impact on cable Until recently, all this had little impact on cable telecommunications. When cable provided only analog video distribution, there was little need for precision synchronization with other networks. Data and local telephony introduced the need for billing timestamps tied to an external reference, and although these timestamps do not require the precision of Stratum One traceability, they follow UTC within a 200 msec variation via the Network Timing Protocol (NTP). An NTP server is specified as a requirement in PacketCable 1.5.

Changing technology and new markets have further increased cable’s synchronization requirements. Next Generation Network Architecture (NGNA) physically separates the media access control (MAC) and physical layer (PHY) levels of the cable modem termination system (CMTS) to allow quadrature amplitude modulation (QAM) sharing. This means that components that had previously communicated via the CMTS backplane can now be a substantial distance apart. Backward compatibility dictates that outboarded QAM devices and CMTS keep the same time relationship as they do on integrated models, which requires synchronization of QAM and CMTS core clocks within 5 ns. A new, highly precise, timing server with associated clients in the separated CMTS components is the answer. It is a requirement in the DOCSIS Timing Interface (DTI) specification for all systems with a modular CMTS (M-CMTS). (Just in case you can’t relate well to measuring 5 ns, it’s the time it takes for light to travel about 50 feet).

When M-CMTS components are not co-located, GPS traceability is a DTI requirement. Since GPS is a Stratum One source, this automatically implies Statum One traceability is available to cable systems with M-CMTS and next generation QAM devices at locations separate from the M-CMTS, such as a hub. GPS traceability is added to a DTI server via connection to a GPS receiver and associated antenna.

Even without M-CMTS, however, business network services, Internet protocol (IP) networks and network convergence drive a need for Stratum One traceability in cable networks. When business services include circuit emulation to transport time division multiplexing (TDM) services such as T-1 across a packet network, the slips resulting from insufficient synchronization will cause error performance to degrade to unacceptable levels. As network convergence drives applications across carrier boundaries, Stratum One traceability is vital. In order to meet interface standards, all digital signals between carriers must be under control of a clock or clocks traceable to a Stratum One source. A DTI server with GPS can provide the required link.

Much of the work behind the DTI was contributed by Symmetricom, a San Jose, CA, company that has been active in the design, development and delivery of network synchronization systems since 1985. In addition to working the specification, Symmetricom has provided a reference design for DTI clients and has implemented the server in its TimeCreator 1000 product, which is the only implementation of the specification available as of this writing.

For those interested in more information on synchronization and the associated requirements of communications networks, I highly recommend Symmetricom’s free distance learning curriculum (22 courses!) at www.syncuniversity.org. (Also see the synchronization feature in this issue.) Promising market Before closing, I’d like to shift gears and note an emerging new market being trialed by switch vendor Cedar Point.

In late February, that company announced the second trial of the Safari C3 Multimedia Switching System as a vehicle for communications on a university campus. In the associated press release, the company touted Safari’s ability to interface with an IP multimedia subsystem (IMS) infrastructure and to leverage existing equipment for new services, such as video telephony and fixed-mobile applications.

Cedar Point personnel carefully avoid the use of the word "PBX," but this is a premises-based application. It will be interesting to watch what follows and consider the possible implications for cable’s business services market. Justin J. Junkus is president of KnowledgeLink and telephony editor for Communications Technology. Reach him at [email protected].