First there was voice over copper. Then there was data over copper in the form of DSL. Now, finally, theres voice and data over copper: Its called Voice over DSL (VoDSL).
Simply put, VoDSL is the packetizing of voice via a DSL line, and, thus, a copper wire. The packets can be either of the IP or TCP/IP variety (VoIP) or, more commonly, the 53-byte ATM cells (VoATM).
Why the hubbub? Its the all-important copper lines, which are already deployed just about everywhere—especially to the small- to midsize-business segment. A single copper line can be leveraged to deploy data and multiple voice circuits simultaneously. VoDSL is less compelling in the residential market where users may have only one or two phone lines, which doesnt justify the cost of the equipment. And typical ADSL lines can satisfy customers with only one phone line requirement.
VoDSL allows deployment of converged services at significantly less cost than a channelized T1 line, or the typical scenario in which data is provided over DSL with extra lines for voice. Another bonus is that while a T1 line must dedicate each of its 24 channels to either voice or data, VoDSL is dynamic—so the bandwidth can be used to its full extent whether the packets contain data or voice.
ATM Now, IP Later While most VoDSL applications are currently ATM-based, many vendors and analysts believe that IP will become the choice method of packetizing voice in approximately five years. The holdup is quality of service (QoS), a guaranteed throughput and delivery level.
VoIP, at least today, cant reliably deliver carrier-class service, the kind of quality we expect when we talk over the phone. IPs QoS standard, Multiprotocol Label Switching, has not been defined fully and theres no widespread implementation. On the other hand, ATM has well-established and standard QoS mechanisms.
One of the most significant ways ATM outshines IP in delivering real-time service such as voice, says Greg Wetzel of the VoDSL Working Group, is simply that IPs packets are large and inconsistently sized. That causes a delay of up to 30ms per packet on a 384Kbps line, because many packets must be assembled and then disassembled before being translated into an understandable voice call. The variation in delay time it takes to assemble and disassemble the inconsistently sized packets is known as "jitter." ATM reduces jitter because its packet size is a constant 53 bytes.
Squeeze the Calls A typical uncompressed, packetized voice call uses 64Kbps of bandwidth. That works out to a maximum of six voice circuits on a single 384Kbps line. Currently, adaptive differential pulse-code modulation (ADPCM), the standard voice-compression technology, cant deliver carrier-class voice quality at bit rates lower than 32Kbps, but Wetzel believes that compression technology will improve, and that carrier-class quality will be attainable at 16Kbps or even 8Kbps in the next few years.
Wake-Up Call So far, theres only one national service provider, mPower (www.mpower.com), deploying VoDSL. Several regional companies are conducting tests, and the incumbent local exchange carriers to date have been slow to adopt the technology. A Qwest representative would state only that the company is "conducting trials."
Expect that to change. Vendors and service providers are predicting that by the end of the year, VoDSL will take off. But TeleChoice Inc.s Adam Guglielmo says that a major carrier is going to have to sign on before VoDSL gains acceptance in the marketplace.
While most vendors believe that VoDSL is an interim technology, smart service providers could make a tidy sum over the next few years while VoIP gets its QoS issues resolved. VoDSL can squeeze more circuits out of a single copper wire at lower costs than analog technology. For that reason alone, its worth looking into.