By STEPHEN HARDY
It's lonely out there on the frontier for the passive-optical-network (PON) pioneers, who have found that coming up with components for their groundbreaking systems is very much a do-it-yourself proposition. The cavalry, however, may be coming to the rescue.
Most PON approaches start with the specifications developed by the Full Service Access Network (FSAN) initiative, an international consortium of carriers supported by input from the vendor community and ratified by the International Telecommunication Union within the G.983.1 standard. An FSAN network comprises three major pieces of equipment: an optical line terminal (OLT) located in the carrier's central office or headend, an optical-network unit (ONU; sometimes this device is called an optical-network terminal, or ONT) that terminates the network, and a passive splitter located somewhere between the OLT and the network's ONUs.
The OLT broadcasts a master signal to the splitter, which divides it into as many as 32 duplicate streams for transmission to the ONUs. Each ONU receives the same incoming signal; electronics within each ONU sorts through the transmission for the information the carrier intended it to receive.
FSAN supports two-way communication. Since the ONUs must share the same passive splitter and transmission path from the splitter back to the OLT on the return leg, the specification provides for a time-division multiple-access scheme, whereby the transmitters within the ONUs operate in burst mode. The OLT sends a ranging cell to each ONU to determine the overall round-trip delay to aid in the assignment of time slots for each ONU. This information enables the carrier's engineers to assign time slots for upstream communication, based on the services the carrier is offering to customers. The OLT then separates the information from the different time slots for transmission back toward the network core. FSAN specifications call for a maximum of 622-Mbit/sec ATM-based transmission downstream and 155 Mbits/sec upstream on a pair of wavelengths-roughly 1,500 nm from OLT to ONU and 1,310 nm in reverse.
Common sense should dictate that with a well-defined set of specifications, piecing a product line together would be fairly straightforward. However, the burst-mode technology central to PONs is a new requirement for optical component manufacturers.
"Those burst-mode transmitters aren't the way SONET devices currently work today. [SONET transmitters are] continuous-they stay on, they don't shut off," says Jeff Masucci, founder and vice president of engineering at Quantum Bridge (Andover, MA), one of the new PON startups. "And the same on the receive side. A SONET receiver, for example-an OC-12, OC-48 receiver-it's expecting to see light coming in all the time and it's supposed to have synchronization there all the time. Here, in these types of architectures, the light coming in comes in for a period of time, then it stops coming in and another burst or packet of light comes in at a different amplitude and with a whole different clock structure essentially. They're out of phase with each other. And current off-the-shelf devices don't work that way."
The dilemma is even worse when companies look to extend capabilities beyond the FSAN specifications. For example, like Quantum Bridge, Terawave Communications (Hayward, CA) wanted to provide symmetrical 622-Mbit/sec capabilities.
"The 622-Mbit aspect of the PON required internal development of the optoelectronics; there is no commercial solution out there today," says Bob Deri, director, lightwave systems at Terawave. "Most of the solutions that you'll find on the market, either from Japanese companies in the form of integrated modules or Lucent in the form of chipsets, are only available for 155 Mbits. And while you may find road maps that have higher bit rates on them, I encourage you to probe more deeply into the reality behind the road maps to see what time scale the engineers really think they'll deliver on. Burst-mode technology is very difficult. It's been recognized as being difficult since the early 1980s in this business."
Most companies in the PON arena have developed in-house optoelectronic and ASIC expertise in the realization that if they want their components done right, they'll have to do it themselves. That said, PON manufacturers don't have to completely reinvent the wheel.
"The transmitters and receivers in some cases involve our own circuitry for burst mode, in which case we're really just buying the laser diode or the photodectector," explains Deri. "Whereas in the other direction of the PON, it is a continuously transmitting link, and so we can use commercial parts."
Quantum Bridge has focused much of its current in-house efforts on developing integrated components. "Your laser transmitter is not just a transmitter," says Masucci. "It's the drive circuitry around that; it's the temperature-compensation circuitry around that. And you pull that together into one module. It makes it much easier to manufacture and also reduces the cost. That also applies to the burst receiver-you have any type of amplifier as well as the optical receiver in one package, in one module."
Even when the right parts are available in the market, finding them can prove a time-consuming task.
"Not all lasers will work to our specifications," Masucci says. "Some will. It's typically not the laser that's of interest; it's typically the circuitry around it that drives it. And there's some filtering and other mechanisms that you use to allow PONs to work. If you looked at just the laser itself, there are many suppliers of those. You have to do a filtering on which suppliers offer the requirements that you're looking for. But then you need to do some work around making those particular lasers work the way you want them to."
With the PON space heating up, component manufacturers have begun to turn their attention toward easing the burdens now placed on systems houses. "The one special part in passive-optical-network technology that many folks are using for cost reductions is a so-called bidirectional module, also known as a duplexer. It basically integrates a photodetector, a laser diode, and very inexpensive WDM in a little package with a pigtail on it," Deri reveals. "These parts can make the system much less expensive than if you bought a WDM and laser transmitter and receiver."
Lucent Microelectronics Group (Allentown, PA) has stepped into the chip breach with a new offering directed at burst-mode PON applications. The chipset includes a clock and data recovery circuit, a laser driver with automatic power control, and another that provides automatic power adjustment in the receiver.
The devices are targeted at 155-Mbit/sec applications. Announced this past July, the chips are already a hit. Shane Gunning, strategic marketing senior manager for analog network and interface products at Lucent, reports that approximately 20 companies had either designed in or sampled the chips before they were announced; since the official debut, Lucent has received another 15 to 20 inquiries.
The company also has its eye on future requirements. "As this application gets deployed in the market, I can already see that instead of 155 [Mbits/sec] downstream, we're going to go to 622 or a gigabit or 2.5 gigabits. And then as the application is deployed to more and more businesses, they'll want symmetrical upstream and downstream communications, so we're going to have to do the same thing in a burst mode. We actually have prototype burst-mode ICs running all the way to 1.25 Gbit," Gunning reveals.
Even with such help on the way, Deri for one doesn't believe that he will no longer be designing ASICs, because his firm will switch to a completely off-the-shelf design approach. "If you want a cutting-edge product, you have to pay the price for that, one way or the other. If you contract the core of your technology out, you're setting yourself up to get burnt. If you own it, you control your destiny," he says. "What we don't do is try and reinvent technology that's already out there. In that case, sure, just go buy it. But for the kind of product that we envision for Terawave, that's going to be a cutting-edge product, and that means I'm employable here."
The requirements that have led startups such as Quantum Bridge and Terawave to look beyond the basic FSAN specifications have not gone unnoticed by the FSAN community. Terawave's Bob Deri reports that the FSAN Optical Access Network Working Group met earlier this year in Chiba, Japan to discuss adding dynamic bandwidth allocation and WDM capability to the FSAN specifications. Terawave has incorporated both capabilities into its current product plan and expects to offer its expertise to the group as it hammers out new specifications.
Codifying symmetrical 622-Mbit/sec transmission also has been discussed, but Deri reports that the other two capabilities will receive first priority. He expressed a hope that the group, chaired by NTT's Kenji Okada, might finish its work by the end of the year.
If you have an idea for a news story or would like to contribute to Lightwave, please contact editorial director Stephen Hardy at tel: (603) 891-9454 or e-mail: [email protected]. Ideas must focus on topics related to fiber optics and optoelectronics in the communications industry.