OFDM promises 100-Gbps alternative
Orthogonal frequency-division multiplexing (OFDM) is a common format for wireless and copper-based communications. So why isn't it used in optical communications? If some companies have their way, it soon will be.
Orthogonal frequency-division multiplexing (OFDM) isn't new; wireless and copper networks use it routinely. It's not even new to optical communications, as those who remember such bubble-era startups as Kestrel Solutions and Centerpoint Broadband Technologies will attest. What is new, say proponents, is that the need to find an alternative to conventional on-off keying (OOK) in the face of 100-Gbps transmission requirements, when paired with advances in digital signal-processing technology, creates a happy confluence of necessity and capabilities that means optical OFDM's time is about to arrive.
OFDM involves encoding a communications stream into multiple subcarriers. In the optical domain, a modulator will take parallel data channels and use inverse fast Fourier transform to create the subcarriers, with some form of quadrature amplitude modulation (QAM) often used to set the phase and amplitude of each subcarrier. These subcarriers can be transmitted within the C-band in a manner compatible with WDM. When received at the other end, the application of fast Fourier transform enables the separation of the subcarriers into their constituent parts.
Splitting the waveform into multiple subcarriers reduces the amount of signal information each subcarrier contains. Since the damage from signal impairments such as chromatic dispersion increases with bit rate, a 100-Gbps signal broken into subcarriers will likely prove more resistant to impairment than a single stream.
In addition, OFDM offers very high spectral efficiency, with a greater percentage of power in the center of the waveform than conventional techniques. This attribute enables OFDM signals to pass through cascaded filters‚ "such as those found in a string of reconfigurable optical add/drop multiplexers‚" more cleanly than conventional signals as well, proponents assert.
Add spectral efficiency to robustness in the face of impairments and you get longer reach and reduced network costs, advocates claim. This, of course, was the value proposition that companies such as Kestrel and Centerpoint offered in the early part of this decade--unsuccessfully, it turned out. However, OpVista (www.opvista.com) currently uses subcarriers in its Dense Multi-Carrier technology for 40- and 100-Gbps applications; although it hasn't touted OFDM as a cornerstone of this approach, OpVista's technology would at least appear to be similar.
Many, many subcarriers
Kestrel and Centerpoint used a relatively small number of subcarriers -- less than 10. Today's OFDM champions, as Australian startup Ofidium (www.ofidium.com) exemplifies, have more ambitious schemes in mind for 100 Gbps.
“The power of OFDM technology is only realized to its full extent if one uses very many subcarriers," asserts Ofidium CEO Jonathan Lacey, who adds that “very many" probably means hundreds.
According to Lacey, the idea behind the company came when cofounder Arthur Lacey was challenged by a colleague at Monash University. "This technology is so compelling and so dominant in these other transmission media--why is it the optics guys don't use it in optics?'" Lacey quotes.
The answer, he admits, lies in the fact that Ofidium's approach--dual-polarized OFDM with coherent detection--is nearly as complex as the dual-polarized quadrature phase-shift keying (DP-QPSK) with coherent detection that the Optical Internetworking Forum (OIF; www.oiforum.org) has dubbed the de facto industry approach to 100 Gbps. Ofidium's optical OFDM, in fact, has much in common with DP-QPSK.
“We will use the same optical and optoelectronic components as our competitors will use," Lacey says. “We'll both use the components that are being standardized by the Optical Internetworking Forum."
Optical OFDM, like DP-QPSK, will also require conversion from analog to digital and back again at rates that will stretch the current state of the art. However, this point is where the difference between “nearly as complex" and “just as complex" resides.
“We do not require a converter to operate at as high speeds as the competing technologies do by a factor of about 1.5," Lacey asserts. “We still require very high-speed converters but not as fast as the competing technology requires."
Thus, while those working on DP-QPSK await commercial availability of converters with the necessary sampling rates, Ofidium has already found a supplier. The company announced last March that it will use the VEGA signal converter line from German company Micram (www.micram.com) in its products, which will take the form of transponders.
The use of off-the-shelf components such as the Micram converters represents a cornerstone of Ofidium's approach. “Our key contribution is really in the architecture of the module and in the digital signal-processing algorithms that we apply to create the OFDM signal and detect it at the far end. We want to as much as possible use off-the-shelf components--both electronic components and optical components--to do the rest," says Lacey.
Those products are still about two years away, Lacey admits. The company currently is engaged in what Lacey termed “proof of engineering" work that will demonstrate that not only do the physics work but that Ofidium can implement the approach using current electrical and optical components. Lacey expects the current effort to stretch into next year, with prototype products available in 2011.
Right on time?
According to Roy Rubenstein, analyst at module market research firm LightCounting (www.lightcounting.com), the 100-Gbps transponder market probably won't begin to take off until 2012. This would appear to make Ofidium's timing felicitous. However, Rubenstein isn't quite so sure, given the weight currently being put behind DP-QPSK.
“It will stand more of a chance or it will become more important as speeds go up even above 100 Gbits," he says of OFDM. “It might be a candidate for a second generation--just like there's more than one generation now for 40 gig--it might make an appearance after this first wave [of OIF-influenced 100-Gbps technology]."
The other question, of course, is how much can Ofidium do on its own to promote interest in OFDM. “You need someone else to run with this," Rubenstein asserts. “Either Ofidium gets acquired by a serious system vendor or a serious system vendor itself starts pursuing this. And they're all doing this in the labs, but they're looking much further out.
“But it's a question of when this starts to appear on a platform, whether it's a partnership with a serious vendor or whether it's a serious vendor themselves have developed this--that's when you start to think OFDM is interesting and has a future. I think it does, but I just think it's very, very early days yet," Rubenstein concludes.
Stephen Hardy is editorial director and associate publisher of Lightwave.
Orange Labs investigates OFDM for access networks
While it may be natural to associate OFDM and high speeds with long-haul networking, OFDM may find use in other applications. France Telecom's Orange Labs, for example, has looked at applying OFDM to next-generation access networks.
According to Naveena Genay, a research engineer at Orange Labs, the appeal of OFDM in access networks is similar to the long haul (another OFDM area of study for Orange Labs). In terms of resistance to dispersion, she says, “For the access network, it means you can increase the reach beyond 20 or 60 km. And, also, you can increase the bit rate and be robust to chromatic dispersion."
OFDM's spectral efficiency also offers advantages to access applications, in terms of the bandwidth requirements of equipment components. “This means you can have low-cost components such as VCSELs and DFBs that aren't meant for 10-Gbps transmission," Genay says.
The IEEE and FSAN currently have standards for 10-Gbps PONs on the drawing board. However, Orange Labs has begun to investigate even greater speeds using OFDM. This includes 40 Gbps, an effort that will use optical components from 3S Photonics (www.3sphotonics.com).
The French company has OFDM transmitter and receiver components under development, says 3S Photonics' Emmanuel Gerard, program leader for pumps and lasers, and Pierre Wolkowicz, product line manager, lasers. The directly modulated laser sources have four primary requirements, they report: high output power, good relative intensity noise, wide bandwidth, and very good linearity. Appropriate figures include 8 to 10 dBm for power, -150 dB/Hz for noise, 10 GHz of bandwidth, and -60 dBm in terms of intermodulation products measured by IMD3.
The receivers clearly must follow similar parameters. 3S Photonics has begun shipping sample lasers; sample receivers should be available this September.
The work at Orange Labs has already demonstrated some of OFDM's potential, according to Genay. For example, OFDM showed better performance than NRZ in experiments using remote modulation and reflective modulators with an eye toward colorless ONTs. “We have seen that in a bidirectional link with remote modulation, OFDM is more robust to Rayleigh backscattering in the fiber," Genay reports. Orange Labs also has demonstrated 100-km transmission in a PON architecture using OFDM.
Orange Labs and 3S Photonics will also collaborate on a PON project called EPOD. Funded by the French National Research Agency, EPOD will examine the use of OFDM to extend PON reach to 200-300 km. Academic partners include LISIF and the University of Limoges/CNRS. The 24-month project launched this past February.
LIGHTWAVE:Expanding Capacity in the Metro Optical Network (Kestrel Solutions)
LIGHTWAVE: Subcarrier Multiplexing: More than Just Capacity (Centerpoint Broadband Technologies)
OFIDIUM:OFDM information resources