A discussion with Larry Marshall of Lightbit


Larry Marshall is CEO of Lightbit. He has more than 16 years of experience as inventor, manager, and entrepreneur in the photonics industry; Lightbit is his third startup. Marshall has a Ph.D. from the Commonwealth Center of Excellence in Australia and has been awarded 11 patents.

WDM: Lightbit has released its first product, an optical 2R [reamplify, reshape] regenerator, which would eliminate OEO conversion. Given the current state of the market, are carriers interested and how difficult is it to implement?

Marshall: Lightbit has always pursued a strategy of marketing directly to carriers, the end user, as well as system OEMs, our direct customer, to answer precisely that question. Carriers are always interested in products and technologies that can lower cost and reduce operational complexity. OEO conversions are expensive because they require a full 3R [reamplify, reshape, retime] transponder for each and every channel, while broadband—or unchannelized—optical solutions divide cost over many channels.

We have crafted a solution that helps carriers and OEMs deploy high-speed WDM networks in a more cost-effective manner. Extending optical reach and reducing cost is very attractive to our customers. Because of its optical transparency, implementation of our optical 2R regenerator is similar to implementation of an EDFA. Now, it's obvious that carrier expenditures are way down, and probably will continue to be down through this year and into next; however, Internet and data traffic continues to grow, and several carriers are currently evaluating their next-generation 10-Gbit/s solution. This solution will, of necessity, be dramatically more cost-effective than what is in the ground today. By this time next year, currently deployed system shelves will be filled with line-cards, and carriers will start deploying new systems, hopefully with our product in them.

WDM: Is 2R adequate and why not 3R?

Marshall: Of course 3R regeneration is better than 2R; however, it is extremely expensive and bulky, both in the electronic and the optical domain. It has also been shown that 2R regeneration allows 10-Gbit/s data transmissions to go farther than any link in the North American network. So, it just becomes a question of economics: What can you build and at what cost, to enable a cost-effective DWDM network?

Our 2R regenerator is far less expensive than any 3R solution because it can be accomplished across many channels at once, in a single unit, without demuxing. Retiming, however, must be accomplished on an individual channel basis because the channels are not synchronized. It is this channel-by-channel 3R requirement—both for optical and electrical 3R solutions—that makes it so expensive. In short, why build a Concorde airliner when a 767 gets the job done at a fraction the cost?

WDM: What's the fundamental technical approach you're using and how does it work?

Marshall: We exclusively licensed the core technology from Stanford University, then added several our own inventions to make it suitable for telecom. Fundamentally, we use a wafer-scale lithographic process to selectively alter the material structure in order to create "designer" active optical properties in a magnesium oxide-doped lithium niobate wafer.

Each waveguide is quasi-phase-matched to a particular pump wavelength, which can be tuned in a number of ways. Based on the particular lithography of a given chip or waveguide (and we create 72 waveguides on a single 5-mm wide chip), we can achieve a number of optical functions. We can also create Fourier synthetic gratings that allow multiple resonance points, or multiple nonlinear interactions simultaneously in the same waveguide. By using a buried waveguide process, we drive up the efficiency of the nonlinear interactions by an order of magnitude, so that even multiple, cascaded, nonlinear processes remain quite efficient. This enables quite complex purely optical functionality to be integrated into the chip.

The processed chip is energized and controlled by the optical pump signal, and the device operates rather like an optical AND gate. The pump interacts with any number of individual channels, yielding true broadband, multiwavelength operations. A single pump interacts with each input channel individually to achieve the desired function. The interactions we can implement enable broadband, multichannel frequency shifting, frequency doubling, phase conjugation, and amplification. System applications of these functions include 2R regeneration, wavelength or band conversion, and ultra-low-noise amplification.

WDM: What is the impact on other optical components and subsystems, like amplifiers, DCMs, receivers?

Marshall: This is another example of why we talked to carriers. As I said, 2R regeneration gets you across pretty much any terrestrial link in North America. However, your market size increases dramatically when you find a way to utilize technology that improves OSNR [optical signal-to-noise ratio] in shorter links. Improved OSNR on all the channels in a DWDM link enables system OEMs to relax a lot of the requirements on their other components.

For instance, by implementing 2R regeneration, carriers eliminate the need for DCMs [dispersion-compensating modules] and mid-stage access EDFAs at every EDFA site. This not only decreases amplification and dispersion management cost, but also takes out 8 to 10 dB of loss from every EDFA, further improving system OSNR. On the transceiver/receiver side, system OEMs can potentially relax performance requirements, allowing them to use cheaper line cards, and extend their reach with 2R regeneration to maintain the same overall system quality. Using mid-reach transmitters on long-haul links gives substantial line card savings that are multiplied by the number of channels deployed.

WDM: Are you focused on 2.5- or 10-Gbit/s transmission? Is there a channel limit or preferred number?

Marshall: Our technology is bit-rate and protocol independent. That said, our optical 2R regenerator is currently targeted at the next round of 10G system deployments. It's been tested at 40 Gbit/s, and the optical processor chips have been tested up to 160 Gbit/s, so we are confident of full transparency and upgradability. There is no channel limit per se; it depends on other optical parameters, but so far we have run 20 channels simultaneously in one waveguide in one device with the same performance as a single channel.

WDM: What other materials look interesting to you, and what other applications or products?

Marshall: Our team has done an outstanding job of developing a platform technology with multiple applications. As a nonlinear optics guy I've always been fascinated with the ability to engineer designer materials, as this opens up a wealth of opportunities. But as a startup CEO, I'm 110% focused on bringing one product to market. Our wafer-scale waveguide and poling process can be transferred to several other nonlinear materials.

One carrier is interested in our all-optical wavelength-band converter, which enables banded network architectures. We have also had interest in our ultra-low-noise optical amplifier. At this point we have only supplied bare chips for those applications. We have also supplied optical processor chips for several nontelecom applications, including biotech.

W. Conard Holton
Editor in Chief

More in DWDM & ROADM