Kodeos Gets Started With $12M
In fact, the money -- which comes in equal amounts from Highland Capital Partners and Jerusalem Venture Partners -- was secured back in January, but the startup waited until now to announce it, says Kodeos's founder and COO Gadi Lenz. Presumably, the startup wanted to put off the dread moment when it had to talk publicly about its plans.
Even now that it is talking, Kodeos is sticking to generalities. According to Lenz, the startup is developing optical transponders, initially for OC192 (10 Gbit/s) line speeds, and later for OC768 (40 Gbit/s). These products will be designed for long-haul 1550-nanometer systems, to improve the reach, lower the bit error rates, and increase the overall capacity of the system.
To attain these goals, Kodeos says it will tackle some of the effects that degrade an optical pulse as it travels down a fiber -- things like chromatic dispersion, polarization mode dispersion (PMD), and optical cross-talk, which occurs in systems with multiple wavelengths.
How will it do this? "We will apply signal processing techniques before launching the light into the fiber," says Lenz. In other words, the transponders will include logic circuits that tweak the electrical signal before it's turned into light and sent out onto the fiber. He won't say what signal processing techniques will be deployed, because that's where the company's key intellectual property lies.
Kodeos's VCs are equally tight-lipped about the technology, but relax a bit when it comes to talking about the startup's founders. "[Gadi Lenz and Jason Stark] were star performers at Bell Labs and are both experts in their fields," notes Glen Schwaber, general partner at JVP.
Lenz worked on a range of projects at Bell Labs, including integration using silicon optical benches, ultrafast optics, erbium-doped fiber, and non-linear fiber Bragg gratings. He worked down the hall from Stark, who headed up the advance modulation and coding project.
Interestingly, Stark was co-author of a recent paper in the journal Nature, which considered and attempted to calculate the effect of so-called "non-linearities" on the amount of data that could be sent down an optical fiber (see Physicists Find Fiber's Limit).
The paper points to a couple of ways in which it may be possible to increase the information carrying capacity of optical fiber. In particular, it suggests using multilevel signaling -- encoding the data in such a way that more than one bit of data is delivered per clock tick -- to reach very high capacities over long distances. Another possibility is frequency modulation.
Kodeos won't confirm or deny that it could be exploiting these ideas, leaving much to the imagination. But given Stark's expertise in advanced modulation schemes, we wouldn't discount them entirely. Indeed, Lenz says that the company's been receiving calls asking if it has anything in common with Kestrel Solutions Inc., a startup that's promoting frequency modulation (see Kestrel Quietly Reconfigures).
Kodeos plans to have product prototypes ready by the end of this year and has enough cash to last until mid-2002, by which time it expects to be shipping samples. It currently numbers 15 people.
— Pauline Rigby, Senior Editor, Light Reading