MIT researchers cram optical circuitry on a silicon chip
By Darren Murph 
posted Feb 14th 2007 1:13AM
posted Feb 14th 2007 1:13AM
It looks like MIT is raising the bar yet again, as this time it's taking a break from crafting autonomous UAVs and stackable vehicles to cram optical circuitry on your everyday silicon chip. In an effort to "integrate the optical circuitry with electronic circuitry" on the same silicon wafer, researchers have devised a method which will harness the "enormous power of light waves in networks" while offering up a way to manufacture the circuitry cheaply. The crew has reportedly already been playing around with a working prototype, and suggests that it could eventually "redefine how optical networks are built." Moreover, the development addresses the existing "signal weakening over distance" issue in fiber optic transmissions by "splitting the light beams as they pass through a circuit, rotating one of the polarized beams, and finally rejoining them on their way out of the circuit, which retains the signals' strength." While there's no projection of when this technology could actually hit the mainstream, anything that makes it less expensive to rollout FiOS (and similar networks) to more people most definitely has our vote.
I still believe the most important finding from the article is the fact that the researchers have found a way to sustain distance light transfer.
This certainly is one of the largest constraints for rolling out fiber here in the U.S., since the nature of fiber does not lend itself well to long distances. At certain intervals, you will need an in-line amplifier. This is done via O-E-O (Optical -> Electrical -> Optical) signal regeneration. This can be pricey equipment for an optical carrier. It's this reason why smaller countries (Korea, UK, NL) have such great dark fiber networks, it's that much cheaper for them to roll out, due to the dramatic price decrease seen from short provisioning distances.
If the research here does in fact show a way to manipulate the light in a fashion that does not require an OEO transfer, then costs will be dramatically reduced, providing an incentive for massive dark fiber rollouts. What I imagine is a physical discrete device that utilizes polarizing filters, mirrors, and ADC monitoring circuitry to condition the original photon transfer, rather than analyzing it, and spitting out a new signal. This would create a light conditioner that may suffice for some of the work traditionally done by expensive amps.
Then there is always the next hurdle, how to go faster than the speed of light.
Dustin - There are EDFA (erbium-doped fiber amplifiers) that are capable of using the incoming signal as an exciter for a laser, so no OEO conversion is necessary. They are capable of replacing regenerators for all but the longest (trans-oceanic) spans.
Engadget - The equipment is a drop in the bucket compared to the cost of burying fiber in streets and hanging it overhead. A half million to a million dollars for a C.O. worth of equipment capable of handling tens of thousands of customers compared to many, many millions to actually get the fiber to people's houses.