The researchers have designed the Firefly architecture, but it is not yet implemented. They have created a simplified, proof-of-concept system to show that their infrared laser can carry the signal and target the receiver. They are transmitting wavelength division multiplexed — multiple signals sent by different colored lights — bi-directional data streams each carrying data at a transmission rate of 10 Gigabits per second from a Bit Error Rate (BER) test set. BER testing determines the number of errors in a signal caused by interference, noise, distortion or sychronization problems.
The proof of concept setup has the bidirectional signal wavelength division multiplexed with a one-way cable television signal. The total data stream goes from fiber-optic cable to the infrared laser, across the room to the receiver and shows the results on a TV and the BER test set. A hand breaking the laser beam shuts off the system, but when the hand is removed, the signal is rapidly reacquired.
The system uses MEMs — microelectromechanical systems — with tiny mirrors for rapid targeting and reconfiguring, Kavehrad said. These MEMs use tiny amounts of electricity from four directions to reposition the mirror that targets the receiver. The movement of the mirrors is so small it is undetectable, but the computer program quickly locates the receiver and then narrows the target to pinpoint accuracy. The laser beam can also be rapidly moved to target a different receiver.
Accurately targeting and sending a signal via infrared laser are only two of the hurdles the researchers need to pass before Firefly is operational. Once the signal arrives at the target it must seamlessly enter the fiber-optic cable. Controlling and managing the data distribution system in an unwired environment is also important.
"We are trying to come up with something reconfigurable using light instead of millimeter waves (radio frequency)," said Kavehrad. "We need to avoid overprovisioning and supply sufficient capacity to do the interconnect with minimal switches. We would like to get rid of the fiber optics altogether."
Also working on this project were Sami R. Das, professor of computer science and Himanshu Gupta, associate professor of computer science and Jon Longtin, professor of mechanical engineering, Stony Brook University; and Vyas Sekar, assistant professor of electrical and computer engineering, Carnegie Mellon University.
The National Science Foundation supported this project.