Breakthrough research from The University of Texas at Arlington and The University of Vermont could lead to a dramatic reduction in the cost and energy consumption of high-speed internet connections.
Nonlinear-optical effects, such as intensity-dependent refractive index, can be used to process data thousands of times faster than what can be achieved electronically. Such processing has, until now, worked only for one optical beam at a time because the nonlinear-optical effects also cause unwanted inter-beam interaction, or crosstalk, when multiple light beams are present.
Currently, to eliminate the noise accumulated during light propagation in optical communication links, telecom carriers must resort to frequent optoelectronic regeneration, where they convert optical signals to electrical via fast photodetectors, process them with silicon-based circuitry, and then convert the electrical signals back to optical, using lasers followed by electro-optic modulators. Since each optical fiber can carry over a hundred different signals at various wavelengths, known as wavelength-division multiplexing (WDM), such an optoelectronic regeneration needs to be done separately for each wavelength, making regenerators large, expensive and inefficient consumers of power.
An attractive alternative to this is processing the optical signal directly, without converting it to electrical and back. In particular, the speed of light propagating in a transparent medium can be slightly modified by a change in the light intensity. This is a manifestation of a nonlinear-optical effect known as “self-phase modulation” or SPM. If light contains both signal and noise, the SPM can help clean the signal from noise by scattering the noise energy into frequencies well outside the signal band, from where the noise can be easily removed by a filter. When applied to light containing useful data, this SPM-enabled noise-removal operation is called “all-optical regeneration,” which can result in optical auto-correction of the signals carrying hundred times faster data rates than what can be processed electronically.