Simpler Technique Measures Group Velocity Dispersion

Richard Gaughan

Group velocity dispersion in optical fiber degrades the quality of transmitted data and limits the data rate. If the dispersion in the system is known, it can be compensated, but measuring it takes sensitive, specialized equipment. A new technique uses stimulated Raman scattering to probe group velocity dispersion with standard laboratory equipment.

When two beams copropagate through a fiber, the glass can mediate energy transfer between them. Specifically, a shorter-wavelength, higher-energy beam transfers energy to the longer-wavelength beam. Researchers at JDS Uniphase Corp.'s branch in Eatontown, N.J., have exploited this effect to measure group velocity dispersion more easily.

In a proof-of-principle experiment, they modulated the output from a 1447-nm, 130-mW diode source and transmitted it down a fiber along with a 6-mW signal beam from a tunable 1500- to 1580-nm laser diode. At a given instant, the modulation on the 1447-nm pump beam provided a spatially variable pump power that coupled with the signal beam. The signal beam, with the modulation of the pump beam impressed upon it, continued to propagate through the fiber, and a photodiode at the end measured the output power. A standard network analyzer compared the initial pump power and final signal modulation power as the frequency swept through a range of 300 kHz to 150 MHz.

Calculations within 5 percent

At a given signal wavelength, the modulation transfer function was measured, and the wavelength was varied through the diode's tunable range. Because the power coupling efficiency directly relates to the group delay difference between the pump and signal, the researchers could fit the data to an analytical expression for the dispersion, with the delay difference measured for wavelengths in the region of interest. Although the noise floor of the network analyzer limited measurements above 60 MHz, the final calculation of group velocity dispersion with the technique was within 5 percent of that determined through more expensive phase difference measurement.

"This technique gets around the phase measurement difficulties present in phase dispersion measurement methods," said Idan Mandelbaum, a member of the research team. "In addition, this technique is relatively less noisy due to the need to take only one derivative, and not two as required by modulation phase shift methods. Compared to pulsed methods, it requires much less bandwidth."

Mandelbaum noted that 40-Gb/s networks are very sensitive to dispersion, so there is an essential need for a system that can be carried into the field to take a quick measurement of the dispersion.

"This method can be made very small and inexpensive," he concluded, "and it also allows the measurement of a fiber in real conditions, under Raman gain, as opposed to just a 'cold' fiber."