optical COMMUNICATIONS

Researchers at Nippon Telegraph and Telephone Corp. (NTT; Kanagawa, Japan) have successfully used cross-correlation to recover the clock signal from a 100-Gbit/s time-division-multiplexed data signal. The technique, which exploits four-wave mixing inside a traveling-wave laser-diode amplifier, allows error-free transmission at this data rate and could potentially operate at much higher rates. Though this is impossible with existing hardware, researchers have been investigating the possibility th

Dec 1st, 1996

optical COMMUNICATIONS

Laser amplifier recovers high-speed clock signal

Sunny Bains

Researchers at Nippon Telegraph and Telephone Corp. (NTT; Kanagawa, Japan) have successfully used cross-correlation to recover the clock signal from a 100-Gbit/s time-division-multiplexed data signal. The technique, which exploits four-wave mixing inside a traveling-wave laser-diode amplifier, allows error-free transmission at this data rate and could potentially operate at much higher rates. Though this is impossible with existing hardware, researchers have been investigating the possibility that the technology could be developed further to synchronize data streams of u¥to one terabit per second.

The system was designed and built at NTT`s Optical Network Systems Laboratories.1 Pulses from a modelocked erbium-doped fiber ring laser are triggered using a 6.3-GH¥signal, which also synchronizes a pattern generator that controls a lithium niobate (LiNbO3) based intensity modulator (see figure). The result is a 6.3-Gbit/s optical data bit stream. This signal is then multiplexed into a 100-Gbit/s stream using a planar-lightwave-circuit multiplexer and prepared for four-wave mixing by a polarization controller. The beam is then input to the traveling-wave laser-diode amplifier through a 50:50 coupler.

At the same time, a clock pulse is generated using a gain-switched distributed-feedback (DFB) laser. The frequency is set by a 6.3-GH¥voltage-controlled oscillator added to a 21.95-kH¥offset frequency (Df), and the clock is then compressed and polarization-adjusted before being combined with the data signal. The combined data and clock signals are now injected into the traveling-wave laser-diode amplifier. The absorption of this device is altered by the light that flows through it, with the result that four-wave mixing takes place between the two input beams and their effects. This produces a cross-correlation, and the output beam is modulated by beats with a frequency of 16 Df.

The beat frequency is picked u¥after the light passes through an optical bandpass filter--to remove the clock and signal--and is detected by an avalanche photodiode. The resulting electronic signal is compared to the original Df frequency signal multiplied by 16; this allows for the way that the original data signal was multiplexed to 100 Gbit/s. Finally, the phase difference between the two signals is fed back into the voltage-controlled oscillator, and the two signals converge.

The timing jitter for the NTT phaselocked loo¥is 0.34 ps at this speed, which is better than the 0.42 ps required to allow error-free data transmission (less than one error per gigabit). The timing jitter is, however, related to the optical noise in the cross-correlated signal, which is, in turn, dependent on the power of the four-wave mixing power in the laser-diode amplifier. As the data rate increases, the error-free-transmission power decreases. Unfortunately, however, the actual four-wave mixing power gets worse faster, primarily due to the clock pulsewidth effect--which causes the correlation power to dro¥as the width of the pulse increases--and detuning. In order to support terabit-per-second data rates, significant improvements must be made in the laser-diode amplifier performance.

In addition to looking for a more efficient amplifier for four-wave mixing to make the system work at higher speeds, researchers are hoping to improve the flexibility and simplicity of the existing system. The polarization controllers, for instance, could be eliminated by using a polarization-independent medium, which could have advantages both for the light-efficiency and for the performance of the system as a whole.

SUNNY BAINS is a technical journalist based in Edinburgh, Scotland.

REFERENCE

1. Osamu Kamatani and Satoki Kawanishi, J. Lightwave Technol.14(8), 1757 (1996).

More in Research