• Korean Journal of Optics and Photonics
• /
• v.34 no.4
• /
• pp.151-156
• /
• 2023

We propose to implement cost-effectively a high-speed short-haul interconnect by transmitting a 200-Gb/s/λ two-channel optical time-division-multiplexed signal generated by a carrier-suppressed optical pulse, which improves the robustness of the multiplexed signal to chromatic dispersion. The multiplexed 200-Gb/s signal is generated in the transmitter by combining two 100-Gb/s 4-level pulse-amplitude-modulated signals (generated using the optical pulse and two Mach-Zehnder modulators). After the signal is transmitted over a fiber, it is amplified by a semiconductor optical amplifier and detected by a photodiode. The amplified spontaneous emission noise is eliminated by an optical band-pass filter. The transmitted signal is reconstructed by a 2 × 2 multiple-input multiple-output equalizer, which compensates for crosstalk. Due to the use of the carrier-suppressed optical pulse, the 200-Gb/s signal can be transmitted over fiber with a chromatic dispersion of 40 ps/nm.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.43 no.5 s.347
• /
• pp.35-44
• /
• 2006

Reflective semiconductor optical amplifiers(R-SOAs) were designed with high gain, wide optical bandwidth, high thermal reliability and wide modulation bandwidth in TO-can package for the transmitter of wavelength division multiplexed-passive optical network(WDM-PON) application. Double trench structure and current block layer were introduced in designing the active layer of R-SOA to enable high speed modulation. The injection power requirement and the viable temperature range of WDM-PON system are experimentally analysed in based on Amplified Spontaneous Emission(ASE)-injected R-SOAs. The effect of the different injection spectrum in the gain-saturated R-SOA was experimentally characterized based on the measurements of excessive intensity noise, Q factor, and BER. The proposed spectral pre-composition method reduces the bandwidth of injection source below the AWG bandwidth and thereby avoids spectrum distortion impeding the intensity noise reduction originated from the amplitude squeezing.

• Korean Journal of Optics and Photonics
• /
• v.15 no.3
• /
• pp.193-199
• /
• 2004

We present the double resonance optical pumping(DROP) spectra in the transition 5P$_{3}$2/-4D$_{3}$2/ and 5P$_{3}$2/-4D$_{5}$ 2/ of ($^{87}$ Rb) and the frequency stabilization in the $1.5mutextrm{m}$ region using those spectra. Those spectra have high signal-to-noise ratio and narrow spectral linewidth, which is about 10 MHz. We could account fur the relative intensities of the hyperfine states of those spectra by the spontaneous emission into the other state. When the frequency of the $1.5mutextrm{m}$ laser diode was stabilized to the DROP spectrum, the frequency fluctuation was about 0.2 MHz fDr sampling time of 0.1 s and the Allan deviation(or the square root of the Allan variance) was about 1${\times}$10$^{-11}$ for averaging time of l00s.

• Korean Journal of Optics and Photonics
• /
• v.16 no.3
• /
• pp.182-190
• /
• 2005

The fiber Raman amplifier(FRA) is a distinctly advantageous technology. Due to its wider, flexible gain bandwidth, and intrinsically lower noise characteristics, FRA has become an indispensable technology of today. Various FRA modeling methods, with different levels of convergence speed and accuracy, have been proposed in order to gain valuable insights for the FRA dynamics and optimum design before real implementation. Still, all these approaches share the common platform of coupled ordinary differential equations(ODE) for the Raman equation set that must be solved along the long length of fiber propagation axis. The ODE platform has classically set the bar for achievable convergence speed, resulting exhaustive calculation efforts. In this work, we propose an alternative, highly efficient framework for FRA analysis. In treating the Raman gain as the perturbation factor in an adiabatic process, we achieved implementation of the algorithm by deriving a recursive relation for the integrals of power inside fiber with the effective length and by constructing a matrix formalism for the solution of the given FRA problem. Finally, by adiabatically turning on the Raman process in the fiber as increasing the order of iterations, the FRA solution can be obtained along the iteration axis for the whole length of fiber rather than along the fiber propagation axis, enabling faster convergence speed, at the equivalent accuracy achievable with the methods based on coupled ODEs. Performance comparison in all co-, counter-, bi-directionally pumped multi-channel FRA shows more than 102 times faster with the convergence speed of the Average power method at the same level of accuracy(relative deviation < 0.03dB).