Wavelength-division multiplexing (WDM) optical networks consists of optical cross-connects (OXCs) and optical add/drop multiplexers as nodes, optical cable as link, and optical amplifiers compensating transmission loss. Its hierarchy involves three optical layers: optical channel layer, optical multiplex section layer, and optical transmission layer. WDM optical networks provide a lot of advantages, such as optical transparency, fault restoration and network reconfiguration, but we cannot assure communication reliability through the conventional performance monitoring at optical repeaters and terminals. Therefore, it is strongly recommended to do performance monitoring of WDM optical networks in the optical layers.
There are five items about the performance monitoring in the optical layers. First, are the end-to-end optical paths correct? If not, where does the fault occur? Second, how much does each channel wavelength deviate from ITU-T standard? Third, are channel powers uniform with the flat gain of the optical amplifiers over optical paths? Fourth, how much does optical signal-to-noise ratio (OSNR) degrade due to amplified spontaneous emission (ASE) of the optical amplifiers? Finally, how much does signal quality degrade due to the signal distortion and noises?
Four items except the last one can be monitored either in an optical domain or in an electrical domain, and the last one only in an electrical domain. In this dissertation, we propose and demonstrate novel performance monitoring schemes in an optical domain for WDM optical networks.
In chapter 2, we explain the concept of “optical path” and propose a novel optical-path supervisory scheme using pilot tones and channel equalizers. The scheme is scalable to the number of nodes and wavelengths. In the experiment, we inputted 4 channels into a 4x4 OXC and monitored the optical-path routing state. The pilot tones superimposed at 4 input ports were 2.0, 2.5, 3.0, and 3.5 kHz. The rising and falling time of the optical-path routing change detection were about 10 msec and the measured power penalty was less than 0.1 dB. We analyzed the effect of the residual pilot tone unsuppressed at each OXC and concluded that it does not give problems in monitoring optical-path routing.
In chapter 3, we propose a low-cost optical channel analyzer using a tunable Fabry-Perot (F-P) filter for measuring wavelength, optical power, and OSNR of each channel in WDM systems. The optical channel analyzer has an automatic temperature controller and two fiber bragg gratings (FBGs) to solve the temperature sensitivity of the F-P filter. It is stable and low-cost because the scheme does not demand precise measurement equipment and an extra light source. By using the modeling of data acquisition process and signal process, we simulated the measurement errors of channel power and OSNR for the variation of optical receiver bandwidth (BW), noises, and F-P filter BW. The simulation results showed that the OSNR measurement errors were very sensitive to the noises and the mismatch of the F-P filter BW. Furthermore, the difference between the transmission spectrum of the commercial F-P filter and the theoretical airy function produces more OSNR measurement errors as the larger ONSR. In order to solve these problems, we measured OSNR by using ASE curve fitting. The experimental results showed that the wavelength measurement errors were within ±30 pm in the range of 1547.72 ~ 1558.98 nm, and that the channel power measurement errors were within ±1 dB in the range of -34 ~ -14 dBm. Finally the OSNR measurement errors were within ±1 dB in the range of 10 ~ 35 dB (200-GHz spacing) or in the range of 10 ~ 30 dB (100-GHz spacing).
In chapter 4, we proposed and analyzed a novel method enhancing the resolving power of tunable optical filters for accurate channel power measurement in dense WDM systems, which is based on filter dithering and lock-in detection. When a tunable FP filter is used, the proposed method always reduces the 3-dB BW by 67 %. We have experimentally shown that 0.04-nm-spaced optical channels are definitely discernible with a FP filter having a 3-dB BW of 0.06 nm and that the channel power measurement errors decrease.
Finally, in chapter 5, we explained the OSNR measurement method using Stokes parameters. The experimental results showed that the measured OSNRs were smaller because the F-P filter characteristics and the experimental setup, and that they largely deviated because of the measurement error of Stokes parameters. We could have solved the former problem by using a narrower F-P filter and the setup with a single-pass configuration. The analysis about the angle adjustment of λ/4 linear polarizers and a quarter waveplate for Stokes parameter measurement showed that we had to be able to adjust the angles with +0.1° in order to measure OSNR up to 25 dB. It is advisable to implement the linear polarizers and the λ/4 phase retader monolithically on a single plate by using integrated optics technologies.