研究目的
To present and validate soft failure localization algorithms for elastic optical networks during commissioning testing and in-operation phases to detect and localize failures early, preventing service level agreement violations and anticipating hard failures.
研究成果
The TISSUE algorithm effectively localizes soft failures during commissioning testing by comparing estimated and theoretical BER slopes, with simulations showing accurate detection of noise-induced failures. The FEELING algorithm achieves high localization accuracy for in-operation failures, especially with filter mask correction, enabling early detection before significant BER degradation. Both algorithms provide automated solutions for soft failure localization in elastic optical networks, reducing human intervention and improving network reliability.
研究不足
The number of OTC modules per node is limited to one OTCTx and one OTCRx, restricting concurrent tests. OSAs are placed only in outgoing links, limiting failure localization granularity to the node level unless more OSAs are added, increasing cost. Simulations may not fully capture real-world network complexities and variations.
1：Experimental Design and Method Selection:
The study uses simulation-based experiments to validate two algorithms: TISSUE for commissioning testing and FEELING for in-operation monitoring. TISSUE involves active monitoring with OTC modules to estimate BER, while FEELING uses OSAs for spectrum analysis and machine learning for failure classification and localization.
2：Sample Selection and Data Sources:
Simulations are conducted with a 250 Mb/s channel over 1000 km of single-mode fiber (ten 100 km spans) for TISSUE, and a 120 Gb/s DP-QPSK lightpath traversing 10 optical filters for FEELING.
3：List of Experimental Equipment and Materials:
Optical testing channel (OTC) modules, optical spectrum analyzers (OSAs), wavelength-selective switches (WSS), optical amplifiers (OA), dispersion compensation fiber (DCF), channel equalizers, tunable lasers, pseudo-random bit sequence (PRBS) generators, and simulation software for BER and spectrum analysis.
4：Experimental Procedures and Operational Workflow:
For TISSUE, OTC modules are allocated in nodes, test signals are injected, BER is measured at intermediate and egress nodes, and slopes are compared to detect failures. For FEELING, OSAs acquire spectrum data, features are extracted, and machine learning algorithms (decision trees and support vector machines) are used for diagnosis and localization.
5：Data Analysis Methods:
BER values are analyzed using slope comparisons in TISSUE. Spectrum features (e.g., central frequency, symmetry, bandwidth) are processed with decision trees and support vector machines in FEELING. Accuracy is measured as the proportion of correct localizations.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
