A CNN-based cost-effective modulation format identification scheme by low-bandwidth direct detecting and low rate sampling for elastic optical networks

Joint modulation format recognition and optical performance monitoring for efficient fiber-optic communication links using ensemble deep transfer learning

“…The true positive (TP), true negative (TN), false positive (FP), and false negative (FN) values are calculated from the confusion matrix. The metrics and their derivative formulas, which are used in the comparison of the imbalance metrics of the proposed algorithm, are given in the following equations: 18 – 22 , 35 , 36 precision=TPTP+FP,recall=TPTP+FN,F−score=2×precision×recallprecision+recall,specificity=TNTN+FP,PLR=sensitivity(1−specificity),NLR=(1−sensitivity)specificity,ROC=TP rateFP rate.…”

“…Successfully, ML implementations are also performed to transform future intelligent optical networks. 8,10 Results from a study [18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33] prove the advantages of using MLbased methods that can be used for futuristic predictions. This paradigm shift results from the various possible advantages provided by ML.…”

“…Khan et al 18 proposed an MFI and OSNR estimation method based on the amplitude histograms (AH) extracted after constant modulus algorithm equalization using hierarchical deep neural networks (DNNs) for each MF individually rather than simultaneously. Du et al 19 proposed a cost-effective CNN-based MFI using asynchronous AH under various ranges of impairment conditions. This work produced a comparative analysis of the results obtained from artificial neural network (ANN) and DNN.…”

“…The true positive (TP), true negative (TN), false positive (FP), and false negative (FN) values are calculated from the confusion matrix. The metrics and their derivative formulas, which are used in the comparison of the imbalance metrics of the proposed algorithm, are given in the following equations: [18][19][20][21][22]35,36 E Q -T A R G E T ; t e m p : i n t r a l i n k -; e 0 0 9 ; 1 1 6 ; 5 5 5…”

“…Recently, CNN techniques have gained momentum for successful application to image-based pattern-recognition tasks 18 – 32 in optical communication applications. Compared with other prediction models, CNN has gained importance in recent times due to its ability to extract features directly from raw images, automatically yielding higher efficiency and accuracy.…”