Power Loss Classification on Shifts Based on SMS (Singlemode-Multimode-Singlemode) Structured Fiber Optic Using Gaussian Naïve Bayes Method

Abstract

Singlemode-multimode-singlemode based optical fiber can be used as a good communication medium, but energy or power carried by light will be weakened (losses) due to leakage or due to lack of clarity or shift in optical fibers. In this study, power losses in fiber optics based on SMS will be classified based on changes in the values of power losses to shifts and divided them according to three classes, there are good, average, and bad. The shift will be used as a classification variable that is between 0 μm to 450 μm with an increment of 50 µm for every interval. The SMS optical fiber structure used is 5.5 with 25 attempts on different optical fibers. The classification method used is Naïve Bayes with a Gaussian distribution. Gaussian distribution is used in Naïve Bayes because the dataset will be processed in the form of continuous values. From the results of testing based on TP+TN = 6, FP = 6 FN = 6 on confusion matrix, the classification accuracy value was 42.86%. This indicates that this classification method is still less effective for classifying fiber optic power losses with an SMS structure. For further study, another classification methods can be used in the power loss classification to get better results.