K-Nearest Neighbors Classifier for Field Bit Error Rate Data

Proactive control strategies based on machine learning can be expected to be very useful for detecting anomalies and soft failures in high-capacity optical networks. In this paper, we propose a bit error rate (BER) classifier based on the K-nearest neighbors (KNN) algorithm. The BER classifier has been used for classifying field BER data collected in a 230-km optical link of the CANARIE network. The results show that the classification accuracy increases up to 97.8% depending on the features considered in the classification process.