Modulation Format Identification Technology And Application Using Support Vector Machine In Elastic Optical Networks

With the emergence of a variety of emerging data services, more and more kinds of modulation formats and line rate signals transmit in the network, lead to a greater flexibility requirement of the optical network. In this complex optical network, the available resources management and monitoring facing enormous challenges, configuration of the corresponding demodulator for different modulation format, then attain the monitoring data of the format to make targeted analysis, this is necessary to detect the optical performance of the intermediate node of the network, which has become one of the hotspots in the next generation of elastic optical network research. Automatic modulation format identification is generally divided into two methods, namely, decision-based theory and feature-based, we adopt the latter, it main focus on three steps: signal preprocessing, feature extraction and classification algorithm,There are many kinds of feature extraction and corresponding signal preprocessing techniques, but there exist many dispute about classification algorithm. Therefore, it is of great significance to explore the classification algorithm for modulation format identification.Aiming at the requirement of automatic modulation format identification in elastic optical network, a new method of optical signal modulation format identification based on support vector machine (SVM)and using asynchronous amplitude histogram (AAH) is proposed. The main work and innovation are summarized as follows:1. Expound the key technical principles (SVM et. al.) of feature extraction, and the characteristics of typical SVM classification functions are analyzed.2. We proposed and implemented the method of feature extraction of optical signal modulation format using AAH. Four typical complex modulated optical signals are taken as examples to validate the proposed method. The amplitude characteristic parameters are obtained, and the method is validated.3. A classification algorithm based on SVM is proposed and implemented for classifier in optical signal modulation format identification. Combined with AAH feature extraction method, four kinds of optical modulation formats are identified.Simulation results show that the classification accuracy of the proposed method is 98.95% and the modeling time is millisecond.