A Novel Network Monitoring Model by Machine Learning Approach for SDN

Due to value of data in current age, analyzing data and extracting useful information and hidden knowledge of them is more common and important than before. Computer networks as main source of producing and transporting of data are considerable in recent researches for data analyzing. Not only data network analysis does not have advantages but also has many necessities. On the other hand AI potentials in network monitoring are going to widespread more and more. So network traffic measurement base on machine learning (ML) is important research topic that is main motivation of this paper. To study more, our research focused on SDN monitoring and measurement data network because SDN has centralized management approach and is valuable source of data produced by data plane and on the other hand analyzing this data is necessary for managing and monitoring of overall network. So we prepare our model for SDN network traffic measurement base on learning and we make relationship between SDN traffic measurement requirements with ML solutions by goals of: network resource management and utilization, resource usage policy and bottleneck switch. Our model and method are implemented by python programming language in Jupyter notebook by decision tree classifier. We use already prepared data set that is SDN specific data set generated by using Mininet emulator and used for traffic classification. The results classify network switches base on their load in three levels: low, moderate and heavy and results was evaluated. Evaluation shows high precision, recall, accuracy, fl score.

کلیدواژه ها:

data ، network monitoring ، measurement ، SDN ، ML ، classification