Design And Implementation Of Digital Currency Network Traffic Identification System Based On Machine Learning

In recent years,as the price of digital currency has soared,more and more electronic equipment has been put into mining activities,resulting in waste of electricity and computing resources.According to IBM’s report,mining Trojan has become the current main malware in cloud.The detection of mining activities is of great importance in terms of energy and network security.Currency identification is a more accurate mining detection method,which helps to carry out more refined management of mining activities.Based on Adaboost algorithm,this paper proposes a mining detection model and currency identification model.The accuracy of mining detection model and currency identification model can reach 97%and 91%respectively.The accuracy performance index of mining detection model and currency identification model based on ID3 decision tree algorithm,CART decision tree algorithm,Naive Bayes algorithm and SVM algorithm is exceeded.This paper implements a network traffic identification system of digital currency for encrypted traffic.The system includes a traffic collection module in the traffic collection layer and a data set generation module in the basic service layer.And the currency identification model training module,currency type annotation module,mining detection model training module and flow type annotation module of classification layer.The system of the experimental results show that the mining detection model of this system in a total of 26906 mining flow and the mining flow mix encryption traffic,to dig traffic identification accuracy reached 97%,currency identification model in a total of 4751 encryption mining flow on four different types of digital currency identification accuracy reached 91%,It has reached the advanced technology level and the expectation of the performance index,which proves the validity of the currency identification method proposed in this paper.