A Network Traffic Prediction Method Based on LSTM

Abstract

The invention discloses a network traffic prediction method based on LSTM. The method is as follows: using a packet sniffing tool to capture network traffic data, deploying a packet sniffing tool on a routing node to capture network traffic data, and collecting all packets in a unit time As a sample, all packets in each sample are stored separately; data preprocessing is performed, features are extracted, and labels are marked. The extracted features include the total number of packets, the proportion of outbound / inbound packets, the total length of outbound / inbound packets, and the number of outbound / inbound packets. Packet average length, outbound / inbound packet length variance, total / average transmission time; use LSTM to model data; use the model to predict new data, and get the predicted value of network traffic. This method combines the long-short-term memory model with the artificial neural network to improve the prediction accuracy of network traffic.

Description

technical field [0001] The invention relates to the technical field of network traffic forecasting, in particular to an LSTM-based network traffic forecasting method. Background technique [0002] At present, the Internet based on TCP / IP technology is developing rapidly, new network technologies are constantly emerging, the scale of network infrastructure is constantly expanding, and network interaction is becoming increasingly active. As an important tool for work, life and study, the Internet has already affected the lives of the public in many aspects such as transportation, medical care, Internet services, and education, and has become an increasingly important part of daily society. Behind the rapid development of the Internet, the increasingly complex network environment also raises more and more questions for network researchers. Among them, a very important issue is, as a network service provider or a network manager, how to effectively acquire and analyze network f...

AI Technical Summary

Problems solved by technology

The training process of these methods is usually offline and time-consuming, but the classification process is more efficient and can be carried out in real time. This method, like the above method, requires network experts to extract a large amount of feature information from network data. , relying heavily on expert experience, it is difficult for the learning model to accurately and automatically extract useful features from network traffic data

Images