In today's world, the number of online services and users is growing rapidly. This leads to a huge increase in internet traffic. Therefore, the task of separating IP traffic is approx. it is important for Internet service providers or ISPs, as well as a variety of government and the private sector for better network management and security. IP traffic separation includes identifying user activity using network traffic flowing into the system. This will also help to improve the network performance. The use of traditional IP traffic Classification methods based on the evaluation of packet capacity and hole numbers dropped significantly because there are so many online apps today that use naturally incorrect port numbers than well-known port numbers. Also, there are several encryption strategies today as a result of when testing the package payload is blocked. Currently, various machine reading techniques are commonly used to differentiate IP traffic. However, not much research has been done on IP fragmentation 4G network traffic. During this study, we did a new database by downloading real-time Internet traffic packets 4G network data using a tool called Wireshark. After that, we released the considered features of the packaged packages using the python script. Then we used five typewriters models, namely, Decision Tree, Vector Support Equipment, K Very Near Neighbors, Random Forest, and Naive Bayes IP splitting traffic. It was noted that Random Forest offered the best almost 87% accuracy