Traffic identification method in 5g/imt-2020 telecommunication networks based on artificial intelligence

The paper analyzes the process of identifying incoming traffic in 5G/IMT-2020 networks, built on the Ultra-reliable and low latency communications technology, identifies its features and research directions to improve the efficiency of traffic identification taking into account the requirements for the functioning of the considered network. To solve the problem of identifying traffic in the 5G/IMT-2020 network, the paper develops and presents an appropriate methodology. The specified methodology includes the formation of metadata arrays of the incoming stream, their modification into a set of training data, the formation of a training software and hardware complex and the development of the neural network structure, the training process of the neural network and its implementation in the process of identifying traffic in 5G/IMT-2020 telecommunication networks. Evaluation of the results of the training process of the proposed neural network and verification of its operation on test data sets in the trained state showed that the neural network presented in the work is able to monitor and identify traffic generated from Internet of Things services with a probability of up to 99.7%. In the process of monitoring and identifying traffic from two or more services, this probability may decrease, but is within the acceptable limits of 80-90%.