Novel AI-Based Scheme for Traffic Detection and Recognition in 5G Based Networks

Abstract

With the dramatic increase in the number of connected devices, the traffic generated by these devices puts high constraints on the design of fifth generation cellular systems (5G) and future networks. Furthermore, other requirements such as the mobility, reliability, scalability and quality of service (QoS) should be considered as well, while designing such networks. To achieve the announced requirements of the 5G systems and overcome the high traffic density problems, new technologies, such as the mobile edge computing (MEC) and software defined networking (SDN), and novel schemes, such as artificial intelligence (AI) algorithms and offloading algorithms, should be introduced. One main issue with the 5G networks is the heterogeneous traffic, since there are enormous number of applications and sub-networks. The main design challenge with the 5G network traffic is the recognition and classification of heterogeneous massive traffic, which cannot be performed by the current traditional methods. Instead, new reliable methods based on AI should be introduced. To this end, this work considers the problem of traffic recognition, controlling and management; mainly for ultra-dense 5G networks. In this paper, a novel AI algorithm is developed to detect and recognize the heterogeneous traffic at the core network. The algorithm is implemented at the control plane of the SDN network, located at the core network. The algorithm is based on the neural network. The system is simulated over a reliable environment for various considered cases and results are indicated.