As an inevitable trend of future 5G networks, Software Defined architecture has many advantages in providing centralized control and flexible resource management. But it is also confronted with various security challenges and potential threats with emerging services and technologies. As the focus of network security, Intrusion Detection Systems (IDS) are usually deployed separately without collaboration. They are also unable to detect novel attacks with limited intelligent abilities, which are hard to meet the needs of software defined 5G. In this paper, we propose an intelligent intrusion system taking the advances of software defined technology and artificial intelligence based on Software Defined 5G architecture. It flexibly combines security function modules which are adaptively invoked under centralized management and control with a globle view. It can also deal with unknown intrusions by using machine learning algorithms. Evaluation results prove that the intelligent intrusion detection system achieves a better performance. 1 Introduction Software Defined 5G architecture will be a crucial tendency in the development of future 5G networks [1]. It takes the advantage of Software Defined Network (SDN) [2] and Network Functions Vir-tualization (NFV) [3] through centralized management and dynamic resource allocation to meet the demands of 5G networks. Besides, the separation of the control and execution planes also facilitate the supervision of network status and the collection of information. With the uprising of novel technologies and attacks, it will also be faced with various challenges and severe security situations. As a result, new network security systems and architectures are desperately needed to enhance the security of Software Defined 5G networks [4]. As an essential technology in network security, intrusion detection systems have received more and more concerns in efficiently detecting malicious attacks. Existing IDS with separate functions are usually deployed locally within restricted areas which are hard to cooperate with each other. Moreover, they are usually signature-based by matching behaviors of incoming intrusions with historical knowledge and predefined rules, which are unable to detect novel attacks intelligently. To overcome the limitation of traditional IDS, Artificial Intelligence (AI) has been employed for intelligent detection. They classify abnormal traffic using machine learning techniques with a self-learning ability [5]. At present, there have been a few researches in the combinations of IDS and AI. However, they are still inadequate for coordinated detection considering the evolution and development of network systems. In this paper, we propose an intelligent intrusion detection system for Software Defined 5G networks. Benefit from the Software Defined technology, it integrates relevant security function modules into a unified platform which are dynamically invoked under centralized management and control. Besides, it implements machine learning to intelligently learn rules fro...

The article presents a throughput maximization approach for UAV assisted ground networks. Throughput maximization involves minimizing delay and packet loss through UAV trajectory optimization, reinforcing the congested nodes and transmission channels. The aggressive reinforcement policy is achieved by characterizing nodes, links, and overall topology through delay, loss, throughput, and distance. A position-aware graph neural network (GNN) is used for characterization, prediction, and dynamic UAV trajectory enhancement. To establish correctness, the proposed approach is validated against optimized link state routing (OLSR) driven UAV assisted ground networks. The proposed approach considerably outperforms the classical approach by demonstrating significant gains in throughput and packet delivery ratio with notable decrements in delay and packet loss. The performance analysis of the proposed approach against software-defined UAVs (U-S) and UAVs as base stations (U-B) verifies the consistency and gains in average throughput while minimizing delay and packet loss. The scalability test of the proposed approach is performed by varying data rates and the number of UAVs.

No abstract