Design and real time performance Evaluation of secure methodologies to Detect and mitigate security breaches In 5g networks

Abstract

The growth of the network users demands the evolution of next-generation networks, namely, 5G. Though 5G offers a plethora of advantages, the medley of protocols, technologies, vertical use-cases, and architecture opens up security loopholes. Hence, our research work takes up the problem of security in the 5G network. In particular, our work revolves around three technologies of the 5G, namely, Software Defined Network (SDN), Internet of Things (IoT), and Fog computing. One of the prime motives of 5G is to support the latency-critical sectors. Hence, it is imperative to detect and mitigate the attacks with improved accuracy at the earliest possible time. We also address the problem of performing a real-time evaluation of the secure methodologies. newlineOur first attempt is to mitigate the attacks in wireless networks. We propose the Secure Deep Neural (SeDeN) framework and deploy it in decentralized SDN controllers. SeDeN collects the traffic over a dynamic period. The traffic is input to the detection algorithm, which has trained Long Short-Term Memory Recurrent Neural Network (LSTM RNN) Machine Learning (ML) model. The mitigation methodology drops the packets if LSTM RNN detects a security breach in the data plane. We test SeDeN in an emulator. SeDeN shows better attack detection time and a reduction in the packet loss rate. We use a supervised LSTM RNN, which has the issue of getting a large labeled dataset with minimal errors. newlineNext, we propose a LEDEM methodology to detect and mitigate the Distributed Denial of Service (DDoS) attack triggered by pernicious wireless IoT devices that victimizes the IoT server. newline newline