IDSoft: A federated and softwarized intrusion detection framework for massive internet of things in 6G network

Abstract

With the rollout of sixth generation (6G)–enabled massive internet of things (IoT) networks, the amount of data generated by IoT devices is expected to grow tremendously. Therefore, existing intrusion detection system (IDS) technology may not be sufficient to meet the scale and accuracy requirements for securing massive IoT. To overcome this shortcoming, edge intelligence–empowered next-generation (NextGen) IDSs utilize federated learning (FL) architectures. Nonetheless, to explore the full potential of federated IDSs, convergence and intercommunication overhead require some attention. One solution is softwarization and virtualization, which can increase network scalability and provide innovative security infrastructure for NextGen IDSs, offering holistic coverage and visibility to the entire network. In this line, we propose IDSoft, a novel softwarized solution that resides across the network infrastructure and leverages 6G enabling technologies, such as network function virtualization, mobile edge computing, and software-defined networking, to support FL-based IDSs. IDSoft is a scalable solution aimed at providing rapid and accurate detection combined with adaptive mitigation of large-scale cyberattacks while optimizing network resources. In this work, we analyze existing anomaly-based IDS approaches in IoT networks and focus on designing a hierarchical FL (HFL) framework for intrusion detection in IDSoft with synchronous and asynchronous aggregation and an additional offloading mechanism to enhance its performance. The numerical results demonstrate that the proposed HFL solution significantly reduces communication overhead, accelerates convergence, and promises greater scalability. Finally, future research trends are discussed.