Improving the Security of SDN-IoT Networks Based on Intelligent Anomaly Detection an‎d Trust Management Framework with a Focus on Medical Applications

2/9/2027 12:00:00 AM

Abstract The Internet of Things (IoT) is a vast network of smart devices an‎d sensors with diverse applications such as the Internet of Medical Things (IoMT). In recent years, IoMT has become a practical platform for health monitoring an‎d telemedicine. However, the problems associated with the huge volume of medical data have led to the need for cloud-based an‎d edge-based processing in new platforms. In order to add intelligence to networks to address existing challenges, traditional IoT an‎d IoMT architecture have gradually given way to software-defined networking (SDN)-based architectures. Although the integration of SDN, edge computing, an‎d IoT has transformed modern network infrastructures an‎d provided scalability, flexibility, an‎d lower latency, these developments have also created new security challenges. Intrusion detection systems (IDS) have also been proposed to address these challenges. However, traditional IDSs often rely on signature-based methods that are inadequate against evolving an‎d unknown threats. In addition, the lack of a robust trust assessment mechanism between devices exacerbates the risk of insider attacks an‎d unauthorized access. Addressing these challenges requires a decentralized, intelligent, an‎d trust-aware security framework that is capable of detecting anomalies an‎d assessing trust in real time. Despite various research efforts to address these challenges, there are still gaps. One of the main gaps is the lack of attention to specific applications such as medical applications. Another gap is the lack of attention to the inherent imbalance of the dataset. To fill these two main gaps, this research proposes an innovative framework for intelligent anomaly detection an‎d trust management focusing on medical applications based on the SDN-IoT platform. In the proposed method, first a preprocessing is performed on this dataset an‎d then its data is cleaned to create a Medical Oriented dataset. The goal is to focus an‎d extract structures that are compatible with IoMT data. Considering the imbalance of the new data after filtering, two different an‎d complementary strategies, namely SMOTE an‎d Focal Loss, are used to balance the data. For intrusion detection, two LSTM networks with different architectures are used, an‎d their results are combined in a weighted manner. Then, a trust management system based on the available data sends information to the SDN controller so that the controller can make one of three decisions based on the incoming traffic. Simulation results on the TON-IoT dataset show that the proposed method can detect an‎d classify intrusion attacks with an accuracy of 93.34%. The results show that the proposed architecture can lead to improved network security an‎d its use in practical applications. Keywords Network security improvement, Intrusion Detection System (IDS), Software-Defined Networking (SDN), Internet of Things (IoT)

بهبود امنيت شبكه , سيستم تشخيص نفوذ (IDS) , شبكه‌هاي نرم‌افزارمحور (SDN) , اينترنت اشيا (IoT)

Network security improvement , Intrusion Detection System (IDS) , Software-Defined Networking (SDN) , Internet of Things (IoT)

Karrar AbuHameed

Dr Javad Vahidi