Efficient Detection of Denial‐of‐Service Attacks in Wireless Sensor Networks Depending on Binarized Simplicial Convolutional Neural Networks for Enhanced Security
R. Vidhya, S. Varunadevi, Murugananth Gopal Raj · International Journal of Communication Systems · 2025
ABSTRACT DoS attacks pose significant threats to wireless sensor networks (WSNs) by disrupting regular network availability. The existing systems face limitations such as limited power, storage, bandwidth, and processing capabilities, making them particularly vulnerable to security risks. Despite these constraints, an effective intrusion detection system (IDS) is essential for detecting such attacks. As denial‐of‐service (DoS) attacks become more frequent and sophisticated, the traditional intrusion detection systems are losing their effectiveness. To overcome these complications, Efficient Detection of Denial‐of‐Service Attacks in Wireless Sensor Networks using Binarized Simplicial Convolutional Neural Networks for Enhanced Security (ED‐DoS‐WSN‐BSCNN) is proposed. The input data are collected from the WSN‐DS dataset. The gathered data are given to the preprocessing stage with the help of the adaptive two‐stage unscented Kalman filter (ATSUKF) for data cleaning, data transformation, and normalization. Then the preprocessed data are given to the classification stage by using the binarized simplicial convolutional neural network (BSCNN) for classifying DoS attacks, such as normal, blackhole, grayhole, flooded, and TDMA. Finally, the Arctic tern optimizer (ATO) algorithm is employed to enhance the BSCNN that categorizes the types of DoS attacks accurately. The performance metrics like accuracy, precision, recall, specificity, F1‐score, computational time, and RoC are taken into account. The performance of the proposed technique is compared with other existing methods. The ED‐DoS‐WSN‐BSCNN technique is implemented in Python. The proposed technique attains 4.05%, 7.52%, and 2.91% higher accuracy, 4.10%, 7.61%, and 5.14% higher precision, 7.46%, 6.92%, and 2.88% higher recall, and 1.06%, 1.75%, and 2.31% higher specificity compared with existing methods: performance analysis of deep learning for DoS attacks identification in wireless sensor network (CNN‐DoS‐WSN), detection of DoS attack in wireless sensor networks: a lightweight machine learning approach (KNN‐DoS‐WSN), and extended evaluation on machine learning approach for DoS detection in Wireless Sensor Networks (RT‐DoS‐WSN), respectively.