Abstract—The rapid evolution of Internet of Things (IoT) devices in advanced urban environments has transformed traditional cities into intelligent ecosystems, which are typically named as smart cities. These interlinked structures improve urban services as well as citizens' lives through smart transportation, public safety solutions, environmental monitoring, as well as energy management. Nevertheless, the incorporation of several IoT devices also increases important security concerns, since they enable potential attack vectors for malicious entities. Therefore, robust Intrusion Detection Systems (IDS) are important to protect sensitive data and ensure the integrity and availability of smart city applications. To mitigate these shortcomings, intrusion detection in IoT-based smart city applications develops Exponentially Weighted Parrot Optimizer_Shuffle Attention Network (EWPO_SA-Net). The IoT simulation is initially carried out, and the data is collected from simulated nodes. The input data is subjected to data normalization, which is done by employing linear normalization. Furthermore, the features are selected by utilizing the Mahalanobis distance. The intrusions are finally detected through the Shuffle Attention Network (SA-Net), which is trained by EWPO. EWPO_SA-Net obtains superior efficiency by analyzing analytics measures like accuracy, sensitivity, precision, F1−Score, and False Positive Rate (FPR) of 93.737%, 93.940%, 93.555%, 93.747%, and 6.521%.


Keywords—internet of things, intrusion detection, smart city applications, exponentially weighted moving average, parrot optimizer


Cite: C. Senthil Selvi, Nithya E., Perumal Sivaraman, and Velliangiri Sarveshwaran, “EWPO_SA-Net: A Hybrid Optimized Attention Network for Intrusion Detection in Smart City IoT Applicati," Journal of Communications, vol. 21, no. 2, pp. 187-200, 2026.


Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).