Blockchain-enhanced SDN framework with deep unfolding spatial frequency attention network and pine cone optimization for smart city IoT cybersecurity

The rapid expansion of the Internet of Things (IoT) in smart cities has revolutionized services by connecting numerous devices to deliver intelligent applications. However, this interconnected environment also exposes IoT networks to severe cybersecurity risks, including data breaches, intrusions, and advanced persistent threats. Traditional detection mechanisms often struggle with high-dimensional traffic, noisy data, missing values, and evolving attack patterns, which significantly reduce their reliability. Moreover, ensuring secure and tamper-proof storage of detection outcomes in large-scale IoT system s remains a major challenge. To address these issues, this paper proposes a blockchain-enhanced IoT cybersecurity framework for smart cities, utilizing a software-defined networking framework with a Deep Unfolding Spatial-Frequency Attention Network and Pine Cone Optimization (DUSFAN-PCO). Initially, input data is collected from UNSW-NB15 and CIC-IDS2017 datasets. The collected data are pre-processed using Diffusion Models for Missing Value Imputation (DMMVI) to handle noise and missing values. Next, Carpet Weaver Optimization (CWO) is applied to select the most informative attributes. These selected features are then fed into the Deep Unfolding Spatial-Frequency Attention Network (DUSFAN), which classifies IoT network traffic into normal and multiple attack categories through real-time anomaly detection. To further enhance detection accuracy, Pine Cone Optimization (PCO) is used to fine-tune DUSFAN’s weight parameters. Finally, the results are securely stored and managed using a Blockchain-Based Secure PBFT Lightweight Consensus (SPBFT-LC), ensuring immutability, transparency, and trust among IoT nodes. The proposed DUSFAN-PCO framework achieves an accuracy of 99.20% on UNSW-NB15 and 99.48% on CIC-IDS2017, with F1-scores of 99.52% and 99.50%, and ROC values of 0.99–1.00 across both datasets. Coupled with high blockchain-based security integrity (> 99%), it ensures robust, real-time, and reliable detection of IoT network intrusions, making it highly suitable for secure and resilient smart city IoT environments.