Entropy-Guided Lottery Sparse Attention for SYN Flood Attack Detection in Smart Networks

The rapid growth of smart networks and Internetbased services has intensified the risk of distributed denial-ofservice attacks, with SYN Flood attacks posing a significant threat to network availability and reliability. Traditional intrusion detection methods often struggle to achieve high accuracy while maintaining low computational complexity in large-scale network environments. To address these challenges, this paper proposes an Entropy-Guided Lottery Sparse Attention framework for SYN Flood attack detection in smart networks. The Search and Rescue Optimization (SARO) algorithm is first employed to select the most informative features. The selected features are then normalized and label encoded before being processed by the proposed sparse attention model. Experimental results demonstrate an accuracy of 99.32% with AUC values exceeding 0.998. Comparative and ablation studies confirm the effectiveness and efficiency of the proposed approach.