SDN-Based Network Architecture Optimization and Secure Data Transmission for Smart Taxation

With the rapid advancement of smart taxation systems, traditional network architectures are no longer capable of meeting the demands for efficient and secure data transmission. This study proposes a software-defined networking (SDN)-based architecture and secure data transmission framework specifically designed for smart taxation environments characterized by high-concurrency workloads, periodic traffic patterns, and complex cybersecurity threats. The proposed framework implements dynamic network resource scheduling and security protection through a three-layer SDN architecture comprising the data, control, and application layers. The framework incorporates a flow table compression algorithm that reduces flow table size by 40%-60%, an LSTM-based dynamic traffic prediction model that captures fiscal calendar periodicities, and an IPSO-CA attack detection algorithm that combines self-organizing maps with decision trees. To address the limitations of conventional SDN optimization methods, the proposed architecture separates audit logging from packet forwarding and introduces a rarity-harmfulness double-validation mechanism to improve the detection of low-frequency ("low-and-slow") attacks. Experimental results demonstrate that the proposed architecture improves controller load balancing by 35% and achieves fault recovery in less than 100 ms. The IPSO-CA algorithm attains a 98.2% detection rate for SYN Flood attacks with a false-positive rate of only 1.3%, outperforming conventional decision tree (90.5%) and support vector machine (93.7%) classifiers. These results demonstrate that the proposed framework provides a scalable, reliable, and secure networking solution that supports the operational requirements of modern smart taxation systems.