Enhance Data Security with Efficient Anomaly Detection in a Real Network Environment

Security in share markets is essential to affirm the integrity, stability, and trust of financial transactions, protecting investors from fraud and cyber threats. Current methods face challenges in high-volume attacks due to their scalability and static rule-based mechanisms. To resolve these issues, this paper develops a Hybrid Isolated Fuzzy Logic to detect anomalies in transactions. The Fuzzy Logic System with the Zebra Optimization Algorithm is utilized to enhance the attack detection accuracy. The isolation forest algorithm computes the threshold abnormal score to distinguish the normal from the malicious transactions. The Reinforcement Learning-based Proximal Policy Optimization algorithm dynamically updates network policies. The Network Function Virtualization for Distributed Denial-of-Service Scrubbing is combined to improve scalability and deliver cost-effective mitigation. The experimental analysis utilizing stock anomaly detection datasets is conducted. The experimental outcomes affirm that the proposed model attains an accuracy of ۹۸.۹۰%, a false positive rate of ۲.۹%, and improved mitigation efficiency than existing methods.

کلیدواژه ها:

Software Defined Networking ، fuzzy logic system ، Zebra Optimization Algorithm ، Reinforcement Learning ، Proximal Policy Optimization