Smartchainguard: A Blockchain, Deep Learning, And AI-Based Framework For Malicious User Detection In 5G And Beyond Cognitive Radio Networks

Abstract

The evolution of 5G and Beyond 5G (B5G) wireless communication technologies has catalyzed a surge in the demand for dynamic and intelligent spectrum access. Cognitive Radio Networks (CRNs) address this need by enabling unlicensed users to opportunistically access underutilized spectrum resources. However, CRNs are highly vulnerable to spectrum sensing attacks, particularly Spectrum Sensing Data Falsification (SSDF) perpetrated by malicious users (MUs). These attacks can severely degrade network performance by disrupting the cooperative sensing process. This paper proposes SmartChainGuard, a novel security framework that integrates blockchain technology, deep learning via Long Short-Term Memory (LSTM) networks, and an AI-based trust management engine to detect and isolate MUs in CRNs. SmartChainGuard ensures tamper-proof logging of spectrum sensing data, models user behavior through sequential anomaly detection, and enforces trust-based access control via smart contracts. We provide a formal mathematical model for trust computation, a secure blockchain-based data integrity layer, and a complete algorithmic workflow. Simulations performed on synthetic and real datasets demonstrate the framework’s efficacy, achieving a 97% detection accuracy, reducing false positives to 4.5%, and maintaining trust stability across 88% of the simulation period. The framework operates in real-time and is scalable to B5G network requirements.