GSR-C2N: Graph Feature Extracted Spar-Raven Optimized CNN Based Crypto Mining Framework

Abstract

The primary objective of this research is to increase the security of computer systems and networks by developing robust tools and techniques. The issue of crypto mining has become increasingly important in the field of cyber security due to the rapid increase in cryptocurrency usage. The main challenge in crypto mining relies on the extraction of the most relevant features and finding the optimal values. To concentrate more on these challenges, the Graph Feature Extracted Spar-Raven Optimized Convolutional Neural Network based Crypto mining framework (GSR-C2N) is proposed that enables the prompt detection and effective mitigation of crypto mining. By doing so, the research aims to address the potential adverse impacts caused, including performance slowdowns, heightened energy usage, and financial losses incurred by both individuals and organizations. The transaction of each block is monitored and controlled by the Block transaction information controller that safeguards security and accuracy. Specifically, the Spar-Raven optimization hybridizes the unique characteristics, including the memory and intelligence characteristics of the Raven with the Spar's keen awareness of predators, to find the global best solution and adaptively fine-tune the hyper-parameters of the GSR-C2N classifier. The performance of the model is analyzed using the crypto-mining-malware dataset, where the accuracy, sensitivity, and specificity of the proposed GSR-C2N were 96.848%, 96.388%, and 97.505% for K-Fold 10, and achieved 96.413%, 96.388%, and 96.633% for Training percentage 80%. Moreover, the proposed approach exhibits better performance and offers rapid processing speed, scalability, adaptability, and seamless deployment across diverse networks, making the GSR-C2N model efficient for performing in a real-time environment.