Blockchain architecture-based malware attack detection using web monitoring model with structure learning in behavioural artifacts analysis

Abstract

This paper proposes a novel malware detection framework that analyzes behavioural artifacts through web monitoring combined with blockchain and structural learning algorithms. The approach focuses on identifying malicious activities by examining application runtime behaviours, including file system operations, network communications, and API requests. Behavioural data collected from web monitoring sources undergoes preprocessing to remove noise and normalize the dataset. The refined data is then analyzed using a reinforcement structural learning model integrated with quantile regression and an adversarial neural network to improve detection performance. A contract-based encryption mechanism implemented through blockchain technology secures the detection results and preserves data integrity. The proposed framework is evaluated using Storm botnet, Waledac botnet, and Ericsson traffic datasets, with performance measured in terms of Accuracy, F1-score, Mean Average Precision, Detection Time, and Processing Time. Experimental results demonstrate that the proposed system achieves 97% accuracy, 96% F1-score, and 95% mean average precision, highlighting its effectiveness for behavioural malware detection.