Malware Behavior Analysis Using Static and Dynamic Analysis Approaches

Abstract

Malware detection is a critical challenge in the evolving landscape of cybersecurity. This study explores the integration of static and dynamic analysis to enhance malware detection accuracy. Static analysis examines malware files without execution, providing insights into their metadata and structural attributes, while dynamic analysis observes behavior during execution in controlled environments. Using a dataset of 5000 samples, including ransomware, trojans, spyware, and worms, tools like IDA Pro, PE Studio, and sandbox platforms were employed. Results indicate that 87% of malware samples utilize code obfuscation to evade detection, and 95% exhibit suspicious runtime activities, such as registry modifications and encrypted network communications. A machine learning model (Deep Neural Networks, Random Forest, Support Vector Machine) trained on hybrid datasets achieved 97.8% accuracy with DNN, demonstrating superiority over single-method approaches. Challenges like high computational demands were addressed through cloud-based implementations.