A novel malware detection framework converts operational code sequences into discrete numerical signals to identify threats targeting Windows Portable Executable and Internet of Things environments. Addressing the rapid growth of Internet usage and the alarming increase in malware threats confronting digital systems worldwide, the methodology applies signal processing techniques directly to Opcode streams for feature extraction. The researchers note that applying signal processing tools to Opcode sequences represents a first-time innovation that enables platform-independent malware detection.
To extract discriminative features from executable files, the framework analyzes overlapping sub-signals using a combination of entropy-based algorithms and fractal dimension methods. The entropy measures include Singular Value Decomposition Entropy, Permutation Entropy, and Spectral Entropy. These statistical metrics are paired with fractal dimension methods, including the Katz, Higuchi, and Petrosian algorithms.
Following feature extraction, the generated feature vectors are fed into a LightGBM classifier to perform binary classification between malware and benign samples. The researchers evaluated the framework across three separate datasets covering Windows Portable Executable files and Internet of Things samples. In these validation tests, the proposed architecture achieved classification accuracies of 94.65 percent, 93.99 percent, and 99.80 percent across the three datasets, respectively.
Assessments focused on the Windows dataset revealed that the framework achieved a 1 percent to 6 percent improvement over conventional approaches. This performance gain was demonstrated across multiple evaluation metrics, including precision, accuracy, recall, and F-measure.
Subject of Research: Signal processing and entropy-based feature extraction for malware detection
Article Title: Singular value decomposition entropy: An efficient feature for malware detection
Article References: Abazar, M., Alkhrijah, Y., Altulyan, M., Bin-Asfour, M., Mohammadi, M., Boostani, R., & Manavi, F. (2027). Singular value decomposition entropy: An efficient feature for malware detection. Cluster Computing, 30(1), Article 34. https://doi.org/10.1007/s10586-026-06794-3
Image Credits: AI Generated
DOI: 10.1007/s10586-026-06794-3
Keywords: Malware Detection, Opcode Sequences, Signal Processing, Singular Value Decomposition Entropy, Permutation Entropy, Spectral Entropy, Fractal Dimension, Katz Fractal Dimension, Higuchi Fractal Dimension, Petrosian Fractal Dimension, LightGBM Classifier, Windows Portable Executable, Internet of Things, Binary Classification, Discrete Numerical Signals, Feature Extraction, Platform-Independent Detection
Cite Scienmag News
Denise Maddox. (November 20, 2026). Signal Processing Framework Improves Windows Malware Detection by 1 to 6 Percent. Scienmag. https://scienmag.com/signal-processing-framework-improves-windows-malware-detection-by-1-to-6-percent/
Denise Maddox. "Signal Processing Framework Improves Windows Malware Detection by 1 to 6 Percent." Scienmag, 20 November 2026, https://scienmag.com/signal-processing-framework-improves-windows-malware-detection-by-1-to-6-percent/. Accessed 20 November 2026.
Denise Maddox. "Signal Processing Framework Improves Windows Malware Detection by 1 to 6 Percent." Scienmag. November 20, 2026. https://scienmag.com/signal-processing-framework-improves-windows-malware-detection-by-1-to-6-percent/








