Robust Encoding and Secure Storage of Executables Using Image Based Encoding Techniques

The ever-increasing sophistication of malicious software poses significant hurdles to the field of cybersecurity, notably in the areas of malware detection and safe executable management.This paper discusses how malware is becoming more complicated and how that affects cybersecurity, especially when it comes to finding malware and managing executable files. It presents a hybrid dual-phase methodology that integrates a machine learning and deep learning-based malware detection system with a secure encoding framework intended to safeguard trusted executables. During the malware detection phase, static features like opcode sequences, API calls, and structural characteristics are taken out of Portable Executable (PE) files. We use feature optimization and k-fold cross-validation to make the system work better. The methodology assesses five algorithms: XGBoost, Random Forest, Gradient Boosting, Deep Learning (utilizing Keras DNN), and SVM (employing RBF Kernel). The performance metrics show that XGBoost has the highest accuracy (99.48%), F1-score (0.991), and AUC (0.9997), with Random Forest and Gradient Boosting not far behind. The Deep Learning model also does very well, with an accuracy of 99.04% and an AUC of 0.9992. This shows that it can recognize complex, non-linear patterns in malware activity. The proposed framework uses a multi-layered encoding system in the secure encoding phase. This system combines Base64 transformation, image-based mappings, and Modified Least Significant Bit (MLSB) embedding techniques. This encoding keeps trusted executables safe from tampering and unauthorized access. It has a 99.2% retrieval accuracy, which is better than traditional encryption methods when it comes to keeping data safe and private. In general, the proposed framework is a clear, scalable, and safe way to classify malware and protect executables. It has a lot of potential to be used in cybersecurity, especially for cloud infrastructures and important systems. The approach plays a big role in making AI-powered systems that can protect against a wide range of digital threats.