Intelligent Real-Time Polymorphic IoT Malware Detection System Using Opcode Analysis and Adaptive Machine Learning
IoT Malware Detection License Framework
📖 Project Overview This project detects polymorphic IoT malware by analyzing executable binaries converted into grayscale images. It uses a Convolutional Neural Network (CNN) trained on the MalImg dataset to classify malware families and benign files in real time.
🗂️ Dataset The MalImg Dataset contains grayscale images generated from malware binaries, categorized by families (e.g., Allaple, Lollipop).
Directory Structure:
├── benign_images/ # Benign IoT firmware images
├── malware_images/ # Malware family images (e.g., Mirai variants)
├── test_images/ # Images for model evaluation
🛠️ Installation Clone the Repository:
git clone https://github.com/your-username/iot-malware-detection.git
cd iot-malware-detection
Install Dependencies:
pip install tensorflow keras numpy matplotlib scikit-learn
Download the MalImg Dataset:
Place the dataset in the malware_images/ folder.
Add benign IoT firmware images to benign_images/.
🚀 Usage
- Train the CNN Model Run the training script to build and save the malware detection model:
python cnnmodel.py
Output: A trained model (saved in /model) and accuracy/loss plots.
- Test the Model Evaluate the model on new images:
python model_testing.py --image_path test_images/sample.png
📂 Project Structure
Copy
├── benign_images/ # Benign IoT firmware images
├── malware_images/ # Malware family images (from MalImg)
├── test_images/ # Test images for evaluation
├── model/ # Saved CNN model (e.g., model.h5)
├── cnnmodel.py # CNN model training script
├── model_testing.py # Script to test the model on new images
└── README.md
📊 Results
Accuracy: ~95% on the MalImg test set.
Confusion Matrix: Confusion Matrix
Real-Time Detection: Processes images in <50ms on a mid-tier GPU.
📚 References MalImg Dataset Paper
TensorFlow Documentation
Polymorphic Malware Detection Techniques