Embedded Internet of Things (IoT) devices are essential components of modern infrastructure, driving automation across healthcare, manufacturing, and smart homes. However, their pervasive deployment makes them highly attractive targets for cyberattacks. Adversaries increasingly deploy specialized malware to compromise these devices, turning them into botnet nodes or pivot points to infiltrate larger corporate networks.

Despite this escalating threat, traditional malware detection tools are fundamentally incompatible with embedded systems. These resource-constrained devices lack the memory, processing power, and energy capacity required to run conventional, heavy-duty antivirus solutions. Consequently, there is a critical need to develop lightweight, efficient malware detectors tailored specifically for the IoT ecosystem. By leveraging static program analysis, defenders can extract features from executable files without the overhead of dynamic execution, paving the way for low-overhead security solutions.

The task of the student is to study the expressiveness of various static features used for malware detection. Based of the results of their study, the student must also develop a lightweight malware detector fit for embedded IoT devices.

Tasks to be performed by the student will include:

- Present the state-of-the-art literature on machine learning-based malware detection for embedded IoT devices with a special focus on features extracted via static program analysis
- Develop a feature extraction pipeline for ELF binaries using static program analysis
- Train a traditional machine-learning-based malware detector for embedded IoT devices based on the extracted features
- Construct a lightweight alternative to the traditional machine-learning-based malware detector to fit the resource constrained nature of embedded IoT devices
- Analyze the best performing features