Machine Learning-based Framework for Malware Detection in Critical Infrastructures for Smart Cities

Technology is an inseparable part of today’s world and is extensively used in every conceivable domain to provide a wide range of applications. Technology has many critical infrastructures, which serve the function of delivering crucial services essential for the smooth functioning of smart cities. These infrastructures produce a large amount of data which is extremely sensitive and very potentially catastrophic if it falls into the wrong hands. Various techniques and concepts have been introduced and developed to deal with the detection of malicious data. But with the passing time, malware threats grow in both frequency and complexity, machine learning and artificial intelligence techniques can be combined to enhance malware detection. Many Machine Learning algorithms exist which may be confusing for users. Different algorithms vary in their results, accuracy, and prediction capabilities depending on the data they are trained and tested on. For data produced by critical infrastructures, the prediction must be as accurate as possible. Encouraged by the aforementioned facts, we propose an approach aiming to contrast and compare different machine learning algorithms for malware detection on portable executable files dataset. Different performance metrics are compared for all the considered algorithms, based on which, the best algorithm for critical infrastructure malware detection can be determined.