Effective Webshell Detection in IoT Networks Employing TF-IDF, Word2Vec and SVM

The Internet of Things is the giant interconnection of sensor devices that allows data sharing, intelligent identification of networked “things,” and administration of these “things.” With increasing pervasiveness, the gadgets in the Internet of Things are playing a very important part in improving the lives of an individual. On the other hand, wide diffusion of Internet of Things devices makes them much more vulnerable to different types of attacks emanating from the Internet, which may result in privacy breaches, data manipulation, or disastrous damage to human beings and society. The stated importance of network security for IoT systems is because one of the main threats is caused by web injection, mainly through web shells. In this paper, we use popular machine learning models for webshell detection and to build secure solutions in IoT networks. This will enable us to develop a more secure Internet of Things system. After the extraction of character-level features using both Term Frequency-Inverse Document Frequency and Word2Vec approaches, classification is done using a support vector machine. It is illustrated through simulation results that the proposed method outperforms those that are regarded as state-of-the-art.