Using Machine Learning to Detect Hidden Information Through Steganographic Techniques in the TCP/IP Network Protocols

This paper presents an algorithm and model for detecting confidential information hidden in TCP and IP network protocols header fields. Conventional tools and methods used to detect steganography are difficult to detect signs of steganographic packets. Therefore, several machine learning methods were experimented with in the study to detect the abnormal behavior of steganographic packets. Finally, a Random Forest algorithm was proposed based on the experimental results. For machine learning of steganographic packets, TCP and IP protocols packet headers use some properties of attributes. For training, network traffic capture software captures steganographic and normal packets. The formula of the relationship between the number of correctly identified results and their total number is used to evaluate the possibility of models. At the end of the study, it was shown by evaluation that the proposed method is more effective than other similar methods.