Hossein Khosraviroshkhari, Mohammad Hossein Yaghmaee Moghaddam, and Hadi Sadoghi Yazdi
SIP, SPIT, feature extraction
VoIP (Voice over Internet Protocol) is gaining popularity due to its manifold advantages and is currently a major technology used in Wireless voice communications. This popularity has made VoIP networks attractive to the users, who use VoIP facilities to generate numerous commercial phone calls or messages (spammers). This vulnerability is named Spam over Internet Telephony (SPIT) and is assumed to be as one of the most important future threats to VoIP networks. In this paper a comprehensive solution for effective SPIT detection, is presented. The proposed framework, is a combination of effective recent solutions, besides new implementations in a modular system. The implemented modules do only rely on the features which are extracted from the SIP header information. Overall decision upon an incoming call is made in a weighting process which yields excellent decision ability. Finally, if the certainty percentage of the decision was not acceptable enough, an interferential subsystem with Audio CAPTCHA prevents the spitter to access the destination phone number. In order to test the proposed framework, the behavior of the Spitters with different skills, is simulated. Furthermore, an efficient evaluation technique is proposed which has overcome the fact that there is no valid universal SPIT database and has demonstrated the success of the presented solution.
Important Links:
Go Back
