Title :
A SPIT Detection Method Using Voice Activity Analysis
Author :
Huang Hai ; Yu Hong-tao ; Feng Xiao-Lei
Author_Institution :
NDSC, Zhengzhou, China
Abstract :
Spam over IP telephony (SPIT) is becoming a large problem with the rise of VoIP. A new SPIT detection method using voice activity analysis is proposed in this paper. The calling and called voice saturation ratio and the called conflict ratio is used as the parameters of the call-behavioral characteristics after an analysis of the difference in call behavior between spam voices and the normal calls. And the parameters of the call-behavioral characteristics are obtained by analyzing voice activity status of the calling and called. Finally, by using support vector machine a spam classifier is designed and the experiment result shows the method can effectively recognize spam call.
Keywords :
Internet telephony; support vector machines; SPIT detection; VoIP; call-behavioral characteristics; called conflict ratio; normal calls; spam classifier; spam over IP telephony; spam voices; support vector machine; voice activity analysis; voice saturation ratio; Ad hoc networks; Batteries; Computer science; Computerized monitoring; Energy resources; Intrusion detection; Mobile computing; Multimedia systems; Speech analysis; Wireless networks; SPIT; SVM; VAD; call-behavioral characteristics;
Conference_Titel :
Multimedia Information Networking and Security, 2009. MINES '09. International Conference on
Conference_Location :
Hubei
Print_ISBN :
978-0-7695-3843-3
Electronic_ISBN :
978-1-4244-5068-8
DOI :
10.1109/MINES.2009.253