A Novel Prevention Technique Using Deep Analysis Intruder Tracing with a Bottom-Up Approach Against Flood Attacks in VoIP Systems

Abstract

AbstractThe Voice over Internet Protocol (VoIP), which is the least expensive system used for voice communication has taken another direction when a layer of vulnerabilities is created to detain its progress. Amongst the many threats described by the Security Alliance official document, the flood attacks are the most difficult ones for end-point servers. In this paper, a deep analysis method for tracing intruders to VoIP networks using a bottom-up approach is been considered. The model can address small rate and high rate attacks with different behaviour. For a 16-s timeframe set by the firewall for the broadcast of each packet to the Session Initiation Protocol (SIP) server in a time interval of 500 ms each, a false alarm rate of zero and a detection rate of 100% for the entire timeframe are observed, irrespective of the attack rates. The proposed model has a performance accuracy of 98.7% in each timeframe of 500 ms, which is better than the value prescribed in previous work from the literature.KeywordsDenial of Service (DoS)Deep analysisDeep learningFlood attacksIntrusion Detection and Prevention System (IDPS)Session Initial Protocol (SIP)Voice over Internet Protocol (VoIP)