Labeled VoIP Data-Set for Intrusion Detection Evaluation

VoIP has become a major application of multimedia communications over IP. Many initiatives around the world focus on the detection of attacks against VoIP services and infrastructures. Because of the lack of a common labeled data-set similarly to what is available in TCP/IP network-based intrusion detection, their results can not be compared. VoIP providers are not able to contribute their data because of user privacy agreements. In this paper, we propose a framework for customizing and generating VoIP traffic within controlled environments. We provide a labeled data-set generated in two types of SIP networks. Our data-set is composed of signaling and other protocol traces, call detail records and server logs. By this contribution we aim to enable the works on VoIP anomaly and intrusion detection to become comparable through its application to common datasets.