SMURFEN: A Knowledge Sharing Intrusion Detection Network

The problem of Internet intrusions has become a world-wide security concern. To protect computer users from malicious attacks, Intrusion Detection Systems (IDSs) are designed to monitor network traffic and computer activities in order to alert users about suspicious intrusions. Collaboration among IDSs allows users to benefit from the collective knowledge and information from their collaborators and achieve more accurate intrusion detection. However, most existing collaborative intrusion detection networks rely on the exchange of intrusion data which raises the privacy concern of participants. To overcome this problem, we propose SMURFEN: a knowledge-based intrusion detection network, which provides a platform for IDS users to effectively share their customized detection knowledge in an IDS community. An automatic knowledge propagation mechanism is proposed based on a decentralized two-level optimization problem formulation, leading to a Nash equilibrium solution which is proved to be scalable, incentive compatible, fair, efficient and robust. We evaluate our rule sharing mechanism through simulations and compare our results to existing knowledge sharing methods such as random gossiping and fixed neighbors sharing schemes.