Grant: Discovery Grants Program - Individual

Recent reports show a trend of increasing professionalization and commercialization of malicious activities, as well as the increase in terms of the number of attacks. Malicious users are collaborating with each other so as to launch attacks more efficiently and effectively. For example, they have built underground forums for exchanging information about new exploits and underground economy servers to trade compromised resources. Unfortunately, on the other side of the attacking-defending competition, there are relatively less collaborations going on among the defenders. The major barrier is the concern that, shared security data may contain sensitive information, which may be disclosed and misused at a later time. And such information may be originally difficult to obtain by adversaries. To address this concern, in the long term, the proposed research program will develop techniques that can preserve privacy while sharing security data within collaborative detection systems, and at the same time ensure the usability of such systems, in a realistic adversarial environment. In the short time, the proposed research program will aim at achieving an appropriate balance between privacy and other properties that are critical to the practical usages of the system. Such properties include availability, accuracy (dependent on quality and quantity of shared security data), and efficiency. The proposed research will have significant impacts on boosting a variety of collaborative security applications, e.g., zero-day malware detection and spam filtering, through strong protection of the system's privacy and usability as well as incentive mechanisms. As a result, both security and cost-effectiveness (in terms of defending attacks) of computer systems, no matter within an Enterprise network or national-wide, will greatly benefit from the proposed research.