SIEGE : Service-Independent Enterprise-GradE protection against password scans

Abstract

Security is one of the main challenges today, complicated significantly by the heterogeneous and open academic networks with thousands of different applications. Botnet-based brute-force password scans are common security threat against the open academic networks. Common defenses are hard to maintain, error-prone and do not reliably discriminate between user error and coordinated attack. In this paper, we present a novel approach, which allows to secure many network services at once. By combining in-app tracking, local and global crowdsourcing, geographic information, and probabilistic user-bot distinction through differential password analysis, our PAM-based detection module can provide higher accuracy and faster blocking of botnets. In the future, we aim to make the mechanism even more generic and thus provide a distributed defense against one of the strongest threats against our infrastructure.