Reputation Systems for Trust Management in the Web PKI

This thesis’ goal is to reduce the attack surface of the existing Web Public Key Infrastructure (PKI) by applying user trust as in real world scenarios of human trust establishment. Reputation systems can be used to exchange Certification Authority (CA) trust information and lower the risk on relying on malicious CAs. The matter in question is connection trust, determining whether a SSL/TLS connection between a client and a server is secure because all involved certificates were issued by trustworthy CAs. A secure connection prevents attackers from reading and manipulating data during a transmission over this connection. Connection trust is no assurance for the trustworthiness of a server, for example the server could run an online shop selling poor quality products. In this thesis, it is discussed how existing reputation system approaches fit into the current Web PKI scenario or how they have to be adapted. The reputation system has to be secure against attacks, since it is an attack vector for connection trust as CAs themselves. Therefore an analysis framework fitting CA reputation system attacks is proposed in this thesis. Within the evaluation, all presented reputation system approaches are compared and evaluated by the attack analysis framework. Both, centralized and distributed CA reputation systems, are discussed. More details about the centralized solution called CA-TMS can be found in ”CA Trust Management for the Web PKI”[4], of which portions of this thesis consist and which contribution was part of this thesis work.