Improving Robustness of PGP Keyrings by Conflict Detection

Abstract

Secure authentication frequently depends on the correct recognition of a user's public key. When there is no certificate authority, this key is obtained from other users. If such users can be malicious, trusting the key information they provide is risky. Previous work has suggested the use of redundancy to improve the trustworthiness of userprovided key information. In this paper, we build on this work and address two issues not previously considered. First, we explain and solve the problems presented by users who falsely claim multiple identities, or who (possibly legitimately) possess multiple keys. Secondly, we show that conflicting certificate information can be exploited to increase the reliability of the key information provided by users. Our methods are demonstrated on actual and synthetically-generated webs of trust (PGP keyrings) and their performance is discussed.

Shared Risk Resource Group Complexity and Approximability Issues

“…The Maximum Number of Color Disjoint st-Paths Problem has been proved NP-Hard in [13] using a reduction from 3SAT. This reduction fits for the case of colors of span at most two.…”

“…Indeed, they can model correlated traffic jams of road networks, as well as cascading failure scenarios in electrical networks. Related concepts are also relevant in the settings of cryptographic PGP (Pretty Good Privacy) keyrings [13].…”

“…It follows that minimizing the total number of colors of 2-Color Disjoint st-Paths is difficult. Moreover, it has been shown in [13] for a cryptographic application, that the problem Maximum Number of Color Disjoint Paths consisting in finding the maximum number of color disjoint st-paths connecting two vertices is also NP-Hard.…”

k-Anonymous data collection

“…Note that this kind of assumptions (of no collusion between certain participants) have been widely used in cryptography (e.g., in [31,48]) and in privacy-preserving data mining (e.g., in [33,55]). The reason is that the involved participants belong to different organizations and have different interests; thus, it is natural to assume they are unwilling to collude with each other to violate anybody's privacy.…”

Two methods for privacy preserving data mining with malicious participants

“…In this case, each adversary can have arbitrary malicious behavior, but his malicious behavior is independent of other adversaries'. An example is[14], in which the 4.1 considers non-colluding malicious adversaries.…”