Privacy Preserving Audit Proofs

Abstract: The increased dependence on computers for critical tasks demands sufficient and transparent methods to audit its execution. This is commonly solved using logging where the log must not only be resilient against tampering and rewrites in hindsight but also be able to answer queries concerning (non)-membership of events in the log while preserving privacy. Since the log cannot assume to be trusted the answers must be verifiable using a proof of correctness. This thesis describes a protocol capable of producing verifiable privacy preserving membership proofs using Merkle trees. For non-membership, a method used to authenticate Bloom filters using Merkle trees is proposed and analyzed. Since Bloom filters are a probabilistic data structures, a method of handling false positives is also proposed.