In response to serious vulnerabilities that plague many of the Internet's core protocols, the last two decades have seen various security infrastructures layered on top of originally insecure protocols (DNSSEC on top of the domain name system, SSL and its public key infrastructure on top of TCP, the RPKI on top of interdomain routing). The security of each is derived from centralized authorities that are trusted to provide information about cryptographic keys or identities. When authorities behave correctly, each security infrastructure protects the underlying insecure system from attack. However, what happens if an authority abuses its power, or experiences a malfunction, misconfiguration, or a compromise by an external attacker? This project is predicated on the observation that the scope, impact, and visibility of abuse by compromised authorities is determined by the architecture of a security infrastructure (e.g., the presence or absence of hierarchy, the scope of an authority's power, etc.). The project finds new ways to balance between the need to provide strong security guarantees for a vulnerable insecure system, and the need to limit the control that a centralized authority can exercise over that system. To do this, the project will use cryptography and mechanism design to design architectures that make abuse easier to detect, or harder to execute. The project has implications on the security of some of the Internet's most crucial systems, and also involves technology transfer to the practitioners involved in the standardization and adoption of Internet systems.
CAREER: Computing on Encrypted Data