New tools in cryptography: mutually independent commitments, tweakable block ciphers, and plaintext awareness via key registration

Cryptographic primitive research comprises much of the body of modern cryptographic research. This kind of research aims to produce useful primitives – building blocks with which we can design systems to enforce complex and subtle security requirements. Cryptographic primitive research often involves noticing a new problem that needs a simple solution, or a different way of setting up a problem; these in turn give rise to precise definitions for a primitive. In addition to this, cryptographic primitive research aims to provide the best, most efficient techniques for implementing each primitive, based on minimal computational assumptions. Our proposed research involves highlighting the advantages of this method of cryptographic research through certain specific examples. In particular, we plan to explore the following: • By refining our notion of what is necessary in a particular definition, we may achieve marked improvements in efficiency. The right tools are not always those that offer the most security or the strongest guarantees. This is exemplified by our work on mutually independent commitments. • By being realistic about the requirements of larger systems, we can design primitives which have the benefit of breaking down the job of solving larger problems in a better way. This simplifies the ultimate goal and allows us to focus our work on improving the efficiency of primitives towards improving certain important primitives. This is exemplified by our work on tweakable block ciphers. • By matching up cryptographic definitions with notions that appear in areas outside of traditional, computational complexity-based cryptography research, we can achieve a synergy that allows us to benefit from work others have done to solve problems we are both interested in. This is exemplified by our work on plaintext awareness via key registration. Thesis Supervisor: Silvio M. Micali Title: Professor