New Frontiers of Attribute-Based Encryption via a General Paradigm and More

Abstract

Attribute-based encryption (ABE) is an advanced form of public-key encryption incorporating fine-grained access control. In such a system, keys and ciphertexts are associated with attributes $x$ and policies $f$, respectively, and decryption is conditioned on $x$ satisfying $f$. Designing ABE schemes is challenging and its objectives include expressiveness, succinctness, efficiency, achieving strong security, and relying on minimal assumptions. This dissertation pushes the frontiers of ABE in terms of these objectives separately and jointly and studies the interaction among them. In the first part, we propose a general paradigm that greatly simplifies the task of constructing ABE schemes. It reasonably distributes the complexities into constituents, making each ingredient and the overall scheme easier to understand, reason about, and potentially improve. It is also versatile and powerful. The benefits are demonstrated by four different instantiations, which achieve various ABE schemes with improved objectives and are related to each other by replacements of ingredients. In the second part, we push the frontiers of ABE outside the paradigm. In one chapter, we resolve a long-standing open problem of constructing depth-unbounded ABE from lattices. In the other, we present the first systematic study of the upper/lower bounds of ABE succinctness and efficiency, showing inherent trade-offs among the objectives and constructing a few Pareto-optimal schemes.