iUC-Secure Distributed File Transfer from Standard Attribute-Based Encryption

Abstract

Attribute-Based Encryption ($\textsf {ABE}$) stands as a cryptographic cornerstone, enabling access control to messages based on user attributes. The security definition of standard $\textsf {ABE}$ is shown to be impossible in Universal Composability (UC) against an active adversary. To overcome this issue, existing formal UC security definitions of $\textsf {ABE}$ rely on additional properties for $\textsf {ABE}$, necessary to prove security against an active adversary, excluding standard $\textsf {ABE}$ by definition. In light of the composability feature offered by UC and the absence of ideal functionality tailored for standard $\textsf {ABE}$, we propose the two following contributions: (1) We construct the first ideal functionality $\mathcal {F}_{\textsf{ABE}}$ for $\textsf {ABE}$ which, under reasonable hypothesis against static corruption, can be realized using an $\text {IND-CCA2}$-secure $\textsf {ABE}$ scheme; and (2) our $\mathcal {F}_{\textsf{ABE}}$ leads us to propose a protocol solving a simple yet highly practical, world-scaled company-focused problem: efficient file transfer. The proposed construction provides data integrity, sender authentication, attribute-based file access, featured with constant data size transferred between users. This is achieved by relying on two efficient building blocks: $\textsf {ABE}$ and signature, which are layered atop of the hash-based distributed storage system $\textsf{IPFS}$. Our protocol, strengthened by a formal security definition and analysis under the Universally Composable (UC) framework called iUC, is proved to realize our problem-oriented authenticated attribute-based file transfer ideal functionality. Finally, we implement our proposal with a proof-of-concept written in Rust, and show it is practical and efficient.