Paper 2024/1555

Private Laconic Oblivious Transfer with Preprocessing

Abstract

Laconic cryptography studies two-message protocols that securely compute on large amounts of data with minimal communication cost. Laconic oblivious transfer (OT) is a central primitive where the receiver's input is a large database $\mathsf{DB}$ and the sender's inputs are two messages $m_0$, $m_1$ along with an index $i$, such that the receiver learns the message determined by the choice bit $\mathsf{DB}_i$. OT becomes even more useful for secure computation when considering its laconic variants, which offer succinctness and round optimality. However, existing constructions are not practically efficient because they rely on heavy cryptographic machinery and non-black-box techniques. In this work, we initiate the study of laconic OT correlations, where the model allows an offline phase to generate the correlations later used in a lightweight online phase. Our correlation is conceptually simple, captured by an inner product computation, and enables us to achieve a private laconic OT protocol where the sender's index $i$ is also hidden from the receiver. Our construction is the first private laconic OT with database-dependent preprocessing based solely on symmetric-key assumptions, achieving sublinear online computational complexity for the receiver. Furthermore, we enhance our construction with updatability and receiver privacy. Finally, we demonstrate the applications of private laconic OT to laconic function evaluation for RAM programs and laconic private set intersection with preprocessing.