量子 oracle 分布切换及其在完全匿名环签名中的应用
Quantum Oracle Distribution Switching and Applications to Falcon and Ring Signatures
AI总结:
本文在量子可访问随机 oracle 模型中,为两种通用环签名构造提供了四个安全性缩减,以提升后量子安全性的理论基础。
摘要:
环签名是一种强大的原始工具,允许成员代表一组人签署,而无需透露其身份。最近,环签名因其作为后量子否认认证密钥交换的成分而受到更多关注,例如用于后量子版本的信号协议,该协议被几乎所有端到端加密信息服务所采用。尽管几种基于后量子假设的环签名构造提供了适用于否认密钥交换的合适安全性和效率,但目前它们仅在随机 oracle 模型(ROM)中被证明是安全的,这对于后量子安全来说是不够的。在本工作中,我们为两种通用的环签名构造提供了四个安全性缩减,这些构造在量子可访问的随机 oracle 模型(QROM)中:两个针对 AOS 框架,两个针对基于环陷阱门的构造范式,其通用骨架我们进行了形式化。AOS 环签名的两个安全性证明在对底层 sigma 协议的要求和紧致性上有所不同。环陷阱门基于的环签名的两个缩减在要求和提供的安全性上表现出各种差异。我们采用测量和重编程技术,基于压缩 oracle 的 QROM 直线提取工具、无历史缩减和 QROM 重编程工具。为了利用 QROM 中的 Rényi 散度属性,我们研究了与具有基于两种不同分布的输出集的 oracle 交互的量子算法的行为。我们提供了统计距离的紧致界,证明了 Rényi 散度不能用于替代整个 oracle,并提供了一个替代方案。
英文摘要:
Motivated by digital signature algorithms ranging from Falcon to fully-anonymous ring signatures used in Signal-style key exchange, such as Gandalf, we revisit a fundamental problem in post-quantum security proofs: distinguishing oracle functions whose outputs are sampled independently from distributions $P$ and $Q$ that are close. In the context of signatures, closeness is often measured via Rényi divergences, which yield multiplicative bounds in the classical setting. A counterexample shows that such multiplicative-error bounds for distinguishers with classical oracle access have no analogue for quantum access, and we provide two alternative approaches based on small-range distributions and reprogramming techniques. We also give a concrete, optimal bound for the case where $P$ and $Q$ are close in statistical distance. We apply these techniques to the motivating constructions. (i) We give the first QROM security proof for Falcon that avoids oracle indistinguishability arguments based on statistical distance. This is crucial, as Falcon's ROM proof relies on Rényi divergence, while the statistical distance induced by its parameters is too large to yield meaningful bounds. (ii) We formalize and abstract the ring signature construction used in Gandalf as a modular framework by defining ring trapdoor preimage-sampleable functions (RPSFs), for which we obtain two QROM proofs. We also provide two QROM security proofs for AOS ring signatures, adapting existing QROM techniques. Together with our results on RPSF-based ring signatures, this yields QROM security proofs for a broad class of fully-anonymous linear ring signature constructions, including Gandalf and the AOS-based constructions Erebor and MayoRS.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。