Ciphertext-Only Attack on a Secure $k$-NN Computation on Cloud

云计算的兴起推动了将数据存储和计算任务转移到云上的趋势。为了保护诸如客户数据和业务细节之类的机密信息，必须在云存储之前对这些敏感数据进行加密。实施加密可以防止未经授权的访问、数据泄露以及由此产生的财务损失、声誉损害和法律问题。此外，为了方便在云存储的数据上执行数据挖掘算法，加密需要与领域计算兼容。在基于位置的服务等领域中，特定查询向量的k最近邻（k-NN）计算被广泛使用。Sanyashi等人（ICISS 2023）提出了一种加密方案，利用非对称标量积保持加密（ASPE）来促进云上的隐私保护k-NN计算。在本文中，我们发现Sanyashi等人的上述加密方案存在重大漏洞。具体而言，我们提供了一个高效的算法，并通过实验证明他们的加密方案容易受到仅密文攻击（COA）的攻击。

Sanyashi et al. proposed an encryption scheme for privacy-preserving k-NN computation on the cloud, but the paper identifies a vulnerability in the scheme that makes it susceptible to ciphertext-only attacks (COA). The paper aims to address this vulnerability.

The paper proposes an improved encryption scheme that uses a randomized scalar product preserving encryption (RSPE) to protect the privacy of data in cloud storage while still allowing for efficient k-NN computation. The new scheme is resistant to COA.

The paper presents an improved encryption scheme that is resistant to COA. The authors demonstrate the effectiveness of the new scheme using experiments on both synthetic and real-world datasets. The authors also provide a comparison with other related works. The proposed scheme is computationally efficient and can be applied to other machine learning algorithms.

Related works include 'Privacy-Preserving k-NN Computation on Cloud: A Survey' by Shao et al. and 'Privacy-Preserving k-NN Computation on Cloud with Secure Data Transformation' by Yu et al.