Privacy-preserving k-NN interpolation over two encrypted databases

Area of Science:

• Cloud Computing Security
• Database Management
• Cryptography

Background:

• Cloud computing offers benefits like cost savings and accessibility but raises security and privacy concerns.
• Encrypted data in the cloud hinders complex operations like query processing.
• Existing k-nearest neighbor algorithms on single encrypted databases face accuracy and reliability issues.

Purpose of the Study:

• To address the limitations of single-database k-nearest neighbor queries in cloud environments.
• To develop a secure protocol for collaborative query processing over multiple encrypted databases.
• To enhance the accuracy and reliability of k-nearest neighbor search in cloud computing.

Main Methods:

• Introduced a secure two-party k-nearest neighbor interpolation protocol.
• Enabled query owners to extract k-nearest neighbors from two distinct encrypted databases.
• Ensured data and query point confidentiality and hidden data access patterns.

Main Results:

• The proposed protocol effectively processes k-nearest neighbor queries over two encrypted databases.
• Demonstrated protection of data confidentiality, query point privacy, and access pattern obfuscation.
• Experimental results confirmed the protocol's efficiency, with running time linearly dependent on neighbor count and data size.

Conclusions:

• Collaborative query processing across multiple encrypted cloud databases improves accuracy and reliability.
• The developed secure two-party protocol offers a viable solution for privacy-preserving k-nearest neighbor search in the cloud.
• The protocol's linear scalability makes it suitable for large datasets and complex query requirements.