Privacy-Preserving Approximate <i>k</i>-Nearest-Neighbors Search that Hides Access, Query and Volume Patterns

Abstract We study the problem of privacy-preserving approximate kNN search in an outsourced environment — client sends encrypted data to untrusted server and later can perform secure updates. design a security model propose generic construction based on locality-sensitive hashing, symmetric encryption, oblivious map. The provides very strong guarantees, not only hiding information about data, but also access, query, volume patterns. implement, evaluate efficiency, compare performance two concrete schemes AVL tree BSkiplist.