A Cryptographic Database Retrieval Scheme Based on Bucket Partition

For the current problem of Internet user privacy data leakage, data encryption storage is a viable option. CryptDB is a typical ciphertext storage technique that uses the onion encryption algorithm to rewrite SQL statements into different onion ciphertext columns based on arithmetic operations semantics, it expose only some of the attributes of the data to perform query tasks. This paper points out the security and efficiency problems of CryptDB caused by removing the randomness of the entire column of data in the retrieval process, and improves its retrieval scheme by proposing a partitioned retrieval scheme which is more effective in resisting frequency attack and efficiency. This scheme is used to partition each attribute value, which ensures that only the partition that the queried data is located in executes the RND layer decryption operation during querying, thus ensuring the data security of the remaining partitions. Experiments show that the improved scheme avoids the disadvantage of exposing the entire column of data privacy due to the decryption of the RND layer, and solves the problem of inefficiency caused by decrypting a column of data, which is more secure and efficient.