Research on Big Data Security Storage Based on Compressed Sensing

Abstract

Big data have the characteristics of volume and variety, which causes huge storage and communication overhead. The traditional encryption system has a high computational complexity and cannot effectively implement the secure storage for the big data. In order to solve these problems, a big data secure storage scheme based on compressed sensing (CS-BDCSS) is proposed. First, the primary encryption of plaintext data is realized by the Data Combined Crossover Random Permutation method, and then, the security-enhanced CS is used to compress and re-encrypt the data to reduce the space overhead and achieve secure storage. Aiming at the problem that the key matrix of the traditional CS algorithm is vulnerable to attack, a Hierarchical Security Key Matrix Generation Scheme is proposed; by expanding the secret key space, the hierarchical security protection for sensitive data of different levels is realized. The concept of pseudo-homomorphism encryption is proposed, and the computation based on ciphertext is realized by using the pseudo-homomorphism encryption, which can improve the efficiency of the data analysis on the basis of security. An adaptive secret key-updating mechanism is designed to automatically update the key in CS-BDCSS system so as to avoid the security risk caused by the repeated use of the same key. The experimental results show that the proposed scheme can achieve efficient data compression and secure encryption storage.