Abstract:

As one of the key technologies to optimize storage space, improve network bandwidth and reduce overall overhead, data deduplication has been an indispensable part of cloud service provider (CSP) solutions on outsourced data management, but it also faces many security issues, such as data confidentiality, integrity and privacy. This paper proposes a secure data deduplication scheme which integrates fault tolerance, confidentiality and efficient key management. The scheme adopts a secret sharing algorithm based on a permutation ordered binary (POB) number system to decompose the data block into multiple random shares, and enhances data security by introducing the proof of ownership (PoW) concept. Moreover, The scheme applies a secret sharing algorithm based on the Chinese Remainder Theorem (CRT) to divide the key into multiple random blocks and sends them to the corresponding key management server (KMS) to minimize the key overhead. Experimental results show that the scheme overwhelms the other schemes in terms of function and efficiency and can effectively resist two types of attackers (i.e. dishonest servers and external attackers) and two types of attack modes (i.e. duplicate faking attacks and erasure attacks).

Key words: data deduplication, secret sharing, proof of ownership (PoW), permutation ordered binary (POB), Chinese Remainder Theorem (CRT)