Date of Publication :13th February 2018

Abstract: With the rapidly increasing amounts of data produced worldwide, networked and multi-user storage systems are becoming very popular. However, concerns over data security still prevent many users from migrating data to remote storage. The conventional solution is to encrypt the data before it leaves the owner’s premises. While sound from a security perspective, this approach prevents the storage provider from effectively applying storage efficiency functions, such as compression and deduplication, which would allow optimal use of the resources and consequently lower service cost. Client-side data deduplication, in particular, ensures that multiple uploads of the same content only consume network bandwidth and storage space of a single upload. Deduplication is actively used by a number of cloud backup providers as well as various cloud services. Unfortunately, encrypted data is pseudorandom and thus cannot be deduplicated: as a consequence, current schemes have to entirely sacrifice either security or storage efficiency. In this paper, we present schemes that permit a more fine-grained trade-off in data chunk similarity. The intuition is that outsourced data may require different levels of protection, depending on how popular it is: content shared by many users. Various deduplication schemes are analyzed and provide experimental results that show proposed secure data chunk similarity provide improved results in real time cloud environments

Reference :