IoT-based Unstructured Data Deduplication Framework using Attention-based DenseRNN with Encryption

Abstract

Due to the rapid expansion of digital data, centralized resources are offered by cloud computing systems to manage the data. Among its various services, data storage is one of the most widely adopted services all over the world. Data deduplication is a fundamental technology in cloud storage systems that help to save space by identifying and eliminating redundant copies of data. To maintain confidentiality, users encrypt their data before uploading it, but most of the traditional deduplication methods are ineffective when data is encrypted. As a result, optimizing deduplication with data encryption has been investigated in many research works to enhance storage efficiency and minimize network bandwidth usage. A major concern in this area is improving the security of deduplication mechanisms to defend against issues like data tampering, file ownership impersonation and fake duplicate generation. Addressing these limitations is essential for advancing robust and secure data deduplication methods in cloud environments. So, an efficient unstructured data deduplication mechanism is designed in the research work with novel techniques. Initially, unstructured data required for the validation is sourced from benchmark resources. Next, the unstructured data is converted into structured data with the help of entity. Then, data deduplication procedures are executed in the structured data using an Attention-based Dense Recurrent Neural Network (ADRNN). Further, the deduplicated data is encrypted using Elliptical Curve Cryptography (ECC) and stored in the cloud platform. Later, various experiments are executed to verify the overall efficiency of the developed framework over existing techniques.