Unified GPU Technique to Boost Confidentiality, Integrity and Trim Data Loss in Big Data Transmission

Abstract

Data integrity, confidentiality and data loss are the issues that arise during transmission because of the use of an inadequate security scheme. These issues become particularly critical for big data transmission due to its own individual overhead causes. Moreover, multiple executions of distinct security algorithms for maintaining confidentiality and integrity reduce throughput and add a large number of additional bits as security overhead that hampers the robustness against data loss. Conversely, an efficient compression technique minimizes data confidentiality, as it eliminates redundant data during compression. Contemporary studies shows the lack of security policies for solving the mentioned issues in a combinatorial manner. The current study proposes an innovative integrated technique to collectively addresses the above security issues. It increases confidentiality and offers a backup for accidental data loss by combining the simplified data encryption standard (SDES) and an advanced pattern generation technique that uses a unique pattern generation table. A novel dual round of error control technique has been introduced to maximize data integrity by considering an arbitrary number of transmission errors. A new compression technique is adopted to enhance the robustness against data loss along with high compression efficiency and resistance against transmission errors. Confidentiality and integrity are further enhanced by integrating advanced audio steganography that uses a distinctive sample selection for hiding bits. Additionally, the implementation of the proposed innovative integrated technique in the graphics processing unit (GPU) environment increases the execution speed and reduces time complexity with extended parallel processing power. Furthermore, the application of a GPU enhances the execution speed at least 28-fold compared to the CPU performance. Experiments are performed using the standard Calgary Corpuses, text files (sized up to 1 TB), and audio files to validate the objectives. The proposed method offers a higher signal-to-noise ratio (SNR), entropy, and avalanche effect (AE) and lower amplitude difference (AD), and uncorrectable error rate (UER) as well as a lower percentage of information loss (IL), which substantiates its potential to offer higher data confidentiality and integrity. The capacity to reduce the computational complexity is further measured with compression ratio (CR) and throughput. The results further depicts the method's superiority in offering confidentiality and integrity over contemporary approaches.