An enhanced privacy preserving framework for data sanitization and restoration

Abstract

Autism Spectrum Disorder (ASD), a neurodevelopmental disorder, is often unveiled in early childhood. This research utilized autism datasets of 24-, 30-, 36-, and 48-months old babies. Some common attacks, such as spoofing, spamming, phishing, novel network attacks, content-based attacks, distributed service attacks, data mining-based attacks, and re-identification threats, affect these data. There are also some popular attacks, for example, the Known Cipher Attack (KCA), Known Plaintext Attack (KPA), Chosen Cipher Attack (CCA), and Chosen Plaintext Attack (CPA). So, data privacy is a critical concern while transferring data to prevent cyber criminals from altering, interrupting, or stealing the information. Consequently, to protect data, researchers employ a variety of techniques, including software or hardware disk encryption, data erasure, data masking, backup, and various encrypted and decrypted algorithms, such as Data Encryption Standard (DES), Advanced Encryption Standard (AES), Blowfish, International Data Encryption Algorithm (IDEA), Rivest Cipher 4 (RC4), and others. A number of these research works make use k-anonymity and query, which need a significant amount of time and substantial computational resources. Moreover, the researchers are employing optimization algorithms to improve the privacy issue. However, there are some limitations, such as no specific duration for updating the key value during the key generation step, not mentioning the key length based on which value, not defining the values of the parameters, an undefined number of key ranges and inappropriate fitness functions. To address these critical and significant concerns, this research proposed a meta-heuristic algorithmic framework called the Enhanced Combined PSO-GWO (Particle Swarm Optimization-Grey Wolf Optimization) Framework. This framework employed two techniques, which are data sanitization and data restoration procedures. Initially, the study creates optimal key, which is employed in the data sanitization process. After that, the same key is employed in the restoration process also to restore the data. This study compared the performances of the proposed framework with the traditional algorithms, such as PSO (Particle Swarm Optimization), GA (Genetic Algorithm), DE (Differential Evolution), CSA (Crow Search Algorithm), and AAP-CSA (Adaptive Awareness Probability-based CSA) against the abovementioned popular attacks and achieved better performances. From the simulation for sanitizing data, it is revealed that the proposed technique, in terms of KPA attack, attained 99.87%, 99.77%, 99.47%, 99.26%, and 99.72%, which are more significantly improved over PSO, GA, DE, CSA, and AAP-CSA, respectively, over the 30 months autism dataset, mostly among the other types of autism data. On the other hand, for restoring data, the model shows from the simulation that it achieved 99.89%, 99.81%, 99.54%, 99.37%, and 99.76%, which are enhanced over PSO, GA, DE, CSA, and AAPCSA, respectively, under the 30 months autism child dataset, mostly among the other types of autism data.