Machine Learning-Based Performance Assessment of Fragmented and Non-Fragmented Encryption in Cloud Data Security

Abstract

With the rapid advancement of cloud-based and network-centric applications, securing confidential data against cyber-attacks over the communication channel has become essential. Protecting against cyber-attacks through the traditional approach will be difficult because the traditional approach encrypts the entire dataset as a single unit, which consumes more computational resources and leaves a single point of failure, leading to the compromise of the entire dataset. So Fragmented-based encryption strengthens security due to limited data exposure. This paper presents a performance evaluation of fragment-based encryption versus conventional encryption based on the performance metrics and security resistance. Here, the proposed method employs splitting the data by column, then the fragments are set by fixed-size, and those fragments are labelled and validated as sensitive using a rule-based algorithm and a Random Forest algorithm. The fragmented data is encrypted using the AES–256 algorithm in GCM Mode with PKI authentication. This analysis shows that fragmented encryption achieves optimal computational resources and improves security by limiting data exposure.