Field-Level Lightweight Encryption Using Record-Dependent Diffusion in a C# and MySQL Environment

Abstract

In this paper, a resource-efficient, lightweight encryption framework to safeguard sensitive data at the field level in relational databases with minimal Central Processing Unit (CPU) and memory overhead is proposed. The framework uses a multi-layer architecture of encryption with the database Record Identifier (RowID) as an active cryptographic modification and "seed" in a cascading diffusion pipeline. Thus, two identical plaintexts can generate different plaintexts. The proposed scheme was tested in a native C-Sharp (C#) environment with a preliminary dataset of 10,000 database records for evaluation of the latency, throughput, diffusion strength and resource consumption. The encryption latency of 9 milliseconds (ms) and decryption latency of 8 ms gives a total of 17 ms of processing time for AES-128 and Blowfish were about 4.1 and 3.8 times slower, respectively, under the evaluated conditions. The average Avalanche Effect was 48.91% (close to the theoretical optimum of 50%), and the highest volatile memory consumption was limited to 393 Kilobytes (KB). These results suggest that the proposed framework offers a feasible compromise between security and efficiency. It's also relatively lightweight in terms of computer resources and memory usage, which could make it a good choice for enterprise database applications with limited hardware resources.