Secure and Resilient Authenticated Encryption Approach Based on Chaotic Neural Networks and Duplex Construction

Abstract

This chapter delves deep into the intricate interplay of procedures that underline the foundations of Authenticated Encryption (AE) and its significance in preserving the confidentiality and authenticity of our digital communications. As the digital age progresses, telecommunication systems have shifted towards digital paradigms, driven not only by the affordability and accessibility of digital components but also by the inherent benefits they bring. Herein, we introduce a distinct approach from the Standard Duplex Construction (SDC) known as the Modified Duplex Construction (MDC). The MDC incorporates two pivotal phases: the initialization phase and the duplexing phase, each encompassing a Chaotic Neural Network Revised (CNNR) defined by a singular-layered neural structure enriched with non-linear functionalities. The chapter further discusses the implementation of MDC in two specific widths of 512 and 1024 bits. A rigorous evaluation of this construction against various cryptanalytic threats showcases its resilience and robustness. In this discourse, readers will encounter the development, realization, and analysis of a novel Authenticated Encryption with Associated Data Scheme (AEADS), conceived from the chaotic realms of neural networks. The chapter explicates the encryption and decryption processes of AEADS, emphasizing the crucial role of variables such as IV, K, AD, and M in encryption, and C and T in decryption. The reliability of the decryption is contingent on the alignment of computed and received tags, dictating either the decryption of the original message or the generation of an error. The decryption intricacies, encompassing variables like C, T, IV, K, and AD, are also elucidated. Two distinct processes have been instituted for message lengths spanning 64 and 128 bytes, providing a comprehensive view of the scheme’s versatility.