Hybrid fusion of machine learning algorithm and chaotic countermeasures for side channel attacks using reconfigurable architecture

Abstract

In the current literature, FPGA architectures have become prominent in every embedded system due to its speed, flexibility and reconfigurability. Mobile phones, smart cards, IoT devices, and electric cars are the diverse sectors of embedded systems. A significant number of these Embedded System-(ES) application programs handle sensitive information (e.g., smart card data on a cell phone/PDA) or perform basic capacities (e.g., therapeutic gadgets or car hardware), and the utilization of security conventions is domineering to look after secrecy, trustworthiness and authentication of these applications. This led the embedded systems to become more secured systems in the current literature. Diverse attacks become vulnerable in these systems for example, side-channel attacks, tamper attacks, fault induction attacks and many others. Elliptic Curve Cryptosystems (ECC) is the most prominent public cryptographic system in terms of security and efficiency. However, its mathematical complexity requires careful analysis and algorithm implementation to achieve efficient implementation. The well-organized execution of ECC is a multidimensional problem involving finite field arithmetic layer levels, elliptic curve grouping operations layer, and hardware and software platform specific problems. We specifically focus on standards-based ECC implementations in the FPGA platforms with the help of machine learning approaches to offer the best performance in both hardware and software implementations of ECC and side-channel stabbing detection. newline