Digital Hardware Implementation of Lightweight Cryptography Algorithm Using Neural Networks

摘要

The security of authentication, data communication and information exchange is extremely important and considered as an essential requirement for modern electronic systems. Neuromorphic circuits, on the other hand, are an emerging era, projecting new solutions for future computing systems based on spiking neural networks. Main idea behind this work is based on using dynamical behavior of nonlinear neuron models with different spiking patterns to extend complexity level and obfuscation of the cryptography system to achieve a higher level of security. Utilizing a novel hardware realization of the Izhikevich neuron model, a low-cost implementation of the system is presented, which can be embedded in any light-weight cryptography system. Simulation results show that the proposed model can precisely reproduce different spiking patterns, as the authentication signature of the secured system. FPGA realization of the system is presented as a proof of concept.