Enhancing IoT Security Using Multi-Layer Feedforward Neural Network with Tree Parity Machine Elements

Erekle Shishniashvili and Lizi Mamisashvili (Iv. Javakhishvili Tbilisi State University, Georgia); Lela Mirtskhulava, Lm. (Iv. Javakhishvili Tbilisi State University & San Diego State University Georgia, Georgia)

To enhance IoT security level of Tree Parity Machines, TPMs, many researchers recently tried to increase the number of neurons in a single hidden layer. In this paper we propose a novel solution of building a feedforward neural network with multiple hidden layers using elements of Tree Parity Machines. As the number of permutations of weights in these layers rises exponentially, it is almost impossible for an attacker to generate the same key, which is also proven by us using simulations, where 10,000 attacker machines try to imitate the key. While prioritizing security level and complex structures, the algorithm is time efficient as well for real time usage. We run our simulations on an Intel Core processor where key generation takes less than 1 second. After long usage of asymmetric cryptographic algorithms, the modern era requires to explore and test many new ways of generating public as well as private keys.

Conference: UKSim-AMSS 22nd International Conference on Computer Modelling and Simulation, UKSim2020

Published: Mar 25, 2020

DOI: 10.5013/IJSSST.a.21.02.37