Secured Wireless Network Based on a Novel Dual Integrated Neural Network Architecture

Abstract

The development of the fifth generation (5G) and sixth generation (6G) wireless networks has gained wide spread importance in all aspects of life through the network due to their significantly higher speeds, extraordinarily low latency, and ubiquitous availability. Owing to the importance of their users, components, and services to our everyday lives, the network must secure all of these. With such a wide range of devices and service types being present in the 5G ecosystem, security issues are now much more prevalent. Security solutions, are not implemented, must already be envisioned in order to deal with a range of attacks on numerous services, cutting-edge technology, and more user information available over the network. This research proposes the dual integrated neural network (DINN) for secure data transmission in wireless networks. DINN comprises two neural networks based on sparse and dense dimensions. DINN is designed for any presence of deep learning-based attack in a physical security layer. DINN is evaluated considering the various machine learning attack such as basic_iterative_method attack, momentum_iterative_method attack, post_gradient_descent attack, and C&W attack; comparison is carried out on existing and DINN, considering attack success rate and MSE. Performance analysis suggests that DINN holds a higher level of security against the above attacks.