Abstract:Recently, learning (DL)-based physical-layer authentication (PLA) has attracted much attention since artificial neural networks (ANNs) can be built to extract useful features from complex wireless environments, thus achieving high authentication performance and lightweight deployment in mobile edge computing (MEC)-Industrial Internet of Things (IIoT) scenario. However, the low latency characteristic of MEC makes it impossible to have much time to obtain sufficient signals for training the authentication system, which will cause over-fitting issues and deteriorate the authentication performance. Data augmentation is an effective method to address this problem. However, existing PLA with data augmentation can not generate representative and high-quality samples, consequently lacking generality in the actual identity authentication. To tackle this problem, a novel channel impulse response (CIR)-based multiuser authentication named latent perturbed neural networks (LPNNs) is proposed in this article, aiming at achieving high authentication performance even when trained a few data. Instead of relying on the generation of synthetic samples, the proposed LPNN adds Gaussian noise in the smooth latent space to avoid underdetermined and poor generalization, which has better interpretability. Specifically, to obtain a better understanding than a black box that connects input CIRs to authentication results, we defined Fingerprint Library and provided post-hoc explanations to answer the following question: which library examples explain the authentication results issued for a given CIR sample? Moreover, the simulations under the static and dynamic IIoT scenarios verify the superiority in authentication accuracy of the proposed LPNN over vanilla deep neural network (DNN) and convolutional neural network (CNN).

Multi-User Identity Authentication in Wireless Communication Based on Deep Learning and Data Augmentation

Toward Multi-User Authentication Using WiFi Signals

MultiAuth: Enable Multi-User Authentication with Single Commodity WiFi Device

Multiuser Physical Layer Authentication in Internet of Things with Data Augmentation

Security Enhancement for Mobile Edge Computing Through Physical Layer Authentication.

Deep Learning-Enhanced Physical Layer Authentication for Mobile Devices

Reconfigurable Intelligent Surface-Aided Physical Layer Authentication with Deep Learning

Multiuser physical layer security mechanism in the wireless communication system of the IIOT

A Survey of Machine Learning-based Physical-Layer Authentication in Wireless Communications

Multi-User Physical Layer Authentication Based on CSI Using ResNet in Mobile IIoT

Authenticating Users Through Fine-Grained Channel Information

Deep-Learning-Based Physical Layer Authentication for Industrial Wireless Sensor Networks.

Multiuser Physical-Layer Authentication Based on Latent Perturbed Neural Networks for Industrial Internet of Things.

Blind Physical-Layer Authentication Based on Composite Radio Sample Characteristics.

Practical user authentication leveraging channel state information (CSI)

Multi-Attribute Weighted Convolutional Attention Neural Network for Multiuser Physical Layer Authentication in Iiot

Authenticating On-Body IoT Devices: An Adversarial Learning Approach

Channel state information based physical layer authentication for Wi‐Fi sensing systems using deep learning in Internet of things networks

Proactive User Authentication Using Wifi Signals In Dynamic Networks

Online Learning Aided Adaptive Multiple Attribute-Based Physical Layer Authentication in Dynamic Environments

Multiple Correlated Attributes Based Physical Layer Authentication in Wireless Networks.