Abstract:Wireless signals are vulnerable to various security threats, like eavesdropping and jamming, due to the inherent broadcast nature of the wireless channel. Encryption may ensure the confidentiality of the data but does not guarantee successful communication among legitimate users in the presence of strong adversaries, like wideband jammers. In this scenario, hiding a secret signal in presence of another mundane ongoing communication is one of the ways to minimize its chances of getting intercepted. Wireless Steganography is a process of embedding a secret signal inside another signal that acts as a cover to hide the signal of interest. In this paper, we propose to encode secret bits into covert signals that are statistically indistinguishable from a hardware noise generated by a low-cost transmitter. As the covert signal resembles hardware noise, it can be transmitted over any waveform, making it adaptable and portable to any communication link. Each generated complex signal sample is merged with a cover signal sample, yielding a 50% embedding rate. We create the encoding and decoding models by creating a pair of complex-valued neural networks (NNs), which is trained in presence of another NN model, critic. The critic model differentiates between true hardware noise and encoder-generated covert signal, thus providing essential feedback to the NN pair to improve the encoding technique. The decoder undergoes a transfer learning process to adapt to the residual channel effects in over-the-air experiments. In an indoor testbed, we successfully decoded the covert communication that mimics a range of hardware noises and is transmitted using different modulation orders of cover OFDM waveform. Our steganalysis indicates that the covert signal can be generated to mimic specific hardware, which remains indistinguishable in different statistical tests. Our method performs an order of magnitude better in statistical steganalysis compared to the state-of-the-art method in this field.

Deep learning for covert communication

Covert and Reliable Semantic Communication Against Cross-Layer Privacy Inference over Wireless Edge Networks

A Comprehensive Study of Deep Learning-based Covert Communication

Deep Learning for Secure Transmission in Reconfigurable Intelligent Surface-Assisted Communications.

Covert Communication Achieved by a Full-Duplex Multi-Antenna Receiver in Wireless Networks

Covert Communication in Autoencoder Wireless Systems

Secure Energy Efficiency in Communication Systems Based on Deep Learning

Covert Communication in Two-Hop Cooperative Cognitive Radio System

Covert Communication for Spatially Sparse mmWave Massive MIMO Channels

Aider: Artificial Intelligent Based Deep Receiver for Wireless Communication Systems

Dual-Functional Artificial Noise (DFAN) Aided Robust Covert Communications in Integrated Sensing and Communications

Learning Robust Representations for Communications over Noisy Channels

Exploiting Deep Learning for Secure Transmission in an Underlay Cognitive Radio Network

Achieving Covert Communication in Underlay Cognitive Radio Networks Using a Multi-Antenna Receiver

ML-Based Detection Approaches for Covert Communication with Multi-D Signal Features

Deep Learning for Joint Noncoherent Differential Detection and Channel Decoding.

Deep Learning-Based Communication Over the Air

AIR: Threats of Adversarial Attacks on Deep Learning-Based Information Recovery

Covert Communications via Adversarial Machine Learning and Reconfigurable Intelligent Surfaces

WISE: Waveform Independent Signal Embedding for Covert Communication