Multi-Dimensional Agile Waveform Design for Covert Communications

Abstract

In adversarial wireless environments, communication systems are increasingly vulnerable to interception and detection due to the statistical regularities of transmitted signals. Conventional approaches, such as frequency-hopping spread spectrum, introduce randomness only in a limited domain and thus provide insufficient protection against advanced detection techniques. To address this issue, this work proposes a multidimensional waveform agility design that jointly exploits the frequency, time, modulation, and code domains to destroy signal stationarity and enhance covert transmission security. Domainspecific covertness and reliability metrics are established, and a unified convex optimization problem is formulated to determine the optimal resource distributions with full cross-domain decoupling, enabling efficient parallel computation. A cryptographic sequence generation mechanism is further developed to realize the optimized distributions in practical waveform generation. Simulation results show that the proposed design achieves a covertness level of 0.99, significantly outperforming single-domain frequency hopping (0.81) and the static waveform baseline (0.63), while maintaining reliable communication. Moreover, the modulation recognition accuracy remains below $\mathbf{1 0} \boldsymbol{\%}$ against deep learning-based classifiers, validating its strong anti-analysis capability of the proposed design.