Signalling for covert passive sensing

Abstract

In this work, we consider the optimality of signalling for covert sensing, where a legitimate receiver intends to estimate unknown variables based on the received signals from a transmitter, while ensuring that the probability of these signals being detected by a warden Willie is negligible. Specifically, we consider additive white Gaussian noise (AWGN) for both the estimation channel and detection channel, based on which we first reveal that Gaussian signalling is not optimal in terms of maximizing the estimation accuracy (e.g., maximizing the Fisher information) subject to a covertness constraint, e.g., guaranteeing a Kullback-Leibler divergence being no large than a specific value. To this end, we explicitly show that a skew normal distribution with an optimized skew parameter can achieve a higher estimation accuracy than a normal distribution subject to the same covertness constraint. Furthermore, we develop a framework based on calculus of variations and the Runge-Kutta method to identify the optimal signalling for covert sensing. As expected and explicitly shown in our numerical results, the identified optimal signalling outperforms both the normal and skew normal distributed signalling, which demonstrates the necessity of optimizing signalling in the context of covert sensing.

Document Type

Conference Proceeding

Date of Publication

1-1-2023

Volume

2023-May

Publication Title

IEEE International Conference on Communications

Publisher

IEEE

School

School of Science / ECU Security Research Institute

RAS ID

62736

Comments

Zhang, Q., Yan, S., Shu, F., Cong, Y., & Ng, D. W. K. (2023). Signalling for covert passive sensing. In IEEE International Conference on Communications (pp. 3902-3907). IEEE. https://doi.org/10.1109/ICC45041.2023.10279567

Copyright

subscription content

First Page

3902

Last Page

3907

Share

 
COinS
 

Link to publisher version (DOI)

10.1109/ICC45041.2023.10279567