Power Allocation for Covert Wireless Communications in Fading Channels

In this paper, we consider covert communications having uncertainty about the noise variance in fading channels where an eavesdropper use a radiometer detector to detect the legitimate signal. Two power allocation policies are proposed to maximize the throughput of the legitimate user, while satisfying two different covertness constraints. In this paper, these two constraints are termed as the error probability constraint and the relative entropy constraint, respectively. It is found that the water-filling power allocation which is the optimal solution for the throughput maximization problem in conventional communications systems is not the optimal any more if the transmission needs to be kept undetectable. Moreover, the achievable throughput under the error probability constraint is higher than that under the relative entropy constraint. The theoretical findings in this paper are all validated via numerical simulations.