Covert Communications in a Dynamic Interference Environment

Much of security research focuses on preventing an adversary from deciphering a message's content, but there are a number of applications that motivate the more challenging goal of “covert” communications: transmitter Alice conveying information to legitimate receiver Bob while preventing a capable and attentive adversary Willie from detecting the presence of the message. Willie detects Alice's transmission by looking for deviations from the background environment; hence, covert throughput can be increased if Willie has uncertainty about that background environment, as might be caused by user mobility or time-varying jamming. Here, we consider how the dynamics of the background environment impact the throughput of covert communications. Our previous work has shown that a background environment varying a constant number of times per codeword duration can greatly increase the covert throughput. In this paper, we consider the covert throughput for all potential rates of variation of the background. Neither extreme of variation is favorable to Alice: (1) when the background environment varies too quickly relative to the codeword length, Willie can average out the variation and hence limit Alice's covert throughput; (2) when the background environment varies too slowly, Willie can estimate the current environment and again limit Alice's covert throughput. We analyze a detector at Willie that provides an upper bound (converse) to the covert throughput of Alice for all potential rates of variation. Next, we provide an explicit construction for a scheme for Alice that provides a lower bound (achievability) on the covert throughput at each rate of variation.