Secure transmissions via compressive sensing in multicarrier systems

In this paper, we consider an encryption scheme based on compressive sensing for multicarrier systems, where artificial noise is selectively transmitted together with a sparse message signal in the frequency domain to make an adversary's attack difficult. In order to minimize the impact of artificial noise on the performance of a legitimate receiver, the channel state information at the transmitter is exploited under the assumption of channel reciprocity in time division multiplexing mode. We consider attack to estimate the measurement matrix using the maximum likelihood approach and find the probability of correct recovery (or successful attack) under certain assumptions which might be favorable for the adversary.