Power Allocation and Time-Domain Artificial Noise Design for Wiretap OFDM with Discrete Inputs
Optimal power allocation for orthogonal frequency division multiplexing (OFDM) wiretap channels with Gaussian channel inputs has already been studied in some previous works from an information theoretical viewpoint. However, these results are not sufficient for practical system design. One reason is that discrete channel inputs, such as quadrature amplitude modulation (QAM) signals, instead of Gaussian channel inputs, are deployed in current practical wireless systems to maintain moderate peak transmission power and receiver complexity. In this paper, we investigate the power allocation and artificial noise design for OFDM wiretap channels with discrete channel inputs. We first prove that the secrecy rate function for discrete channel inputs is nonconcave with respect to the transmission power. To resolve the corresponding nonconvex secrecy rate maximization problem, we develop a low-complexity power allocation algorithm, which yields a duality gap diminishing in the order of O(1/\sqrt{N}), where N is the number of subcarriers of OFDM. We then show that independent frequency-domain artificial noise cannot improve the secrecy rate of single-antenna wiretap channels. Towards this end, we propose a novel time-domain artificial noise design which exploits temporal degrees of freedom provided by the cyclic prefix of OFDM systems {to jam the eavesdropper and boost the secrecy rate even with a single antenna at the transmitter}. Numerical results are provided to illustrate the performance of the proposed design schemes.