MIMO Wireless Secure Communication Using Data-Carrying Artificial Noise

Abstract

In MIMO wireless channels, channel estimation can provide a source of randomness common to sender and receiver that may be used to generate secret keys. These keys can provide information theoretically secure communication. Alternatively, a known MIMO channel allows transmitting artificial noise in the null space of the main channel to secure communication. In this paper, we show that both methods can be combined to further enhance secrecy. Symbols encrypted at the transmitter simultaneously act as message for the intended receiver and as noise for any potential eavesdroppers. We derive expressions for the minimum-guaranteeable secrecy rate as a function of number of symbols encrypted. Given at least one symbol can be encrypted, our proposed method outperforms the standard artificial noise scheme in terms of both secrecy rate and power expenditure. We first analyze the system assuming keys can be instantaneously extracted from channel measurements. We then derive bounds for rates assuming the overhead time required for key generation produces a Gaussian channel-estimation error.