Adaptive Linear Secrecy Codes with Feedback

Abstract

We identify a novel method of using feedback to provide enhanced information-theoretical security in the presence of an eavesdropper. This method begins with a fixed linear coset code providing both secrecy and error detection/correction, as has been described by several authors. The legitimate receiver then sends the syndrome information for the received codeword, and based on this feedback, the transmitter can provide further error correction information specifically tailored to assist only the legitimate receiver. We show that this method allows secure communication with the legitimate receiver even when the eavesdropper’s channel is superior to that of the legitimate receiver.