Secure Communication Under Channel Uncertainty and Adversarial Attacks

Information theoretic approaches to security have been examined as a promising complement to current cryptographic techniques. Such information theoretic approaches establish reliable communication and data confidentiality directly at the physical layer of a communication network by taking the properties of the noisy channel into account leading to unconditional security regardless of the computational capabilities of eavesdroppers. The provision of accurate channel state information is a major challenge particularly in wireless communication systems, especially information about the channels to eavesdroppers. In addition, there might be malevolent adversaries who jam or influence the channel of the legitimate users. This paper surveys different models for secure communication under channel uncertainty and adversarial attacks and reviews the corresponding secrecy capacity results, which characterize the maximum rate at which information can be sent to legitimate receivers while being kept perfectly security from eavesdroppers.