Secure Communications over Wireless Networks Even 1-bit Feedback Helps Achieving Security
Zouheir Rezki · King Abdullah University of Science and Technology Repository (King Abdullah University of Science and Technology) · 2016
Recently, there have been a surge toward developing sophisticated security mechanisms based on a cross layer design. While an extensive progress has been realized toward establishing physical layer security as an important design paradigm to enhance security of existing wireless networks, only a little effort has been made toward designing practical coding schemes that achieve or approach the secrecy capacity. Most of existing results are tied to some simplifying assumptions that do not seem always reasonable (passive eavesdropper, perfect channel state information (CSI), etc.). Furthermore, it is still not very clear how to exploit physical layer security paradigms, together with existing cryptosystems, in order to add a supplementary level of protection for information transmission or to achieve key agreement. In this talk, we address the first part of the above problematic, i.e., the effect of channel uncertainty on network security. Particularly, we show that even a coarse estimate of the main channel (channel between the transmitter and the legitimate receiver) can help providing a positive secrecy rate. Specifically, we assume two types of channel uncertainty at the transmitter. The first one is a rate-limited feedback in a block fading channel where the feedback information can be proactive (at the beginning of the coherence block) or of ARQ-type. The second type of uncertainty takes the form of a noisy estimate of the main channel at the transmitter in a fast fading channel. In both cases, we provide upper and lower bounds on the secrecy capacity. We argue how our achievable schemes and upper bounding techniques extend to multi-user setting (broadcasting a single confidential message or multiple confidential messages to multiple legitimate receivers) and to multiple antenna channels.