Strong Secrecy for Multiple Access Channels

TL;DR

This paper establishes strongly secret achievable rate regions for two wiretap multiple-access channel scenarios, highlighting the roles of common randomness and conferencing in enabling secure communication.

Contribution

It introduces new achievable rate regions for secret communication in multiple-access channels with limited or no common randomness, and demonstrates how conferencing can enable secrecy.

Findings

Achievable regions depend on access to common randomness.

Conferencing can enable secret transmission where it was previously impossible.

Without common randomness or conferencing, secret transmission is not achievable.

Abstract

We show strongly secret achievable rate regions for two different wiretap multiple-access channel coding problems. In the first problem, each encoder has a private message and both together have a common message to transmit. The encoders have entropy-limited access to common randomness. If no common randomness is available, then the achievable region derived here does not allow for the secret transmission of a common message. The second coding problem assumes that the encoders do not have a common message nor access to common randomness. However, they may have a conferencing link over which they may iteratively exchange rate-limited information. This can be used to form a common message and common randomness to reduce the second coding problem to the first one. We give the example of a channel where the achievable region equals zero without conferencing or common randomness and where conferencing establishes the possibility of secret message transmission. Both coding problems describe practically relevant networks which need to be secured against eavesdropping attacks.