Secrecy Analyses of a Full-Duplex MIMOME Network

TL;DR

This paper analyzes the secrecy performance of a full-duplex MIMOME network with unknown eavesdropper CSI, proposing power optimization and channel estimation strategies to enhance security against eavesdroppers with multiple antennas.

Contribution

It introduces power optimization techniques and two novel anti-eavesdropping channel estimation methods for full-duplex MIMOME networks with unknown eavesdropper CSI.

Findings

Power optimization significantly increases the maximum number of antennas Eve can have without compromising security.

The analysis provides bounds on secure degrees of freedom based on Eve's antenna count and packet size.

Blind detection methods offer approximate secrecy rates when Eve has no prior CSI knowledge.

Abstract

This paper presents secrecy analyses of a full-duplex MIMOME network which consists of two full-duplex multi-antenna users (Alice and Bob) and an arbitrarily located multi-antenna eavesdropper (Eve). The paper assumes that Eve's channel state information (CSI) is completely unknown to Alice and Bob except for a small radius of secured zone. The first part of this paper aims to optimize the powers of jamming noises from both users. To handle Eve's CSI being unknown to users, the focus is placed on Eve at the most harmful location, and the large matrix theory is applied to yield a hardened secrecy rate to work on. The performance gain of the power optimization in terms of maximum tolerable number of antennas on Eve is shown to be significant. The second part of this paper shows two analyses of anti-eavesdropping channel estimation (ANECE) that can better handle Eve with any number of antennas. One analysis assumes that Eve has a prior statistical knowledge of its CSI, which yields lower and upper bounds on secure degrees of freedom of the system as functions of the number (N) of antennas on Eve and the size (K) of information packet. The second analysis assumes that Eve does not have any prior knowledge of its CSI but performs blind detection of information, which yields an approximate secrecy rate for the case of K being larger than N.