Secrecy Rate of Cooperative MIMO in Presence of a Location Constrained Eavesdropper

TL;DR

This paper introduces a cooperative MIMO system called Reconfigurable Distributed MIMO, which enhances secrecy rates by leveraging cluster cooperation and eavesdropper location constraints, using practical precoding and artificial noise.

Contribution

It presents a novel cooperative MIMO architecture with closed-form secrecy rate approximations and optimization methods considering eavesdropper location constraints.

Findings

Secrecy rate significantly improves with eavesdropper location constraints.

Increasing cluster size further enhances secrecy performance.

Proposed approximations facilitate effective secrecy rate maximization.

Abstract

We propose and study the cooperative MIMO architecture to enable and to improve the secrecy transmissions between clusters of mobile devices in the presence of an eavesdropper with certain location constraint. The cooperative MIMO system in this paper (referred to as Reconfigurable Distributed MIMO) is formed by temporarily activating clusters of nearby trusted mobile devices, with each cluster being centrally coordinated by its corresponding cluster head. We assume that the transmitters apply a practical eigen-direction precoding scheme to transmit the confidential signal and artificial noise, while the eavesdropper can be located in multiple possible locations in the proximity. We first obtain the expression of the secrecy rate, where the required average mutual information between the transmitters and the eavesdropper is characterized by closed-form approximations. The proposed approximations are especially useful in the secrecy rate maximization, where the original non-convex problem can be solved by the successive convex approximations. Numerical results show that the secrecy rate can be significantly improved by leveraging the location constraint of the eavesdropper, compared to the existing result. We also demonstrate that the secrecy rate can be further improved by increasing the cluster size.