On eavesdropping attacks and countermeasures for MU-MIMO systems.

MU-IMMO beamforming offers great potential for the AP with multiple antennas to serve multiple receivers concurrently. A key factor to implement the MU-MIMO system is the CSI feedback mechanism. However., it might be exploited by malicious attackers to threaten the transmission security of benign clients. To better explore the attacks through false CSI feedback in MU-MIMO systems, this paper proposes a mathematically formulated strategy of eavesdropping attack, called polynomial attack. Based on the basic polynomial attack strategy, we investigate two kinds of advanced polynomial attacks: the passive polynomial attack and the active polynomial attack. They are discriminated by the different abilities in approaching to the eavesdropped victims. Besides proving the attack effects theoretically, we also develop defensive methods against the two attacks. Especially against the active attack, we propose a novel MU-MIMO scheme called AntiPoly, under which each client holds a pair of constant keys. One is the feedback key used for interacting with the AP in the CSI training period, and the other one is the transmission key used for decoding signals in the transmission period. With intensive experiments performed on actual measured CSI, the attack effects of polynomial attacks are illustrated, and our developed defensive methods are validated.