Feasibility of Passive Eavesdropping in Massive MIMO: An Experimental Approach
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Massive MIMO systems have the potential for preventing passive eavesdropping as the signal transmitted by a large antenna array becomes highly focused. Prior works showed that passive eavesdropping becomes negligible when the number of BS antennas approaches to infinity in independent Rayleigh channel from a secrecy rate perspective. However, in practical massive MIMO systems, the number of BS antennas is in the order of a hundred, not infinity. Also, channels in the real world are not ideally independent. Furthermore, secrecy rate does not directly indicate whether a transmission can be decoded by the eavesdropper in practical wireless transmission.
In this work, our analysis is based on real channel measurements from a 96-antenna ArgosV2 BS in 2.4 GHz band indoor environment with a LOS component. Instead of the asymptotic behavior, we focus on how the increasing number of BS antennas affect passive eavesdropping. Also, we propose to use the SNR difference between the intended user Bob and the eavesdropper Eve as a metric to determine how resistant to passive eavesdropping a system is.
From our analysis based on real channel measurements, we find that increasing the number of antennas at the BS improves the ability of preventing passive eavesdropping, and a 96-antenna BS has the potential to prevent passive eavesdropping in the indoor LOS environment with careful power control. However, compared to the independent Rayleigh channel, indoor LOS environment is less passive-eavesdropping resistant under the same number of BS antennas. Furthermore, the marginal benefit of increasing an antenna in the indoor LOS environment decreases much faster than in the independent Rayleigh channel scenario.
Description
Advisor
Degree
Type
Keywords
Citation
Yeh, Chia-Yi. "Feasibility of Passive Eavesdropping in Massive MIMO: An Experimental Approach." (2018) Master’s Thesis, Rice University. https://hdl.handle.net/1911/105588.