PAPR-Limited Precoding in Massive MIMO Systems with Reflect- and Transmit-Array Antennas (1912.00485v1)

Abstract: Conventional hybrid analog-digital architectures for millimeter-wave massive multiple-input multiple-output (MIMO) systems suffer from poor scalability and high implementational costs. The former is caused by the high power loss in the analog network, and the latter is due to the fact that classic MIMO transmission techniques require power amplifiers with high back-offs. This paper proposes a novel hybrid analog-digital architecture which addresses both of these challenges. This architecture implements the analog front-end via a passive reflect- or transmit-array to resolve the scalability issue. To keep the system cost-efficient, a digital precoder is designed whose peak-to-average power ratio (PAPR) on each active antenna is tunable. Using the approximate message passing algorithm, this precoder is implemented with tractable computational complexity. The proposed architecture allows for the use of power amplifiers with low back-offs which reduces the overall radio frequency cost of the system. Numerical results demonstrate that for low PAPRs, significant performance enhancements are achieved compared to the state of the art.