接头(建筑物)
计算机科学
电子工程
电气工程
工程类
滤波理论
数字滤波器
控制理论(社会学)
信号处理
声学
带宽(计算)
物理
矩阵代数
作者
Uichan Park,Donghun Kang,Jungsuek Oh
标识
DOI:10.1109/tmtt.2026.3734488
摘要
This article proposes a novel system-level framework that enables, rather than replaces, standard digital predistortion (DPD) to operate effectively in dual-polarized (DP) arrays. Conventional DPD models typically fail in this environment due to two fundamental physical-layer challenges: 1) severe power amplifier (PA) nonlinear variations caused by dynamic load-impedance (LIP) modulation and 2) DPD training corruption from cross-polarization interference (XPI) in the feedback path. Our solution overcomes these challenges by employing two complementary novel techniques that precondition the system before DPD adaptation: the reference phase (RP) control for LIP alignment and XPI cancellation. The RP adjustment technique actively tunes the relative phase between vertical and horizontal polarization (V, HP) paths to align the LIP seen by each PA without altering the far-field beam direction. This alignment minimizes per-path impedance variations and makes the PAs’ nonlinear responses statistically more similar, allowing a single DPD to linearize the entire array consistently. In parallel, the proposed XPI cancellation scheme treats the DP transmitter and observation receiver (OR) as a$2\times 2$MIMO system, estimating and removing the cross-coupling between polarizations in the feedback path. By eliminating leaked distortion from the DPD training signals, the predistorter adapts only to the PAs’ intrinsic nonlinearity. Using a measurement-validated simulation testbed, we demonstrate that the combined approach significantly improves linearity and broadens the effective linearity beamwidth of the array. In experiments, the proposed method achieved an order of magnitude reduction in error vector magnitude (EVM) and doubled the angular range of satisfactory EVM compared to a conventional DPD.
科研通智能强力驱动
Strongly Powered by AbleSci AI