一种基于深度学习的自适应数字预失真器设计方法及系统
By combining a deep learning model with LSTM and FNN modules on an FPGA platform, online adaptive adjustment of the digital predistorter was achieved, solving the problem of fixed model parameters in existing technologies and improving the compensation effect and system robustness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHANGGUANG SATELLITE TECH CO LTD
- Filing Date
- 2025-10-11
- Publication Date
- 2026-07-17
AI Technical Summary
Existing deep learning-based digital predistorters lack online adaptive capabilities in practical applications, failing to adapt to environmental changes, resulting in a significant reduction in compensation effectiveness and failing to meet practical linearization requirements.
An adaptive digital predistorter is designed, including an offline training module, a forward inference module, a loss calculation module, and a backpropagation module. Real-time loss calculation and backpropagation are implemented on an FPGA platform, and online adaptive adjustment is performed using a deep learning model combining LSTM and FNN modules.
It enables continuous optimization of model parameters in dynamic environments, improves the system's robustness to temperature drift, frequency offset and nonlinear drift, and significantly enhances compensation accuracy and engineering feasibility.
Smart Images
Figure CN121125405B_ABST