APA_200MHz
收藏资源简介:
OpenDPDv2是一个统一的框架,用于PA建模、DPD学习和模型优化,以减少功耗同时保持高线性化性能。该框架包含一个名为APA_200MHz的测试信号数据集,由3.5 GHz Ampleon GaN功率放大器(PA)在41.5 dBm平均输出功率下测量得到。该数据集包含5个信道,每个信道40 MHz(200 MHz)的测试信号,为社区提供了精确的PA测量数据,可用于建模、DPD学习和其他数字信号处理算法和系统的潜在开发。
OpenDPDv2 is a unified framework for PA modeling, DPD learning, and model optimization, designed to reduce power consumption while preserving high linearization performance. This framework includes a test signal dataset named APA_200MHz, which was measured using a 3.5 GHz Ampleon GaN power amplifier (PA) at an average output power of 41.5 dBm. The dataset contains 5 channels, each with a 40 MHz bandwidth (totaling 200 MHz) of test signals, providing the research community with precise PA measurement data that can be utilized for modeling, DPD learning, and the potential development of other digital signal processing algorithms and systems.
OpenDPD 数据集概述
数据集基本信息
- 名称: OpenDPD
- 开发团队: 代尔夫特理工大学高效机器智能实验室 (Lab of Efficient Machine Intelligence)
- 框架: PyTorch
- 用途: 功率放大器(PA)建模和数字预失真(DPD)的端到端学习框架
版本信息
OpenDPDv1
- 相关论文:
- 数据集:
- DPA_100MHz
- DPA_160MHz
- DPA_200MHz
- 支持模型:
- GRU, LSTM, GMP, RVTDCNN, VDLSTM, DGRU, TCN
- 性能指标:
- DGRU (1041参数): ACPR -58.4 dBc, EVM -39.1 dB
OpenDPDv2
- 相关论文: arXiv
- 数据集:
- APA_200MHz
- APA_200MHz_b
- 支持模型:
- PGJANET, DVRJANET, TRes-DeltaGRU
- 性能指标:
- TRes-DeltaGRU (996参数): ACPR -59.4 dBc, EVM -42.1 dB
数据集结构
. └── backbone # 特征提取器配置文件 └── datasets # 测量PA数据集 └──DPA_200MHz # 200 MHz OFDM信号的数字功率放大器 └── dpd_out # 自动生成的输出文件 └── log # 实验日志数据 └── modules # 主要功能模块 └── save # 保存的模型 └── steps # 实现步骤 └── utils # 工具库
训练流程
-
数据获取与预处理:
- 收集并预处理PA的基带I/Q信号
- 数据按8:2:2比例划分为训练、测试和验证集
-
PA建模: bash python main.py --dataset_name DPA_200MHz --step train_pa --accelerator cpu
-
DPD学习: bash python main.py --dataset_name DPA_200MHz --step train_dpd --accelerator cpu
-
量化感知学习: bash python main.py --dataset_name DPA_200MHz --step train_dpd --accelerator cpu --DPD_backbone qgru --quant --n_bits_w 16 --n_bits_a 16
-
验证实验: bash python main.py --dataset_name DPA_200MHz --step run_dpd --accelerator cpu
引用
bibtex @INPROCEEDINGS{Wu2024ISCAS, author={Wu, Yizhuo and Singh, Gagan Deep and Beikmirza, Mohammadreza and de Vreede, Leo C. N. and Alavi, Morteza and Gao, Chang}, booktitle={2024 IEEE International Symposium on Circuits and Systems (ISCAS)}, title={OpenDPD: An Open-Source End-to-End Learning & Benchmarking Framework for Wideband Power Amplifier Modeling and Digital Pre-Distortion}, year={2024}, volume={}, number={}, pages={1-5}, keywords={Codes;Transmitters;OFDM;Power amplifiers;Artificial neural networks;Documentation;Benchmark testing;digital pre-distortion;behavioral modeling;deep neural network;power amplifier;digital transmitter}, doi={10.1109/ISCAS58744.2024.10558162}}
贡献者
- Chang Gao - 项目负责人
- Yizhuo Wu - 核心开发者
- Ang Li - 核心开发者
- Huanqiang Duan - 贡献者
- Ruishen Yang - 贡献者




