We highly respect reproducible research, so we try to provide the simulation codes for our submitted papers. Please refer to the following paper for more details.
@ARTICLE{11018390,
author={Xiao, Jian and Wang, Ji and Liu, Yuanwei},
journal={IEEE Communications Letters},
title={Channel Estimation for Pinching-Antenna Systems (PASS)},
year={2025},
doi={10.1109/LCOMM.2025.3575166}}
Dataset link: https://pan.baidu.com/s/1iqhJiSGO3rq3VN25GghYAA?pwd=5u1v
You can directly run the “train_CMoE_mixer2.py” file to obtain the channel prediction results. Now, this code is a preliminary version composed of a few redundant statements, we will try my best to release the clean codes and add the necessary annotations in the future. Note that in the training stage, the different hyper-parameters setup will result in slight difference for final channel estimation performance. According to our training experiences and some carried attempts, the hyper-parameters and network architecture can be further optimized to obtain better channel estimation performance gain, e.g., the training learning rate, batchsize and epochs. However, these adjustments will not affect the conclusion of the paper. What we care about is a general framework, not the hyper-parameter adjustment work.
The author in charge of this simulation code package is: Jian Xiao (email: jianx@mails.ccnu.edu.cn). If you have any queries, please don’t hesitate to contact me.