Training and Testing Datasets for Detection Network

该数据集是为了训练和测试一种基于1位压缩感知的超叠加信道状态信息反馈的深度学习方案而生成的。这些数据集是在无噪声条件下，使用QPSK调制在随机生成的伯努利二进制序列上生成的。数据集的规模如下：训练集包含60,000个样本，验证集包含20,000个样本，测试集也包含20,000个样本。该任务旨在检测上行用户数据序列和下行信道状态信息。

This dataset was generated for training and testing a deep learning scheme based on 1-bit compressive sensing for superimposed channel state information feedback. It was constructed under noise-free conditions using QPSK modulation over randomly generated Bernoulli binary sequences. The dataset is divided into three subsets with the following specifications: the training set contains 60,000 samples, the validation set includes 20,000 samples, and the test set also has 20,000 samples. The target task of this dataset is to detect uplink user data sequences and downlink channel state information.