5G CFR/CSI dataset for wireless channel parameter estimation, array calibration, and indoor positioning

Channel frequency response (CFR) dataset used for "In-Situ Calibration of Antenna Arrays for Positioning With 5G Networks" paper (IEEE Transactions on Microwave Theory and Techniques, in print, doi: 10.1109/TMTT.2023.3256532, preprint link: https://arxiv.org/abs/2303.04470). Since the CFR of the wireless channel, which is also known as the channel state information (CSI) in many cellular or Wi-Fi positioning literature, contains the angle and delay information of both the direct path and reflected paths, we believe the following research fields can benefit from this dataset:1. direction-of-arrival (DOA) estimation and array calibration,2. time-of-arrival (TOA) estimation,3. joint parameter estimation for indoor positioning and navigation,4. line-of-sight (LOS) and non-line-of-sight (NLOS) identifications,5. wireless channel parameters estimation. We use a baseband unit (BBU), a four-channel picocell sub-6GHz remote radio unit (RRU), and a user terminal (UT) to collect the wireless CFRs, all of them are commercial off-the-shelf products with 5G R16 specification supported. The 5G UT with a single omnidirectional cylindrical antenna is mounted on an autonomous vehicle, which communicates to the gNB and generates the ground-truth locations with an accuracy of several centimeters. This CFR dataset is collected in an underground parking lot of an office building. During experiments, the UT is placed at 476 different locations in the area of [−35 m, 10 m] × [8 m, 33 m]. At each location, the uplink wireless channel is measured repetitively via transmitting a series of uplink sounding reference signal (UL-SRS) symbols and their corresponding CFR data are logged. Aside from CFR data from these 476 locations, other assisted measurement data, including the ground-truth locations of the UT, the six degrees of freedom (6-DoF) pose of the RRU, the locations of the dominant obstacles (i.e. the pillars) in the environment, are also provided.

该频段信道频率响应（CFR）数据集用于《基于5G网络的阵列天线现场校准定位》论文（IEEE微波理论与技术汇刊，待刊发，DOI：10.1109/TMTT.2023.3256532，预印本链接：https://arxiv.org/abs/2303.04470）。鉴于无线信道频段响应（亦称信道状态信息，CSI）在众多蜂窝或Wi-Fi定位文献中，包含直达路径和反射路径的角度及延迟信息，我们认为以下研究领域可从本数据集中获益：1. 方位角（DOA）估计与阵列校准，2. 到达时间（TOA）估计，3. 室内定位与导航的联合参数估计，4. 视距（LOS）与非视距（NLOS）识别，5. 无线信道参数估计。我们使用基带单元（BBU）、四通道亚6GHz微蜂窝远程射频单元（RRU）以及用户终端（UT）来收集无线CFR，所有设备均为符合5G R16规范的商用现货产品。配备单全向圆柱形天线的5G UT被安装在自动驾驶车辆上，与gNB通信并生成数厘米精度的地面真实位置。该CFR数据集是在一座办公大楼的地下停车场收集的。在实验过程中，UT被放置在[−35 m, 10 m] × [8 m, 33 m]区域内的476个不同位置。在每个位置，通过发送一系列上行链路声学参考信号（UL-SRS）符号及其对应的CFR数据，反复测量上行链路无线信道。除了476个位置的CFR数据外，还包括其他辅助测量数据，如UT的地面真实位置、RRU的六自由度（6-DoF）姿态、环境中主要障碍物（即柱子）的位置。

搜集汇总

背景与挑战

背景概述

该数据集是一个5G CFR/CSI数据集，专为无线信道参数估计、阵列校准和室内定位设计。数据采集于真实室内环境（办公楼地下停车场），包含476个不同位置的上行链路信道频率响应样本，每个位置有100次重复测量，并提供地面真实位置、远程无线电单元姿态等辅助数据。数据集适用于方向到达估计、时间到达估计、联合参数估计等多个研究领域，支持MAT和Excel格式。