Download Report IEEE DataPort : Multi-Static OFDM Radar Dataset for Human Activity Analysis: Three USRP X410 based Measurements with Motion Capture Ground Truth - 2025
BIN, TXT, JSON by Bixing Yan, Kwadwo Mensah Obeng Afrane, Jiteng Ma, Zhiyu Wang, Achiel Colpaert, Sofie Pollin, Mark Beach, Yang Miao
Information
Format: BIN, TXT, JSON Publisher: IEEE DataPort Publication Date of the Electronic Edition: 12/05/2025
?
ISBN: 10.21227/zz73-wj77
$15
$3Discount Coupon
Delivery time: Maximum 24 hours
Description
This dataset presents comprehensive multi-static OFDM radar measurements for human activity analysis, collected using three USRP X410 software-defined radio platforms. The system operates in the 6.7-6.9 GHz frequency band with custom patch antennas, employing different frequency configurations (6.74 GHz, 6.8 GHz, and 6.86 GHz) to enable both monostatic and bistatic radar sensing capabilities. Each USRP is configured with one transmit and three receive channels, achieving a 20 MHz sample rate per channel. The frequency configuration enables each USRP to transmit on a unique frequency (USRP0: 6.74 GHz, USRP1: 6.8 GHz, USRP2: 6.86 GHz) while all three receive channels on each USRP simultaneously monitor all three transmission frequencies (6.74 GHz, 6.8 GHz, and 6.86 GHz), establishing a comprehensive multi-static radar network with both co-located and distributed sensing capabilities.The dataset includes raw transmitted and received OFDM signals, synchronization timestamps, and corresponding motion capture (MOCAP) ground truth data for human activity scenarios. The transmit OFDM signal design features 2048 subcarriers with the first and last 64 subcarriers designated as guard bands to prevent interference with adjacent frequency bands, and the cyclic prefix length is 512 samples. Two synchronization words with distinct peaks are incorporated ahead of every 100 OFDM symbols to facilitate precise time alignment between transmitted and received signals. The transmit signal contains 500 data symbols, and during measurements, this signal is transmitted repeatedly to enable continuous radar sensing. Raw data are stored in complex float 32 format. For ground truth, the subject's activity was recorded by a motion capture system using eight markers, providing 3D position data with millimeter precision in a controlled indoor environment.This dataset supports both traditional radar signal processing methods and machine learning-based approaches for human motion analysis, including activity recognition, range estimation, and velocity estimation. The synchronized motion capture system provides precise ground truth for algorithm validation and performance benchmarking. The multi-static configuration and diverse measurement scenarios make this dataset valuable for advancing radar-based human sensing research, cell-free radar system development, and AI-driven radar signal processing applications.
$15
$3Discount Coupon
Delivery time: Maximum 24 hours