MobRFFI: A WiFi RF Fingerprinting Dataset with Granular Multi-Receiver Signal Capture
收藏资源简介：
MobRFFI is a WiFi device fingerprinting and re-identification dataset collected in the Orbit testbed facility in July and April 2024. The dataset contains raw IQ samples of WiFi transmissions captured at 25 Msps on channel 11 (2462 MHz) in the 2.4 GHz band, using Ettus Research N210r4 USRPs as receivers and a set of WiFi nodes equipped with Atheros AR5212 chipsets as transmitters. The data collection spans two days (July 19 and August 8, 2024) and includes 12,068 capture files totaling 5.7 TB of data. Each capture file contains a two-second signal capture, during which we performed a transfer of randomly generated data via UDP protocol between a transmitter and a WiFi access point, with the USRP receiver performing independent signal capture. The dataset has several key advantages that may be useful for developing novel WiFi-based RFFI methods. First, we perform signal capture simultaneously across multiple USRP receivers (4 on day 1 and 3 on day 2). Second, we perform repeated rounds of signal capture, which is useful for evaluating method performance on multiple hours (24 hours on day 1, and 4 hours on day 2). Third, we perform signal capture on two separate days, with sufficient sensor overlap for evaluating multi-day method performance. Finally, we provide a suite of signal processing tools and a reduced-size dataset for faster onboarding and experimentation. For more details, please refer to our GitHub repository: https://github.com/i-sense/mobrffi-paper If you find this dataset useful, please consider citing our recent publication: "MobRFFI: WiFi Device Fingerprinting and Re-identification for Mobility Intelligence."
MobRFFI是一款WiFi设备指纹与重识别数据集，于2024年7月及4月在Orbit试验床设施中采集。该数据集包含2.4GHz频段11号信道（2462MHz）上以每秒25百万次采样（Msps）速率捕获的WiFi传输原始IQ样本，采集设备采用Ettus Research N210r4通用软件无线电外设（USRP）作为接收机，搭载Atheros AR5212芯片组的一组WiFi节点作为发射机。本次数据采集周期跨越两日（2024年7月19日与8月8日），共包含12068个捕获文件，总数据量达5.7 TB。每个捕获文件对应一段时长2秒的信号捕获过程，期间我们通过用户数据报协议（UDP）在发射机与WiFi接入点之间传输随机生成的数据，同时由USRP接收机独立完成信号捕获。 本数据集具备多项可用于开发新型基于WiFi的射频指纹识别（RFFI）方法的关键优势。其一，我们可通过多台USRP接收机同步开展信号捕获（首日使用4台，次日使用3台）；其二，我们开展了多轮重复信号捕获，可用于评估方法在数小时级场景下的性能（首日捕获时长24小时，次日为4小时）；其三，我们在两个独立日期完成信号捕获，且传感器部署存在足够重叠，可用于评估跨日期方法的性能；最后，我们配套提供了一套信号处理工具集与精简版数据集，以加速模型入门与实验开展。 如需了解更多细节，请参阅我们的GitHub仓库：https://github.com/i-sense/mobrffi-paper 若您认为本数据集对研究有所助益，请引用我们的最新学术论文：《MobRFFI：面向移动智能的WiFi设备指纹与重识别技术》。
