RFID localization is considered the key enabler of automating the process of inventory tracking and manage- ment for high-performance logistic operations. A practical and deployable RFID localization system needs to meet re- liability, range and throughput requirements. This paper presents RF-CHORD, the first RFID localization system that simultaneously meets all three requirements. RF-CHORD features a one-shot multisine-constructed wideband design that can process RF signal with a 200 MHz bandwidth in real-time to facilitate one-shot localization at scale. In addition, multiple SINR enhancement techniques are designed for range extension. On top of that, a kernel-layer-based near-field localization framework and a multipath-suppression algorithm are proposed to reduce the 99% long-tail errors. Our empirical results show that RF-CHORD can localize more than 170 tags 6 m away from a reader within 1 second and with 99% long-tail error smaller than 1 m, achieving a 0% miss reading rate and ~0.01% cross-reading rate in the warehouse and fresh food delivery store deployment.

RF-CHORD: Towards Deployable RFID Localization System for Logistics Network

Bo Liang, Purui Wang, Renjie Zhao, Heyu Guo, Pengyu Zhang, Junchen Guo, Shunmin Zhu, Hongqiang Liu, Xinyu Zhang and Chenren Xu

RF-Chord paper

RF-Chord hardware design (Google drive)

RF-Chord evaluation dataset (Google drive)

We are ready to provide more wideband RFID localization dataset and add more meta data. We will gradually open source firmware and algorithms of RF-Chord.

This data set is the evaluation data set of our paper RF-Chord: Towards Deployable RFID Localization System for Logistics Network, which builds a wideband sniffer system for RFID localization in logistics. Our target is to create a deployable RFID localization system featuring high-reliability, high-throughput, and long-range. This work is supported by National Natural Science Foundation of China (Grant No. 62022005, 62272010, and 62061146001) and Alibaba Innovative Research.

The most critical step to high reliability is wideband. Standard UHF RFID works at the ISM band of 902-928 MHz, while our RFID works at a wider band of 800-1000 MHz. Generally speaking, the wider bandwidth, the better the time resolution when locating the target. We build RF-Chord, a system with customized hardware, firmware, and software to realize wideband channel information estimation and localization. The word *Chord* means three or more musical notes played at the same time, and our system uses 16 different tones to build wideband. Refer to our paper for more details about RF-Chord.

To evaluate RF-Chord, we deployed it at different experiment environments and collected data. This data set is from one of the deployment environments, called geek lab in Peking University, Beijing, as shown in the following figure. As the figure shows, there are multiple mental reflectors (e.g., metal furniture, low ceilings, and walls). The evaluation range is the area of 6 x 3.2 m ahead of the antenna. We divide the evaluation space into 20 cm grids and use guide rails to move the tags. All the tags are facing the array. The dataset contains about 20k wideband RFID channel information measurements at 384 locations. We used a guide rail of 60 cm x 60 cm to set the tags at suitable locations. As shown in the following figure, we set a stick on the guide rail and attached five tags to the stick. We can move the stick at 0, 20 cm, 40 cm, and 60 cm at the x or y-axis. Therefore, every time we put the guide rail in one place, we can obtain 4 x 4 = 16 samples. We set the guide rail at 24 locations, extracting the channel information from 16 x 24 = 384 locations. This is also how the file is named. For example, *data_02_04.mat* means the guide rail is at the 2nd location, and the stick is at the 4th location. `tagLocation` in the file records the global location ground truth. All the measurement locations are shown as follows: The green points are the mearsurement locations. The red points are the Rx antennas and the blue points are the Tx antenna.

You can use our example localization script *hologram_localization_script.m* to process the data and see the localization result. It implements basic hologram algorithm in the paper. We also provide the direct path enhancement algorithm in the paper, called *DPE.m*. Or you can develop your own localization algorithm with the wideband channel informaiton of tags.