Research On Indoor Localization Technology Based On Multi-source Wireless Fusion Sensing

Indoor localization has wide applications in the Internet of Things(IoT)fields,such as warehouse management.Many wireless technologies including Bluetooth,WiFi and Radio Frequency Identification(RFID)have been used in indoor localization.Among them,RFID has advantages of low cost,easy deployment and fast identification,etc.So it has a natural advantage in indoor localization.Besides,RFID can also support functions such as object posture sensing and liquid composition recognition.Therefore,RFID-based sensing technology has important research value in the field of IoT.Due to the periodicity of RFID phase signals,the existing indoor localization works based on RFID usually need to deploy lots of antennas and readers for localization.These systems are designed for large-scale stocktaking in warehouse management.These systems are not cost-effective when dealing with some applications,such as finding a specific object in a small range.In this paper,considering the limitations of the existing RFID indoor localization works,we have studied the indoor localization based on the moving antenna by combining the computer vision technology as an auxiliary tool.The main contributions are as follows:(1)By deploying a camera on the RFID antenna,computer vision technology is used to recover the moving trajectory of the antenna,which provides more flexibility for the antenna movement and supports the hand-held antenna for signal acquisition.(2)To locate the target object with only one antenna,we use phase point pairs to generate probability holograms,which can eliminate the interference caused by the periodicity of the phase and the deviations induced by hardware devices.In addition,the influence of outliers can be eliminated by setting thresholds.(3)We implemented a prototype system using commercial devices,and evaluated the performance of the proposed algorithm in the real environment.Experimental re-sults show that the system can keep the average localization error within 20 cm in the confidence interval of 95%.