Nowadays, thanks to the development of China's network technology and information technology, indoor positioning technology has also developed rapidly. As we all know, GPS positioning has unique advantages in outdoor positioning, and has good anti-interference and confidentiality, but it can not play a role in the field of indoor positioning, and radio frequency identification (RFID) technology can make up for this shortcoming, it has Low cost, non-line-of-sight, non-contact, and good positioning performance, so it has a wide range of applications in the field of indoor positioning. Although the research of indoor positioning system based on RFID has become more and more mature, due to the complexity of the indoor environment, it is easy to receive other interference between the signals, and it becomes more and more difficult to improve the accuracy of the positioning algorithm. This paper designs an indoor positioning system based on RFID. First, it briefly describes RFID, and introduces the background significance and development history of RFID. The advantages and disadvantages of indoor positioning technology and positioning algorithms, including common positioning algorithms and classic RFID-based systems (LANDMARC system and VIRE system) are summarized. Next, the overall framework of the indoor positioning system based on RFID is described, including the overall system design, system composition, functional module design, and the design of the database, client, and middleware. This paper proposes a method of adding virtual beacons based on the weighted centroid algorithm, which can reduce the hardware cost while improving the positioning accuracy. Tests by deploying a positioning system in an indoor environment prove that the positioning system can dynamically display patient positioning information and real-time voice alarm function in real time, especially the positioning accuracy has been greatly improved. The experimental data shows that the weighted centroid algorithm based on virtual nodes not only greatly improves the positioning accuracy, but also reduces the implementation cost of hardware. The sampling time of the system can also be controlled within 1s, achieving the desired effect.