Research on RFID Indoor Positioning Algorithm Based on Attention
Jintao Tang,Yishu Qiu,Lvqing Yang,Yihui Deng,Jiangsheng Zhao,Shaoqin Shen
DOI: https://doi.org/10.1109/ICETCI53161.2021.9563444
2021-08-27
Abstract:With the improvement of the quality of life, the amount of time people spend indoors is constantly increasing, and the demand for indoor positioning is also increasing. In recent years, many wireless indoor positioning technologies have emerged, among which RFID has been favored by researchers due to its stability, low power consumption and price advantages. In order to reduce the great influence of RFID signal fluctuations on the positioning results, this paper proposes an indoor positioning technology based on Attention. The mobile reader collects the Received Signal Strength Indication, Time of Arrival and Angle information of the positioning tag as timing features, and the positioning coordinates as the target to train regression model; The model proposed in this paper is characterized by using the Attention mechanism to work as a filter, followed by a fully connected layer for features combination, to predict coordinates. The experimental results show that the naive Attention positioning model can predict the highest proportion of the error range of 0.2m, and the Multi-head Attention positioning model performs best in the overall positioning results.
Engineering,Computer Science