Industrial Scene Area Positioning Technology Based on LED Endogenous Identity Characteristics

Due to the unique advantages of visible light positioning (VLP), its application in the industrial scene has received extensive attention. However, the implementation of VLP usually requires complex equipment and huge amount of computation, which is a huge challenge for the industrial scene with many production factors. Therefore, this paper proposes an industrial scene area positioning technology based on LED endogenous identification characteristics, which uses the same set of visible light equipment to realize communication and positioning, effectively avoiding the problems faced by current VLP applications in industrial production. The principle of this technology is to identify the LED of the transmitted signal according to the signal waveform received by the mobile terminal, and then use the corresponding position information of the LED to achieve positioning. To improve the accuracy of LED recognition, we propose a combined network of wavelet scattering and long short-term memory (WLS-LSTM) to extract signal waveform features and classify. Otherwise, we verify the classification accuracy of the WLS-LSTM through experiments. The experimental results show that the trained WLS-LSTM network model has a classification accuracy of 98.5% for six different models of LEDs, and the lowest classification accuracy of 90.2% in several different areas under each model of LED, showing good generalization. Finally, through comparison, it is found that the classification accuracy of WLS-LSTM is significantly higher than that of other common classification algorithms.