Convolutional neural network-based simultaneous display-communication system

Due to the overlapping emission and detection spectra of quantum well (QW) diodes, they inherently possess the dual functions of light emission and detection. In this paper, we integrate a 4 x 4 array of QW diodes and combine it with a programmable circuit and a convolutional neural network algorithm, ultimately proposing a simultaneous display-communication system. This system not only displays visual content but also receives external signals via wireless light communication and classifies and recognizes the signal content with an accuracy exceeding 95%. The QW diode array operates within a temperature range of -40-85 degrees C and is easily scalable, making it suitable for both on-chip and off-chip integration. Moreover, the channels are mutually independent, meaning the channel capacity is theoretically proportional to the number of QW diodes. This system has significant potential for secure transmission and intelligent display applications: while the screen displays a certain image, it may also be secretly transmitting other information in the background.