Fig. 4: S-DNN for integrated sensing and communication. | Light: Science & Applications

Fig. 4: S-DNN for integrated sensing and communication.

From: Super-resolution diffractive neural network for all-optical direction of arrival estimation beyond diffraction limits

Fig. 4

a The reflective LC RIS is controlled by 400-channel voltages through FPGA. The voltage applied to the liquid crystal layer changes its dielectric constant for modulating the phase of incident EM fields. b Schematic illustrating the application of DOA estimation with passive and reconfigurable S-DNNs for RIS-based communications. c Experimental output energy distribution of the three-layer passive S-DNN for angular estimation of users and base stations (top). Based on the results of the S-DNN, RIS accurately steers the base station beam to the user, improving the receiving gain (bottom). d RIS-based communication systems using the angular estimation result of reconfigurable S-DNNs substantially improve the detected signal strength. e, S-DNN advanced over the conventional DOA estimation with MUSIC in terms of the snapshots and angular resolution. The S-DNN is more robust and achieves higher angular resolution than MUSIC at low input SNR

Back to article page