Indoor Positioning and Posture Recognition of Human Body Applying Integrating‐Type Intelligent Metasurfaces Based Sensing System

Abstract The electromagnetic (EM) detection and sensing research of human signs are becoming much more attractive, especially with the rise of intelligent metasurface which can strongly enhance the control abilities and functionalities of transmitting and receiving EM fields. It proposes the design of integrating‐type intelligent metasurface to realize the indoor positioning and posture recognition of the human body using the support vector machine algorithm and a new deep learning algorithm model it proposed called MetaNet. The integrating‐type metasurface with the in‐plane feed rather than horn excitation is real low profile, and the varactor based programmable phase shifters achieving 360‐degree cover and continuous control of phase are independently embedded in each unit cell for further synthesized manipulation of apertures. One pair of integrating‐type intelligent metasurfaces are fabricated for transmitting and receiving EM fields, respectively. Also, the corresponding prototype of radar transceiver system is developed and manufactured to rapidly acquire massive data. The total experimental flow can include two main steps that first, the transmitting and receiving beams of metasurfaces should scan dynamically and synchronously under the method of far‐field synthesis for detecting the direction of the human. Second, the transmitting beam should keep the detected direction and the receiving apertures of metasurface should be varied by randomly control the phase shift of each unit cell for multiple sensing. Based on the technology of deep learning to process the massive receiving data, the profile images and key points of various human postures can be reconstructed very well in the measurement that can verify the validity of the proposed sensing system.