I2SAC: AI-enabled Digital-Twin Framework on a mmWave 64-element Phased Array for Integrated Imaging, Sensing, and Communication

This article presents an AI-enabled unified digital-twin framework for Integrated Imaging, Sensing, and Communication (I2SAC) using a multi-static aperture comprising a 64-element dual-polarized electronically scanned phased array and four open-ended waveguide RF probes operating at 25–30 GHz. The physical layer is emulated using a computationally efficient, physics-informed dipole-moment propagation model to generate large synthetic datasets over spatially distributed targets and vector fields. These data train a deep neural network to learn the inverse channel matrix across space, frequency, and polarization. The digital twin is refined using sparse hardware-in-the-loop calibration for physical accuracy. Using the calibrated model, we experimentally demonstrate: (1) robust sensing of mmWave channels under hardware non-idealities, (2) computational imaging with 1.5 cm cross-range resolution at 1 m stand-off, and (3) 64-QAM communication over 1 GHz bandwidth with 3.5% average EVM.