Variational Bayesian Learning for 3-D Localization of Extended Targets in mmWave MIMO OFDM ISAC Systems

Abstract

Variational Bayesian learning (VBL)-aided extended target localization is conceived for orthogonal frequency division multiplexing (OFDM) based-mmWave MIMO systems using the OFDM integrated sensing and communication (ISAC) waveform. The proposed framework also considers the intercarrier interference (ICI) effects encountered in mobile scenarios and the clutter present in the environment. The proposed algorithm is based on a hybrid mmWave MIMO architecture, where the number of radio frequency (RF) chains is significantly lower than the number of antennas. A range, Doppler and angular (RDA)-domain representation of the target in three-dimensional (3D) space is conceived for accurate target parameter estimation. The proposed algorithm exploits the four-dimensional (4D) sparsity arising in the RDA domain of the scattering scene and employs the powerful VBL framework for the estimation of target parameters, such as elevation angle, azimuth angle, range and velocity. To handle a practical scenario where the actual target parameters typically deviate from their finite-resolution grid, a super-resolution-based improved off-grid VBL is developed for recursively updating the parameter grid, thereby progressively improving the estimates. We also determine the Cramér-Rao bound (CRB) and Bayesian CRB for the estimation of the target parameters in order to bound the estimation performance. Our simulation results validate the superior performance of the proposed approach in comparison to the existing algorithms.