Beam-Squint Aware Sparse Techniques for Massive MIMO-OFDM Integrated Sensing and Communication

Novel techniques are conceived for direction-of-arrival (DoA), range, velocity, and reflection coefficient estimation of multiple moving targets in a beam-squint aware massive multiple-input multiple-output (mMIMO) orthogonal frequency division multiplexing (OFDM) system designed for integrated sensing and communication (ISAC) relying on a hybrid beamforming architecture. A grid-aligned sparse Bayesian learning (GA-BL) solution is proposed for determining the DoA, delay, and Doppler shift parameters of these targets. Further improvements are made by formulating a multi-measurement vector (MMV) based sparse problem for estimating the target parameters. Then, by leveraging this model, we dispense with the grid-aligned assumption, and its gridless Doppler-based MMV BL (GLD-MBL) counterpart is introduced. The latter offers three key advantages in comparison to GA-BL: superior Doppler estimation performance, the ability to estimate gridless Doppler shift, and significantly reduced computational complexity. To further enhance the estimation performance, a block-sparse problem is formulated for the estimation of the target parameters under the assumption of gridless Doppler. The gridless Doppler block sparse BL (GLD-BLKBL) algorithm is then harnessed for exploiting both the block sparsity and intrablock correlations, thus achieving notable estimation performance improvements over the GA-BL and GLD-MBL approaches in the high signal-to-noise ratio (SNR) regime. Finally, simulation results are provided to illustrate the improved performance of the proposed methodologies across multiple performance metrics. These results highlight the superiority of our solutions over conventional sparse recovery techniques.