Channel Estimation and Signal Detection for Massive MIMO-ODDM With Hybrid Beamforming
Abstract
Massive multiple-input multiple-output (MIMO) systems combined with orthogonal delay-Doppler division multiplexing (ODDM) waveform constitute a robust paradigm for next-generation high-mobility communications. However, the high-dimensional coupling of time delays, Doppler shifts, angles of departure (AoD), and angles of arrival (AoA) poses a significant challenge to reliable signal detection and accurate channel estimation, especially under the hardware constraints of hybrid beamforming architectures. To address these challenges, we first derive an input-output (IO) relationship for the massive MIMO-ODDM system with hybrid beamforming. Leveraging this formulation, we develop a singular value decomposition (SVD)-based hybrid beamforming strategy, along with a low-complexity iterative block successive interference cancellation (SIC) linear minimum mean square error (LMMSE) detector to ensure robust transmission. Furthermore, to acquire accurate channel state information (CSI), we reformulate the estimation task as a partially-decoupled four-dimensional (4D) sparse signal recovery (SSR) problem, which enables separate parameter estimation to circumvent the complexity of full joint estimation. Building on this formulation, a 4D hierarchical sparse Bayesian learning (HSBL) framework is proposed that adopts a coarse-to-fine strategy to reconcile estimation accuracy with computational complexity. Simulation results demonstrate that the proposed hybrid beamforming and detection scheme ensures robust error performance, while the estimation framework achieves superior accuracy with significantly reduced complexity compared to conventional off-grid SBL methods.