A Unified Successive Channel Model and Switch-Free Channel Estimation for MMWave MIMO Systems

Abstract

Large-aperture millimetre-wave (mmWave) massive multiple-input multiple-output (MIMO) systems can span near-, transition-, and far-field propagation regimes. Conventional channel estimators typically select a field-specific representation, such as a far-field angular-sparse channel or a near-field polar-sparse channel. This model switching becomes fragile in the transition region, where weak wavefront curvature can be missed or incorrectly absorbed into a planar-wave atom. This paper first proposes a unified successive channel model for mmWave MIMO systems, in which the channel phase is represented by a linear phase trend and a curvature-related term so that the far-field case is naturally recovered as the zero-curvature limit. Based on this model, we develop a switch-free hierarchical OMP estimator that represents the channel using full-matrix atoms parameterised by phase slope and curvature, with low-curvature-aware candidate screening and local refinement to enhance robustness in the transition region. Simulation results under a 28 GHz wideband Rician multipath mmWave MIMO setting show that the proposed unified OMP estimator consistently achieves lower NMSE than the far-field OMP, near-field OMP, hybrid-field OMP, LS, and MMSE baselines across the evaluated near-, transition-, and farfield distances.