Simulation study demonstrates accurate near-field channel estimation in distributed MIMO systems, indicating improved efficiency without pilot matrix inversion.
Key Points
To develop an efficient off-grid near-field channel estimation framework for distributed MIMO systems that mitigates spherical-wave modeling errors and avoids high computational costs.
Formulated a geometry-coupled spherical-wave model mapping local direction-range parameters into a unified, two-dimensional jointly sparse coordinate system across collinear base station arrays.
Implemented independent-vector variational Bayesian inference directly on received pilot matrices to solve user-specific subproblems without matrix inversion.
Embedded a two-dimensional skewed off-grid expectation-maximization procedure to iteratively refine angle and range offsets on a coarse dictionary.
Numerical simulations demonstrated effective channel recovery across distributed base stations while mitigating basis mismatch errors.
Direct processing of raw pilot matrices eliminated pilot-matrix inversion, preventing distortion of noise statistics under high-dimensional multiuser conditions.