Neural Network-Assisted Hybrid Model Based Message Passing for Parametric Holographic MIMO Near Field Channel Estimation

TL;DR

This work tackles near-field HMIMO channel estimation governed by the Dyadic Green's function, where the channel depends on a small set of geometric parameters. It introduces an NN-assisted hybrid channel model that replaces the original complex nonlinear mapping with a simpler surrogate, enabling offline training and reliable parametric estimation. Building on this model, the authors formulate a Bayesian estimation problem and solve it with a factor-graph representation using unitary approximate message passing (UAMP), including a forward pass and a Taylor-linearized backward pass. They further extend the framework to a hybrid receiver with reduced RF chains by cascading two UAMP procedures to jointly estimate the pseudo-observations and the channel parameters. Simulations demonstrate substantial performance gains over LS and a mismatched approximate model, with promising accuracy for both channel estimation and user localization, highlighting the method's practical impact for HMIMO sensing-enabled systems.

Abstract

Holographic multiple-input and multiple-output (HMIMO) is a promising technology with the potential to achieve high energy and spectral efficiencies, enhance system capacity and diversity, etc. In this work, we address the challenge of HMIMO near field (NF) channel estimation, which is complicated by the intricate model introduced by the dyadic Green's function. Despite its complexity, the channel model is governed by a limited set of parameters. This makes parametric channel estimation highly attractive, offering substantial performance enhancements and enabling the extraction of valuable sensing parameters, such as user locations, which are particularly beneficial in mobile networks. However, the relationship between these parameters and channel gains is nonlinear and compounded by integration, making the estimation a formidable task. To tackle this problem, we propose a novel neural network (NN) assisted hybrid method. With the assistance of NNs, we first develop a novel hybrid channel model with a significantly simplified expression compared to the original one, thereby enabling parametric channel estimation. Using the readily available training data derived from the original channel model, the NNs in the hybrid channel model can be effectively trained offline. Then, building upon this hybrid channel model, we formulate the parametric channel estimation problem with a probabilistic framework and design a factor graph representation for Bayesian estimation. Leveraging the factor graph representation and unitary approximate message passing (UAMP), we develop an effective message passing-based Bayesian channel estimation algorithm. Extensive simulations demonstrate the superior performance of the proposed method.

Figures (12)

• Figure 1: Illustration of antenna patches and their coordinates.
• Figure 2: The real part of $h^{xx}_{11}$ (a) and (c) and the real part of $h^{xx}_{11}/\exp(ik_0 r_{11})$ (b) and (d).
• Figure 3: Architecture of the neural network.
• Figure 4: NMSE versus the number of hidden nodes.
• Figure 5: Factor graph representation of \ref{['eq:factorization']}