Enhanced Channel Estimation for Flexible Intelligent Metasurface-Aided Communication Systems

Abstract: Flexible intelligent metasurface (FIM) has recently received considerable interest due to its advantage in realizing a better channel condition by dynamically morphing its surface shape. An FIM consists of multiple elements deposited on a flexible substrate. These elements can not only transmit signals, but also adapt their displacements in a direction perpendicular to the FIM surface via an attached controller. In this paper, we consider the channel estimation problem for the uplink of an FIM-enhanced communication system via customizing the orthogonal matching pursuit (OMP) method. Specifically, we formulate an optimization problem of minimizing the column coherence of the measurement matrix by optimizing the FIM's surface shape, subject to the morphing range constraint. Based on the estimated direction of arrival (DOA) and channel gain, we further investigate the signal-to-noise ratio (SNR) improvement in the FIM-enhanced downlink multiple-input single-output (MISO) system. Numerical results demonstrate that an FIM significantly outperforms a conventional rigid uniform planar array (UPA), thereby showing that FIM can substantially improve channel estimation accuracy and achieve SNR improvement, even when using estimated channel parameters.

• The paper presents a DFP and compressed sensing-based approach that reduces uplink NMSE by up to 9.8 dB under FIM morphing, outperforming rigid architectures.
• It employs optimized shape trajectories to minimize measurement matrix coherence, thereby enhancing channel estimation accuracy in multipath environments.
• Results demonstrate a 1.72 dB SNR improvement in downlink transmissions with imperfect CSI, validating the FIM's robust performance.

Enhanced Channel Estimation for Flexible Intelligent Metasurface-Aided Communication Systems

Overview

This work presents a systematic study on uplink channel estimation and downlink SNR enhancement in communication systems augmented by Flexible Intelligent Metasurfaces (FIMs) (2603.29098). Unlike conventional rigid UPAs or RISs, FIM architectures offer programmatically controlled, rapid and accurate surface morphing. The primary contributions are: (1) formulating a customized compressed sensing-based uplink channel estimation protocol grounded on OMP and measurement matrix coherence minimization via FIM morphing; (2) adapting DFP-based shape optimization for simultaneous reduction of matrix coherence and SNR maximization in the downlink, even under imperfect CSI acquired via the morphing-enhanced estimation routine; (3) comprehensive numerical evaluation highlighting significant gains over rigid benchmarks.

System Model and Channel Representation

The FIM-augmented uplink considers an SIMO topology where the surface shape of an $N$-element FIM is a constrained variable over multiple measurement intervals. Channel modeling is executed in the multipath domain, where steer vectors are parameterized not only by incident path geometry but also by real-time vertical displacements (morphing) of the FIM meta-atoms. The parametric form $h(\mathbf{y}) = \sum_{l=1}^L \xi_l g_l(\mathbf{y})$ naturally incorporates shape-dependent variations, thus motivating direct optimization over the morphing vector $\mathbf{y}$ for enhanced system observability.

Compressed Sensing Channel Estimation and Optimization

Capitalizing on the sparsity of wideband millimeter-wave MIMO in the angular domain, the channel estimation pipeline leverages OMP. Critically, instead of fixed array geometries, the measurement matrix $\Psi(\mathbf{y})$ is treated as a function of the FIM's morphing profile at each training slot. By orchestrating shape trajectories to minimize mutual coherence between columns of the aggregated measurement matrix, the system indirectly tightens the restricted isometry property of the underlying sparse recovery problem, directly benefiting estimation accuracy. The optimization is non-convex and high-dimensional. The authors adopt a DFP-based iterative descent with Armijo-Goldstein line search, under strict morphing constraints, ensuring practical feasibility. The update rules analytically compute gradients with respect to morphing variables, exploiting the differential structure of the steering vectors.

Downlink SNR Maximization with Estimated CSI

A subsequent SNR maximization stage is devised for FIM-based downlink MISO. The real-time estimated DOAs and complex gains from the OMP-enhanced uplink are interpolated to reconstruct effective (possibly imperfect) channel state information. The transmit FIM's shape is then re-optimized (again via DFP) to maximize received SNR under an MRT beamforming regime. Notably, the analysis explicitly models and quantifies the impact of non-ideal channel knowledge on achievable SNR, demonstrating robustness of the FIM morphing protocol to upstream estimation errors.

Numerical Results

The empirical results are comprehensive and underscore several strong numerical conclusions:

• NMSE Reduction: When morphing is employed, the FIM reduces uplink channel estimation NMSE by 9.8 dB over rigid UPA benchmarks (SNR = 8 dB, $L=2$ paths) and sustains 3.8–4.1 dB NMSE improvements even as $L\to 5$–$9$.
• Measurement Gains: Augmenting the number of training slots $N_s$ further lowers NMSE, attributing to more orthogonal measurement acquisition via additional shape diversity.
• Morphing Range: Larger allowable morphing range $b$ yields diminishing but noticeable returns in NMSE, with computational convergence of the DFP routine typically requiring $\sim10$ iterations.
• Downlink SNR: The SNR gain at the UE increases with improved channel estimates. FIM achieves 1.72 dB SNR improvement (for $h(\mathbf{y}) = \sum_{l=1}^L \xi_l g_l(\mathbf{y})$0) over UPA even using estimated CSI, and nearly closes the gap to the perfect CSI upper bound when the environment is sufficiently sparse ($h(\mathbf{y}) = \sum_{l=1}^L \xi_l g_l(\mathbf{y})$1).
• The proposed DFP algorithm is guaranteed to converge under resource-constrained hardware actuator limits.

Implications and Future Directions

The results validate that FIM-based architectures, when jointly optimized using coherence-aware shape morphing, substantially outperform rigid array platforms for both channel estimation and data transmission—especially under sparse multi-path regimes relevant for 5G/6G mmWave and THz applications. This work establishes the viability of closed-loop metasurface control, where physical geometry is a core variable in PHY layer tasks.

A significant implication is that spatial DoF expansion via morphing can become a key enabler in environments with limited scattering or in high-mobility/hand-off regimes. The demonstrated resilience to imperfect CSI suggests rapid morphing FIMs can maintain high performance with practical estimation errors.

Future work remains to incorporate multi-user MIMO, fast time-varying channels, actuator nonidealities, and energy constraints. Modeling lifetime and wear-out effects of flexible substrates and controllers, and considering the impact of morphing latency on overall system coherence, could further enhance applicability.

Conclusion

This paper rigorously establishes the practical superiority of FIM-assisted communication systems in channel estimation and SNR enhancement, enabled by direct geometric surface optimization intertwined with compressed sensing and sparsity-based recovery (2603.29098). By closing the loop between physical morphing and digital signal processing tasks, the methodology advances the integration of intelligent surfaces into next-generation wireless architectures. Progressing towards full-stack optimization (including higher-layer protocols and hardware constraints) will further empower intelligent metasurfaces as active components in adaptive wireless infrastructure.