eess.SPNov 14, 2026

Over-the-Air Semantic Alignment via Dual Stacked Intelligent Metasurfaces

Authors: Mario Edoardo Pandolfo, Kyriakos Stylianopoulos, George C. Alexandropoulos, Paolo Di Lorenzo

Organizations: the DIAG and DIET Departments, Sapienza University of Rome, Italy · the National Inter-University Consortium for Telecommunications (CNIT), Parma, Italy · the Department of Informatics and Telecommunications, National and Kapodistrian University of Athens, Greece · the DIET Department, Sapienza University of Rome, 00184 Rome, Italy

Abstract

Semantic communications enable task-oriented information exchange, but suffer from latent-space mismatch when transmitter and receiver employ heterogeneous machine learning models. Existing semantic alignment methods rely on digital processing, introducing computational and latency overhead. This paper proposes a dual stacked intelligent metasurfaces (dual-SIM) architecture that performs semantic alignment entirely in the wave domain. By jointly exploiting programmable metasurfaces at both ends of a link and the wireless channel, the proposed architecture realizes an end-to-end linear transformation between heterogeneous latent spaces. We show phase-only multilayer SIM can approximate both supervised linear aligners and zero-shot Parseval-frame equalizers. We investigate single-SIM, disjoint asymmetric and symmetric dual-SIM, and jointly optimized dual-SIM configurations, considering channel equalization and constructive channel exploitation. For joint design, we develop an alternating projected gradient algorithm with convergence guarantees to first-order stationary points. We further introduce a deep-unrolled network that unrolls gradient-based iterations to reduce computational cost of SIM reconfiguration under channel variations. Numerical results with heterogeneous vision transformer encoders show accurate semantic alignment from moderate to high signal-to-noise ratios. The joint and symmetric disjoint dual-SIM designs achieve the highest accuracy over compact channels, the joint design leads at shallow stack depths, and the advantage over single-SIM designs diminishes at equal depth.

Figures & tables

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

Mar 13, 2026

A Learnable SIM Paradigm: Fundamentals, Training Techniques, and Applications

Stacked intelligent metasurfaces (SIMs) represent a breakthrough in wireless hardware by comprising multilayer, programmable metasurfaces capable of analog computing in the electromagnetic (EM) wave domain. By examining their architectural analogies, this article reveals a deeper connection between SIMs and artificial neural networks (ANNs). Leveraging this profound structural similarity, this work introduces a learnable SIM architecture and proposes a learnable SIM-based machine learning (ML) paradigm for sixth-generation (6G)-andbeyond systems. Then, we develop two SIM-empowered wireless signal processing schemes to effectively achieve multi-user signal separation and distinguish communication signals from jamming signals. The use cases highlight that the proposed SIM-enabled signal processing system can significantly enhance spectrum utilization efficiency and anti-jamming capability in a lightweight manner and pave the way for ultra-efficient and intelligent wireless infrastructures.
Oct 26, 2026

XL MIMO Extreme Learning Machines: Wave-Domain Inference with Stacked Metasurfaces

The recently envisioned goal-oriented communications paradigm requires machine learning inference to be performed directly on wirelessly transferred data, by splitting neural network computations between the transmitter, the Receiver (RX), and the wireless channel. Such approaches aim to offload large portions of the computation onto the channel itself, which is often controlled by reconfigurable MetaSurfaces (MSs). Typical MS-parametrized systems, however, enable only limited computation offloading in the wave domain, usually corresponding to a single neural network layer, due to adverse channel fading effects and the almost linear operation of the channel and conventional MS models. In this paper, we demonstrate that an eXtremely Large (XL) Multiple-Input Multiple-Output (MIMO) system equipped with Stacked Intelligent MSs (SIM) at the RX side, together with a single antenna and its accompanying radio-frequency chain, may act as a two-layer neural network operating exclusively over the air. The first (uncontrollable) network layer corresponds to the transformations offered by wireless fading, while the second (controllable) layer is implemented via SIM operations. In particular, only the first, front-end MS layer needs to exhibit nonlinear response to serve as an activation function. Leveraging the Extreme Learning Machine (ELM) framework, the proposed XL MIMO system architecture is shown to be a universal approximator, and the optimal linear-SIM-based combining is derived in closed form, allowing for fast training even and partial signal observability. Numerical evaluations across diverse datasets showcase that the proposed XL-MIMO-ELM framework achieves classification accuracy comparable to idealized digital models, thereby, proving the viability of low complexity, wave-domain neural computations.
Aug 27, 2026

Over-The-Air Extreme Learning Machines with Nonlinear Stacked Intelligent Metasurfaces

The recently envisioned goal-oriented communications paradigm requires machine learning inference to be performed directly on wirelessly transferred data. This paper presents an eXtremely Large (XL) Multiple-Input Multiple-Output (MIMO) system that operates as an Extreme Learning Machine (ELM) to execute Over-The-Air (OTA) binary classification. To reduce hardware complexity, the receiver is equipped with cascaded metasurfaces terminating in a single radio-frequency chain. A front metasurface layer applies a fixed nonlinear response to the incoming signal, acting as the ELM's activation function. Subsequent tunable linear metasurface layers physically approximate the trained network weights directly in the wave domain. Numerical evaluations across diverse datasets showcase that our XL MIMO architecture achieves classification accuracy comparable to idealized digital models, thereby proving the viability of low-complexity, wave-domain OTA learning.