One of the main events of SIC’26 is the poster session on the afternoon Tuesday July 7th. The poster session is an excellent opportunity for workshop attendees to get to know each other and to discuss the research topics in more detail. We hope that this will result in a relaxed atmosphere for networking, informal discussions, and the exchange of research ideas.
CALL FOR POSTERS NOW OPEN!!!!
If you are willing to present a poster with recent research results, please register here.
Posters must be A0 (841 x 1189 mm) size and in portrait orientation.
Posters List
Pedro Crespo Bofill, Iñigo Moreno
DERIVATION OF Q. ERROR CORRECTION CONDITIONS VIA THE Q. COMPLEMENTARY CHANNEL
We derive the Knill-Laflamme Quantum Error Conditions for quantum codes by ensuring that the information encoded by an error correcting code does not leak to the enviroment.
Mar Elizo, Alejandro Almodóvar, Patricia A. Apellániz, Santiago Zazo, Juan Parras
KOLMOGOROV-ARNOLD CAUSAL GENERATIVE MODELS
Causal generative models provide a principled framework for answering observational, interventional, and counterfactual queries from observational data. However, many deep causal models rely on highly expressive architectures with opaque mechanisms, limiting auditability in high-stakes domains.We propose KaCGM, a causal generative model for mixed-type tabular data where each structural equation is parameterized by a Kolmogorov-Arnold Network (KAN). This decomposition enables direct inspection of learned causal mechanisms, including symbolic approximations and visualization of parent&ndash.
Liz Gámez Picó, Tomás Domínguez-Bolaño, Ángel Carro-Lagoa ,Valentín Barral, Carlos J. Escudero, José A. García-Naya
ROBOLOC-G: AN OPEN MULTIMODAL WIRELESS SENSING DATASET FROM A CALIBRATED GANTRY ROBOT TESTBED
RoboLoc-G is a high-precision multimodal dataset for indoor localization, radio map estimation, and wireless sensing research. It was acquired with a calibrated gantry robot testbed that executes repeatable trajectories while synchronously recording UWB, Wi-Fi FTM, mmWave radar, IMU, video, and OptiTrack reference measurements. The dataset addresses common limitations of existing resources, including reliance on synthetic data, weak ground-truth traceability, and limited synchronization across modalities. Ground truth is generated from gantry encoder positions, calibrated coordinate transformations, and sensor-specific offsets estimated with OptiTrack. The resulting dataset provides controlled dynamic trajectories, millimeter-level reference validation, FAIR-compliant distribution, and open code for reproducible benchmarking of localization, sensor fusion, and GenAI-based wireless sensing methods.
Fabián González, Deniz Akyildiz, Dan Crisan, Joaquín Míguez
MODEL MISSPECIFICATION IN NONLINEAR FILTERING: STABILITY AND ROBUSTNESS UNDER RANDOM OBSERVATIONS
We study the problem of quantifying the error in Bayesian filtering when the model dynamics are misspecified. Specifically, we consider an optimal filtering recursion governed by a family of predictive operators $K={K_t}_{t\ge 1}$, associated with the true underlying dynamical model, and an approximate filter following a perturbed family of operators $\widehat{K}={\widehat K_t}_{t\ge 1}$, both combined with a common family of update operators $U={U_t}_{t\ge 1}$ associated with the observation model.To analyze the propagation of the error over time, we adopt a fully probabilistic setting in which observations are treated as random variables rather than fixed realizations. This perspective enables us to characterize the evolution of the filtering distributions in expectation and to derive a linear recursion for the mean measures of the optimal filter. The contraction of predictive operators interacts with the probabilistic structure of the random observation updates and with the accumulation of model discrepancies over time, jointly shaping the evolution of filtering errors under misspecified dynamics.Within this framework, we obtain explicit bounds for the discrepancy between the optimal filter and its misspecified counterpart. These bounds decompose the total error into two key contributions: the contraction induced by the dynamics and the perturbation introduced by the model mismatch $\widehat{K} – K$. As a consequence, the analysis of error propagation becomes closely connected to the stability and contraction of the predictive operators (K) and (\widehat{K}), which can be quantified through their Dobrushin coefficients. Under suitable contractivity conditions on the dynamics, we show that the cumulative error stabilizes at a steady-state level determined by the magnitude of the model misspecification. Finally, we provide several analytical examples corresponding to realistic scenarios in which our results hold. Specifically, we prove that if the transition functions between states are bounded and the transition noise belongs to a family of elliptically symmetric distributions (e.g., Gaussian, Student t, Laplace, etc.), then the error between the true and misspecified filters stabilizes as predicted by the theory.This approach establishes a theoretical foundation for understanding how random observations and model perturbations interact, enabling a systematic investigation of filter robustness.
Hiba Hawess, Felipe Gómez Cuba
MMWAVE AOA ESTIMATION IN MULTI PANEL ARRAYS WITH 3GPP CHANNEL SPATIAL CONSISTENCY MODEL
he sparsity of Millimeter-Wave (mmWave) multi-path channels enables Compressed Sensing (CS) estimation from a limited number of observations to overcome the bottleneck imposed by Hybrid Beam-Forming (HBF). However, future Sixth Generation (6G) cellular systems in centimeter to Terahertz bands using eXtra-Large MIMO (XL-MIMO) demand more challenging models beyond the planar wave. This paper proposes a practical model for multi-panel arrays based on the Third Generation Partnership Project (3GPP) channel model spatial consistency extension. The proposed model is designed for practicality and simplifies multi-panel array modeling while retaining validity for practical array sizes in commercial applications. We compare CS channel estimation performance under a single-panel baseline, a uniform planar wave multi-panel model and the proposed 3GPP spatial consistency multi-panel model. Simulation results show that the proposed model supplies an additional number of observation to the CS algorithm and achieves a consistent Normalized Mean Squared Error (NMSE) improvement over the single-panel baseline, while the planar wave two-panel model fails to represent the channel correctly for link distances below the Fraunhofer distance.
Francisco Javier López, Martínez Taissir Elganimi, Pablo Ramirez-Espinosa, David Morales-Jimenez, F. Javier López-Martínez
CHANNEL ESTIMATION AND RECONSTRUCTION IN FLUID ANTENNA MULTIPLE ACCESS: MYTHS, MISCONCEPTIONS AND CRITICAL QUESTIONS
Fluid antenna systems (FAS) represent a paradigm shift in which antenna elements (ports) emulate the illusion of motion or fluidity within a spatial aperture to optimize performance. One of FAS’s key use cases is the provision of open-loop fluid antenna multiple access (FAMA), enabling multiplexing gains through spatial interference nulling without requiring channel state information (CSI) at the transmitter side. However, this comes at the price of requiring a precise channel reconstruction at the receiver to successfully identify the optimal port. Current research efforts map this sensing task to a legacy MIMO-style estimation problem focused on minimizing global reconstruction errors such as normalized mean-squared error (NMSE). In this work, we argue that because FAS is inherently selection-based, NMSE-like approaches often lead to excessive training overhead and reduced net throughput. We revisit the problem of channel estimation and reconstruction in FAS, challenging some prevalent myths related to (i) the adequacy of global error metrics.
David Luengo, Thomas Trigano, Antonio Squicciarini
JENSEN-TSALLIS DIVERGENCE FOR SUPERVISED CLASSIFICATION UNDER DATA IMBALANCE
In supervised classification problems, the loss function is typically based on the Kullback-Leibler divergence. However, alternative entropic divergence formulations, such as the Jensen-Shannon Divergence (JSD), have recently garnered attention for their unique properties. In this study, we delve deeper into the interpretation of the JSD and its generalized form, the Jensen-Tsallis Divergence (JTD), as alternative loss functions for supervised classification. When provided with one-hot encoded distributions for the true label probabilities, we demonstrate that these novel divergences impose an intrinsic output confidence regularization that prevents overfitting. Additionally, the q non-extensive parameter of the JTD directly influences the structure of the regularizer, offering increased flexibility in the formulation of the loss function. Through experiments conducted on artificially imbalanced versions of CIFAR-10, MNIST, and fashion-MNIST we showcase how JTD outperforms JSD and other traditional loss functions in terms of generalization performance, especially for highly imbalanced datasets.
Roberto Maneiro Catoira, Julio Brégains, José A. García-Naya, Luis Castedo
TIME-MODULATED ARRAYS WITH SIGNAL ENCRYPTION FOR ENHANCED DIRECTIONAL MODULATION SECURITY
Directional modulation (DM) implemented with time-modulated arrays (TMAs) offers a competitive size, weight,power, and cost (SWaP-C) profile, high-resolution beam steering, and multi-beam capability, making TMAs a promising alternative to conventional phased arrays for secure communications. TMAs implement a direction-dependent distortion process (D3P)through mechanisms that are inherently non-invertible, such as spectral aliasing of harmonic signal replicas or direct symbolmodulation at the antenna. As a result, these D3Ps cannot provide signal encryption and are not compatible with generic pre-modulated signals in the second case. In this letter, we present a novel DM-TMA with signal-encryption capability that can operate as an add-on antenna for generic transmitters. By pseudorandomly varying the time delays applied to the periodic TMA modulating signals, two independently beamsteered harmonic beams are synthesized, providing DM with enhnanced security.
Francesc Molina, Albert Guillén i Fàbregas
CHANNEL CAPACITY UNDER STOCHASTIC DECODING
Reliable communication limits are typically established under the assumption that perfect knowledge of the channel likelihood is available to the receiver. In many practical settings, however, this assumption is overly idealistic
Rodrigo Oliver Coimbra, Ricardo Vazquez Alvarez, Alejandro Lancho, Stefano Rini
NARRAS: EDGE-TRIGGERED DISTRIBUTED INFERENCE FOR CSI-BASED LOCALIZATION IN VEHICULAR IOT NETWORKS
CSI-based localization with spatially distributed antenna arrays exposes a basic resource trade-off. Each array can provide a rich view of the channel, but forwarding observations from all arrays to a fusion center is wasteful when only a few carry useful information, and the shared uplink supports only a limited number of simultaneous transmissions. We let each array decide locally whether its current observation is worth reporting, subject to a budget on the average number of active transmitters. We refer to this abstraction as Edge-Triggered Distributed Inference (ETDI). It captures a broader class of task-oriented communication problems where resource-constrained devices share an access channel for a common inference task. We instantiate ETDI for CSI-based localization, a common scenario in vehicular IoT networks. Spatially distributed remote antenna arrays (RAAs) encode local channel state information (CSI) from user equipment (UE) transmissions into latent features, and the fusion center estimates the UE position from the subset of reported features. We propose NARRAS, a decentralized reporting policy in which each RAA combines a recurrent summary of its recent observations with a memory of the last latent it transmitted. Training controls an explicit activity budget through differentiable activity penalties and validation-calibrated deterministic thresholds, and uses channel-chart regularization to shape the latent geometry. Experiments show that, at comparable uplink activity, NARRAS improves localization accuracy over learned and heuristic sparse-reporting strategies, while dense full-report models remain useful budget-free references. In low-activity regimes, chart regularization further reduces high-percentile localization errors, suggesting that geometry-aware latent representations are more robust under sparse reporting.
Darian Perez, Adan Oscar Fresnedo, Pedro Suarez-Casal, Luis Castedo, John S. Thompson
LEARNED LATTICE ENCODERS FOR LOW-LATENCY ANALOG JSCC IN SIMO WIRELESS SENSOR NETWORKS
We address the transmission of spatially correlated analog sources in wireless sensor networks (WSNs) over fading SIMO multiple access channels under low-latency constraints. The proposed learned encoder relies on a multi-dimensional lattice-based analog joint source-channel coding (JSCC) mapping. The two most computationally demanding conventional encoder operations&mdash
David Ramírez
DISTRIBUTION OF THE FIRST-ORDER DETECTOR FOR MATCHED SUBSPACE DETECTION ON GRAPHS
Matched subspace detection (MSD) is a powerful tool recently generalized from Euclidean data to graph signal processing. In a previous work we derive the first-order generalized likelihood ratio (GLR) test that operates without knowledge of the noise variance and the GFT coefficients by estimating them via maximum likelihood. In this work we derive the distributions of the first-order GLR.
Miguel Rivas Costa Carlos Mosquera
DISTRIBUTED SOURCE LOCALIZATION WITH ANTENNA ARRAYS AND CRB-BASED FUSION
Source localization with distributed networks is hindered by the high communication overhead associated with raw data transmission from nodes. We propose a framework for antenna arrays, applicable to both mobile platforms and
Vahid Vahidpour, Roberto López-Valcarce, Josep Sala-Álvarez
CSIT-AWARE SPECTRAL PRECODING FOR OUT-OF-BAND POWER REDUCTION IN OFDM
We propose a CSIT-aware spectral precoding scheme for multicarrier modulation that reduces out-of-band radiation (OBR) under transmit-power and mutual-information constraints.The precoder is computed through a fixed-point algorithm that combines reverse waterfilling with semi-unitary spectral shaping. By exploiting the spectral invariance of unitary rotations, CSIT is incorporated without changing the transmit PSD, enabling ZF and SIC receiver designs. Simulations over block-fading multipath channels show that the proposed scheme achieves stronger OBR suppression and better symbol-error-rate performance than PSP-OFDM and null-subcarrier precoding, with SIC performing best at medium-to-high SNR.
Marc Vilà Insa, Jaume Riba
SINGULAR DETECTION IN NONCOHERENT COMMUNICATIONS
This work proposes a general analysis of codeword detection in noncoherent communications. Motivated by the existence of error floors in various regimes, fundamental characteristics of signal design are investigated. In particular, the necessary and sufficient conditions for asymptotically singular detection (i.e. error-free in the limit) are derived from classical results in detection theory. By leveraging tools from linear algebra and the theory of Hilbert spaces, we are able to characterize asymptotic singularity in two main scenarios: the large array and high SNR regimes. The results generalize previous works and extend the notion of unique identification, as well as re-contextualize the geometry of Grassmannian constellations from an alternative perspective.
