Signal Processing
Stacked Intelligent Metasurfaces Assisted UAV Communications
Chandan Kumar Sheemar, Giovanni Iacovelli, Sourabh Solanki et al. (+2 more)
In this paper, we investigate an unmanned aerial vehicle (
UAV) communication system assisted by stacked intelligent metasurfaces (SIMs), which enable programmable wave-domain signal processing through multiple cascaded metasurface layers. By shifting part of the beamforming functionality from the RF/digital domain to the electromagnetic domain, SIMs allow the realization of energy-efficient hybrid beamforming architectures suitable for aerial platforms. We formulate the joint design of digital precoding, SIM phase configuration, and
UAV positioning for multi-user downlink sum-rate maximization. To solve the resulting non-convex problem, we develop an alternating optimization framework that guarantees monotonic improvement of the objective. Numerical results demonstrate that the proposed SIM-assisted architecture significantly improves spectral efficiency while maintaining low hardware complexity, and highlight the impact of the number of metasurface layers and size of each layer on system performance.
Read the full abstract ▾
All authors Chandan Kumar Sheemar, Giovanni Iacovelli, Sourabh Solanki, Wali Ullah Khan, Symeon Chatzinotas
Systems and Control · Cross-listed from cs.RO
Model Predictive Planner for UAV Navigation in Non-Convex Air Corridors
Henrique Silva Jr., Marcelo A. Santos and Guilherme V. Raffo
This work presents a motion planning framework for
UAV navigation in non-convex urban air corridors. The planner is based on a mixed-integer tracking model predictive control formulation that enforces corridor feasibility and dynamic consistency within a single optimization problem. To guarantee convergence to the target and mitigate the occurrence of local minima induced by non-convex geometry, a shortest-path-based offset cost with feasibility constraints is embedded directly into the planning problem. Numerical simulations show that the proposed formulation generates dynamically valid trajectories that satisfy the corridor constraints and converge to the target without relying on external global planning stages.
Read the full abstract ▾
Signal Processing
Toward Alias-Free Channel Extrapolation in Upper Mid-Band Systems: A Spatial-Frequency-Temporal Tensor Learning Approach
Jiawei Zhuang, Hongwei Hou, Yafei Wang et al. (+4 more)
Upper mid-band massive multiple-input multiple-output (
MIMO) offers a favorable capacity-coverage trade-off for next-generation wireless systems, but its large
antenna arrays, wide bandwidths, and faster temporal variation substantially increase the pilot overhead required for accurate channel state information (CSI) acquisition. To reduce this overhead, this paper establishes a tensor-structured multi-domain channel extrapolation framework that exploits the limited-scattering nature of practical propagation environments to recover complete CSI across the spatial-frequency-temporal (SFT) domains from limited observations. Specifically, we develop a Tucker-based SFT-domain signal model to represent the complete CSI, where the factor
matrices are parameterized by angle-delay-Doppler (ADD)-domain grids. Thanks to this representation, we reveal that limited SFT-domain observations imposed by uniform pilot patterns and
antenna-port selection inherently induce ADD-domain aliasing, so that multiple physically distinct ADD-domain components become indistinguishable within structured ADD aliasing groups. To tackle this issue, we introduce a support-prior-assisted ADD-domain de-aliasing mechanism that leverages coarse-grained support information. Since exact closed-form characterization of this mechanism is difficult to derive, we propose a tensor-structure-aware axial-
attention neural network (TANN), which integrates axis-wise
attention with a lightweight multi-scale CNN-based gating module to incorporate support priors for ADD-domain de-aliasing. With tensor-structure modeling and mixed-configuration training over different pilot decimation factors, TANN yields a unified model that generalizes across pilot configurations without retraining. Numerical results demonstrate the effectiveness and strong generalization of the proposed framework over
benchmark methods under diverse scenarios.
Read the full abstract ▾
All authors Jiawei Zhuang, Hongwei Hou, Yafei Wang, Xinping Yi, Wenjin Wang, Jiangzhou Wang, Björn Ottersten
Audio and Speech Processing
Revisiting Vocos: That Phasiness Business in Time-Frequency Neural Vocoding
Ünal Ege Gaznepoğlu, Frank Zalkow, Mohammad Joshaghani et al. (+3 more)
Recently, time-frequency neural vocoders have been approaching the state-of-the-art quality of time-domain neural vocoders. Vocos is a notable example due to its efficiency, but its audio quality lags behind the time-domain vocoders and the reasons remain debated. Thus, in this study, we revisit Vocos from a phase reconstruction perspective. First, we quantify the gap between time-domain and time-frequency domain vocoders using bandlimited mel spectrograms as inputs. Later, via an ablation study, we verify the Vocos architecture is effective for magnitude modeling, but less so for phase. We then adapt the Vocos backbone to predict phase differences, a precursor for phase reconstruction, and identify 1D convolutional layers are hindering their accurate prediction. Our findings indicate that future research needs to focus on inductive biases that allow the architecture to better model the time-frequency structure of
speech signals, without sacrificing the support for arbitrary input representations.
Read the full abstract ▾
All authors Ünal Ege Gaznepoğlu, Frank Zalkow, Mohammad Joshaghani, Emanuël A.P. Habets, Nils Peters, Christian Dittmar
Systems and Control · Cross-listed from cs.RO
Continual-RL for Generalization in Autonomous Racing on the RoboRacer Platform
Joel Siegert, Edoardo Ghignone and Michele Magno
A key challenge in modern robotics is to adapt to changing environments, a challenge that is exacerbated when simulations cannot encompass every possible real-world configuration, and therefore Reinforcement Learning (RL) in the physical world becomes necessary. Continual Reinforcement Learning provides the tools to address this challenge; however, both the frameworks and the methods remain underexplored. Autonomous Racing and in particular the RoboRacer competition provide a testing ground for such methods, as learning to drive on a new track-floor combination with the least amount of new experience naturally frames a continual learning problem. This work tries to address this gap by proposing a continual RL framework based on Continual Backpropagation that is able, with only real-world data, to train a generalistic policy on a set of tracks and then fine- tune it within 15 minutes to outperform classical controllers. Furthermore, a comparison method based on offline RL is proposed, and a simulation analysis of the plasticity properties of the methods is conducted.
Read the full abstract ▾
Signal Processing
High-Altitude Platform Station-Aided Terahertz Satellite Communication Systems with Hardware Impairments
Evla Safahan Ahrazoglu, Eylem Erdogan and Ibrahim Altunbas
The utilization of terahertz (THz) frequencies in satellite-aerial-ground communication systems stands out as a promising solution to accomplish both global connectivity and extreme data rates requirements of the sixth-generation networks. In the current literature, the impact of non-ideal equipment on the performance of THz satellite-aerial-ground communication systems remains unexplored, which is critical for practical implementations. Hence, this paper analyzes the performance of high-altitude platform station (HAPS)-aided THz satellite communication system in the presence of $\alpha$-$\mu$ fading, pointing errors, absorption loss, and hardware impairments for different atmospheric conditions. In the system of interest, it is assumed that variable-gain amplify-and-forward protocol is utilized at HAPS nodes (systems), and the HAPS system, which provides the maximum end-to-end signal-to-noise ratio (SNR), is selected for transmission. To evaluate the outage, asymptotic outage, and ergodic capacity bounds for the system, the probability density function, cumulative distribution function (CDF), and asymptotic CDF related to the upper bound of the end-to-end SNR are obtained. By using these statistics, the effects of hardware impairment levels, zenith angles, and atmospheric conditions on the system performance are examined. The results have shown that hardware impairments cause power loss in outage performance and reduce the system capacity. Moreover, it is demonstrated that the outage probability depends on either fading or pointing error characteristics in high SNR region and also that the system performance almost remains the same for lower zenith angles in HAPS-to-ground link regardless of the atmospheric conditions.
Read the full abstract ▾
Signal Processing
RPG-VST: Robust Poisson-Gaussian Variance Stabilization for Blind RAW Denoising
Wenbin Wang, Xiaotong Luo, Yuan Gao et al. (+1 more)
Variance stabilization with the generalized Anscombe transform (GAT) enables frozen Gaussian denoisers to process Poisson–Gaussian (PG) RAW noise, but its reliability depends on fitted shot/read-noise parameters. In blind single-image deployment, these parameters are estimated from low-texture RAW statistics that are often corrupted by residual texture, clipping, defective pixels, and read-noise floors. Such contamination yields heavy-tailed log-variance residuals, making ordinary least-squares PG calibration brittle and causing severe tail failures despite favorable average PSNR. We propose RPG-VST, a robust no-reference variance-stabilization framework for blind RAW denoising. RPG-VST estimates PG parameters separately for each color filter array (CFA) plane using a Student-$t$ log-variance objective with robust tile statistics and physical constraints. It then estimates the stabilized-domain noise level $\sigma_z$ from tile variance ratios and uses it as a reliability signal. For each image, RPG-VST selects the robust fit or the conventional OLS fit according to which produces $\sigma_z$ closer to unit variance, requiring no clean reference or learned threshold. On SID Sony SID$50$, SIDD, and ELD with frozen SwinIR and Restormer denoisers, RPG-VST improves mean PSNR in all six
dataset–backbone settings. It reduces severe tails, defined as cases whose PSNR gain over Direct is below $-1$ dB, in four settings and leaves them unchanged in the other two. On SIDD, it yields $+1.83/+1.92$ dB and reduces severe tails from $44/36$ to $7/4$. Ablations show that the $\sigma_z$ gate prevents regressions of ungated robust fitting on read-noise-dominated ELD captures.
Read the full abstract ▾
All authors Wenbin Wang, Xiaotong Luo, Yuan Gao, Wenjun Zeng
Systems and Control · Cross-listed from math.OC
A Nonlinear CTH-Based Adaptive Cruise Controller With Safety and String Stability Guarantees
Dionysios Theodosis, Amirhossein Samii and Nikolaos Bekiaris-Liberis
This paper studies the problem of longitudinal control, with safety and string stability guarantees, in vehicle platoons. A nonlinear adaptive cruise controller based on the Constant Time Headway (CTH) policy is designed for platoons of heterogeneous vehicles, that incorporates a speed-dependent nonlinear gain guaranteeing bounded vehicle speeds and a saturation-type nonlinearity to guarantee bounded (or non-aggressive) accelerations, while preserving collision-avoidance properties of time-headway-based spacing policies. By introducing a suitable spacing-error variable, explicit nonlinear error dynamics are obtained, allowing a direct analysis of the closed-loop system. It is shown that the collision-free region is positively invariant and that the proposed controller guarantees convergence of the spacing errors. Furthermore, nonlinear string-stability properties are established directly from the closed-loop dynamics without resorting to linearization or frequency-domain arguments. In particular, sufficient conditions are derived for $L_p$ string stability, $1\le p\le\infty$, with respect to disturbances, external inputs acting on the leading vehicle, and perturbations of the initial conditions.
Read the full abstract ▾
Signal Processing
Beam-Domain Channel Estimation for mmWave MIMO using Sub-6 GHz Out-of-Band Information
Faruk Pasic, Mariam Mussbah, Stefan Schwarz et al. (+1 more)
Future wireless multiple-input multiple-output (
MIMO) systems will integrate sub-6 GHz and millimeter wave (mmWave) bands to support high data rates for latency-critical applications. The coexistence of these bands enables the use of reliable out-of-band sub-6 GHz information to assist mmWave link establishment, particularly for channel estimation. In this paper, we propose a novel beam-domain channel estimation method for mmWave
MIMO that leverages sub-6 GHz beam-domain information to improve estimation accuracy. We compare the proposed method with conventional in-band methods. Simulation results show that our proposed method outperform existing baselines in terms of spectral efficiency, in both line-of-sight (LOS) and non-line-of-sight (NLOS) scenarios.
Read the full abstract ▾
All authors Faruk Pasic, Mariam Mussbah, Stefan Schwarz, Markus Rupp
Signal Processing
ETSI ISG MAT: Bridging Multiple Access Techniques Research and 6G Standardisation
Dheeraj Raja Kumar, Mert Ozates, Yong Jin Daniel Kim et al. (+11 more)
Improved spectral efficiency and inter-user interference mitigation are important aspects of IMT-2030 and the ongoing 3GPP 6G study. ETSI established the Industry Specification Group on Multiple Access Techniques (ISG MAT) as a research and pre-standardisation activity to build wider consensus on downlink MAT for 3GPP-based 6G systems. Its first report, ETSI GR MAT 001 V1.1.1, provides a standards-oriented comparison of 3GPP-specified techniques, including OMA, MU-
MIMO, and MUST, with candidate techniques comprising power-domain NOMA, RSMA, and cache-aided MU-
MIMO. The report identifies operating conditions in which candidate MAT can improve spectral efficiency and assesses implications for transceiver processing, network assistance information, and reference-signal requirements. Ongoing work includes realistic 5G NR link-level evaluations and the study of MAT for non-terrestrial networks, providing timely technical evidence for 3GPP 6G standardisation discussions.
Read the full abstract ▾
All authors Dheeraj Raja Kumar, Mert Ozates, Yong Jin Daniel Kim, David Vargas, Onur Dizdar, Fehmi Emre Kadan, Carles Antón-Haro, Sebastian Wagner, Bruno Clerckx, Malte Schellmann, Xavier Mestre, H. M. Furqan, Kenza Abela, Shaima Abidrabbu
Systems and Control
The role of symmetries on the stability of power grids
Jingyang Fang, Feng Gao, Xu Yang et al. (+1 more)
The stability of power grids principally depends on the synchronization of multiple generating units. So far, the relationship between grid stability and
symmetries remains unclear. This article fills in the research gap by innovatively using the mathematical tool of group theory. We find that grid
symmetries involve two aspects-network and component
symmetries, which reflect the homogeneities of network structures and generating units, respectively. As disclosed, the stability of power grids has a tight bearing on network and component
symmetries, where corresponding
automorphism group and symmetric subgroup orders reflect their levels of
symmetries, respectively. In the perspective of network
symmetries, where grid stability and synchronization speed are dictated by the algebraic connectivity, the stability ranking list of grids with four generators is K4 > C4 > K13 > R4 > P4 > X4, which fully agrees with that of network
symmetries. In terms of component
symmetries, the stability of power grids largely benefits from improved component
symmetries yet with R4 as an exceptional case, where it is recommended to avoid the centralized effort of a single generator. These conclusions have significant implications for the control of modern power grids with distributed resources and controllable generators, including energy storage and renewable power generators interfaced by grid-following/-forming converters.
Read the full abstract ▾
All authors Jingyang Fang, Feng Gao, Xu Yang, Stephan Goetz
Systems and Control
Analytics for the Optimization of the Soybean Oil Purification Process
Henrik Meyer, Lars Ahlers, Pedro Querini et al. (+4 more)
Machine Learning, Artificial Intelligence, among others, are very promising methodologies and technologies that are emerging for implementing a broad spectrum of analytics within digitalized eco-systems. Analytics containing adequate analytical models are generating a burgeoning interest from Business Intelligence-, Information Technology (IT)- and Operational Technology (OT)-professionals, who are able to exploit the huge amount of internally and externally available data and information that lies behind digitalized components and systems and their associated processes. In this paper, the authors present the essential specifications of an analytical model developed and implemented applying the Knowldege Discovery in Databases (KDD) approach. The analytical model is the essential part of an analytics component, positioned as an digitalized asset within an Industry 4.0 compliant (RAMI 4.0) infrastructure, and used to optimize the industrial Soybean Oil Purification Process associated to the digitalized eco-system.
Read the full abstract ▾
All authors Henrik Meyer, Lars Ahlers, Pedro Querini, Erica Fernandez, Maria L. Caliusco, Martín A. Bär, Armando W. Colombo
Systems and Control
Grid-Forming Enhanced STATCOMs for Wind Power Plant Stability: Design, Modeling, and Control
Prabhat Ranjan Bana, Jean-Philippe Hasler, Rasool Heydari et al. (+2 more)
The increasing integration of renewable energy sources (RES) into modern power systems, particularly in weak grid environments, demands advanced solutions for voltage and frequency stability. This paper demonstrates that Enhanced STATCOMs (E-STATCOMs) with grid-forming capabilities outperform synchronous condensers (Syncons) in wind farm applications. Using a power-admittance-based linear modeling framework and validated through detailed EMT simulations, the study shows that E-STATCOMs provide superior damping of sub-synchronous resonance (SSR), faster dynamic response, and effective suppression of DC components, transient overvoltages, and new oscillatory modes introduced by series compensation. Unlike Syncons, which are limited by mechanical inertia, E-STATCOMs offer tunable control, virtual impedance emulation, and scalable fault current support. These advantages make E-STATCOMs a more robust and flexible solution for stabilizing renewable-dominated grids, especially those with long transmission lines and series compensation. The paper also aims to provide a
benchmark E-STATCOM design criterion that enables stakeholders, such as transmission system operators (TSOs), to assess its suitability without the need for exhaustive EMT simulations.
Read the full abstract ▾
All authors Prabhat Ranjan Bana, Jean-Philippe Hasler, Rasool Heydari, Andrew J. Owens, Francisco Chiuminatto
Signal Processing
Hierarchical THz Near-Field Localization with Subarray Processing and Covariance Correction
Ahmad Dkhan, Yazan Dayoub, Jana El Haj et al. (+1 more)
Terahertz (THz)-band near-field (NF) localization offers high spatial resolution due to short wavelengths and distance-dependent wavefront
curvature in NF multi-
antenna systems. However, large arrays and dense deployments, necessary to mitigate THz path loss, raise the received signal dimensionality, creating computational overhead for localization. Furthermore, traditional two-dimensional (2D) subspace
algorithms suffer from excessive complexity and poor robustness under coherent sources. This paper proposes a hierarchical localization framework based on subarray (SA) processing. The first step performs 1D estimation per SA to estimate local angles. The second step combines SA outputs to estimate distances, reducing the 2D search to two 1D searches. To handle the drawback of coherent sources, a
transformer-based network predicts a covariance correction, refining subspace estimation. Simulations show that the proposed hierarchical
algorithm lowers complexity by four orders of magnitude. The
transformer-based covariance correction improves angular accuracy by 85 % and reduces range error by 6.5 m at 5 dB signal-to-noise ratio in a coherent scenario compared to multiple signal classification (MUSIC).
Read the full abstract ▾
All authors Ahmad Dkhan, Yazan Dayoub, Jana El Haj, Hadi Sarieddeen
Systems and Control
Sensor-Limited Observability and Carrier-Induced Reachability of Low-Order Rotor-Coupled NVH in Production Electric Drives: A Magnetic Co-Energy, Gramian, and Active Projection Framework for Production-Signal Feasibility Analysis
Meng-Chou Wu
This paper is motivated by a practical rotor-coupled Order-2 NVH problem observed under load in a production electric drive. Production controllers typically lack direct measurements of rotor vibration, air-gap deformation, or housing acceleration. The central question is therefore whether a mechanically important low-order vibration can be inferred or influenced using only dq currents, voltage commands, resolver angle, and resolver speed. The selected vibration component is represented as an equivalent air-gap deformation mode in a
magnetic co-energy-based electromechanical state-space model. Finite-time Gramians evaluate current-based observability, resolver-based observability, and plant-level voltage-to-air-gap reachability. The analysis identifies flux-linkage sensitivity as the common bottleneck for passive current observability and passive voltage reachability, while resolver observability depends on a separate torque-projection condition. A mechanically significant deformation may thus remain first-order invisible to passive dq signals because production measurements observe a winding-level global projection rather than local air-gap flux redistribution. To address this degeneracy, the paper introduces carrier-induced active projection. A realizable electromagnetic carrier may create additional measurement and force projections, enabling carrier-on observability, reachability, and sideband signatures in production residuals. Two numerical studies examine passive low-order projection in a 48-slot, 8-pole
symmetry proxy and sideband detectability under finite-window noise and near-order disturbance leakage. They demonstrate the computational structure of the framework but do not constitute FEM-calibrated motor validation.
Read the full abstract ▾
Signal Processing
On-Site Beam Calibration for RIS-Aided Positioning Systems
Mengting Li, Hui Chen, Sigurd S. Petersen et al. (+5 more)
High precision positioning is a key enabler for next-generation communication applications such as smart transportation and augmented reality. Reconfigurable intelligent surface (RIS) technology can enhance positioning by providing additional angular information and improving coverage under obstructed propagation conditions. However, true RIS beams can differ significantly from the simplified or ideal beam response models commonly used in RIS-aided positioning, leading to beam model mismatch and an elevated positioning error floor. This paper proposes an on-site RIS beam calibration framework that reduces this error floor by estimating a realistic 3D RIS beam response model from on-site measurements. The proposed calibration
algorithm first extracts the RIS-reflected channel response from signals received by a calibration
agent sampling the angular range of interest, using delay-domain sparse recovery, and then estimates the beam model parameters with a gradient-based
estimator. To validate the proposed framework, 3D beam patterns under 66 phase modulations were measured and incorporated into simulations. With an angular sampling step of 1 deg, the calibrated model achieves an average beam response similarity of 88.5% with respect to the ground truth, compared with 43.7% for the ideal model. The probability that the absolute lower bound of the positioning error is below 0.5m increases from 0.52 without calibration to 0.74 after calibration, showing that on-site RIS beam calibration effectively reduces the positioning error floor caused by true beam model mismatch.
Read the full abstract ▾
All authors Mengting Li, Hui Chen, Sigurd S. Petersen, Alireza Pourafzal, Huiping Huang, Ming Shen, Mikko Valkama, Henk Wymeersch
Image and Video Processing · Cross-listed from cs.CV
When Low CER is Not Enough: An Analysis of Hallucinations in Vision-Language OCR Systems on Historical Uruguayan Documents
Marina Gardella, Camilo Mariño, Diego Belzarena et al. (+3 more)
Optical Character Recognition (OCR) is a key component in the digitization of historical archives. Recently,
Vision-
Language Models (VLMs) have emerged as strong alternatives to traditional OCR systems, achieving state-of-the-art performance on standard
benchmarks. However, their suitability for archival transcription remains insufficiently understood. In this work, we
benchmark traditional OCR systems and VLM-based approaches on the Berrutti
dataset, a challenging collection of Uruguayan dictatorship-era documents derived from microfilm scans. While VLMs consistently outperform traditional methods in terms of Character Error Rate (CER) and Word Error Rate (WER), we show that these improvements hide a more complex picture. Through a detailed qualitative analysis, we uncover systematic failure modes that are invisible to standard metrics, including orthographic normalization, spurious content generation, and semantic substitutions that preserve fluency while altering meaning. Errors affecting named entities are particularly critical, as they can introduce substantial semantic distortions with minimal impact on CER and WER. These findings reveal a critical gap between quantitative OCR performance and transcription fidelity in real-world archival settings, and highlight the need for evaluation frameworks that go beyond character-level accuracy to capture the semantic reliability of generated transcriptions.
Read the full abstract ▾
All authors Marina Gardella, Camilo Mariño, Diego Belzarena, Ignacio Ramírez, Gregory Randall, Jean-Michel Morel
Systems and Control
Data-Driven Sequential Market Optimization for Front-of-the-Meter Battery Energy Storage Systems
Steffen Kortmann, Hannah Sanders and Andreas Ulbig
The growing integration of Battery Energy Storage Systems into electricity
markets has highlighted the importance of coordinated participation across energy and balancing services to fully exploit their operational flexibility. However, existing revenue-stacking models often simplify
market sequences and neglect the impact of rolling forecasts, leading to unrealistic scheduling and overestimated revenues. This paper addresses this gap by introducing a sequential, data-driven optimization framework for Front-of-the-Meter Battery Energy Storage Systems that mirrors actual
market operations. The framework explicitly models
market mechanisms and its respective Gate Closure Times across Frequency Containment Reserve, automated Frequency Restoration Reserve, Day-Ahead Auction, and Intraday Continuous
markets. Each
market stage optimizes expected revenue over all remaining
markets using updated price forecasts while maintaining feasibility within both technical and regulatory limits. A key contribution is the opportunity-cost-based bidding strategy, which endogenously derives
market-consistent bid prices and quantities from residual capacity optimization. Validation for a representative operating day demonstrates that the framework yields consistent and feasible schedules, effectively adapts to updated forecasts, and minimizes deviations between planned and realized dispatch, thereby enhancing the realism and profitability of BESS operation.
Read the full abstract ▾
Signal Processing · Cross-listed from cs.AI
A Cyclic Adaptation-Generalization Framework with Uncertainty-Guided Self-Paced Learning for Long-Term Brain-Machine Interfaces
Jiyu Wei, Di Hong, Zhanjie Zhang et al. (+3 more)
Brain-Machine Interfaces (BMIs), which link the
brain to external devices, hold great potential in rehabilitation, human performance augmentation, and human-centered robotics. However, invasive BMIs face a critical challenge for long-term deployment due to neural drift, which degrades decoding performance over time and necessitates frequent recalibration. Existing methods designed to mitigate neural drift typically rely on either domain adaptation (DA) or domain generalization (DG) alone and often fail to capture fine-grained distribution shifts across neural subdomains, resulting in limited performance. To overcome these limitations, we propose Uncertainty-guided Self-paced Cycling (UnSPC), a robust framework that synergizes DA and DG for target domain refining under an Uncertainty-guided Self-paced Pseudo-labeling (UnSPL) mechanism. To handle subdomain neural drift across domains, UNSPL is proposed to iteratively mine reliable pseudo-labeled samples with a noise-robust ranking strategy for further fine-tuning. Leveraging these high-quality samples, we introduce a novel Cycling Adaptation and Generalization (CycAG) strategy, which integrates DA and DG within an iterative cycle to progressively mitigate both global and subdomain drift. This cyclic process enables effective
alignment to evolving target distributions while preserving robust and transferable representations, thereby mitigating performance degradation under long-term neural drifts. Extensive experiments on multiple neural decoding
datasets demonstrate the effectiveness and robustness of UnSPC. To our knowledge, our proposed UnSPC is the first to cyclically integrate DA and DG with pseudo-labeling, paving the way toward stable long-term BMI controls.
Read the full abstract ▾
All authors Jiyu Wei, Di Hong, Zhanjie Zhang, Dazhong Rong, Qinming He, Yueming Wang
Signal Processing
Resource-Aware Topology Management for ISAC-Enabled TDOA Localization in IoUT Networks
Ruhul Amin Khalil
Reliable localization in the Internet of
Underwater Things (IoUT) is hindered by limited acoustic bandwidth, long propagation delays, multipath, Doppler shifts, and energy-constrained nodes. This letter proposes an integrated sensing and communication (
ISAC)-enabled
topology management and multi-stage adaptive estimation (
ISAC-TM-MAE) framework for time-difference-of-arrival (TDOA) localization in IoUT networks. In the proposed framework, heterogeneous
underwater sensors, seabed anchors, autonomous
underwater vehicles (AUVs), and surface buoys exploit acoustic
ISAC packets for joint communication and localization. The IoUT network is modeled as a graph, in which informative node pairs are selected subject to acoustic bandwidth, computation, energy, and link-reliability constraints. A D-optimal criterion maximizes the Fisher information
matrix (FIM) over the
underwater region of interest, while the MAE stage refines the source estimate under acoustic noise and channel uncertainty. Numerical results show that the proposed
ISAC-TM-MAE improves localization accuracy, resource efficiency, scalability, and robustness compared with centralized, distributed, and
benchmark TDOA localization schemes.
Read the full abstract ▾
Signal Processing · Cross-listed from cs.AI
Self-Supervised Consistency Enhanced Disentangled Learning for Neural Decoding Generalization in Brain-Machine Interface
Jiyu Wei, Di Hong, Zhanjie Zhang et al. (+3 more)
Brain-Machine Interfaces (BMIs) provide a direct communication pathway between the
brain and external devices, enabling humans to control assistive and robotic technologies, with potential applications in rehabilitation, human motor augmentation, and human-centered robotics. However, due to neural drift, the performance of BMIs decreases over time, posing challenges for long-term viability, particularly for invasive BMIs (iBMIs). Existing solutions suffer from two main drawbacks: (i) difficulty in learning robust neural representations, and (ii) neglecting that neural drift varies across motor parameters (e.g., velocity, direction, and speed). To overcome these limitations, we propose Self-Supervised Consistency enhanced Disentangled Learning (SSCDL), a neural decoding generalization framework built on two key innovations. We first design a backbone model named Consistency enhanced Neural Decoder (CND), using a novel teacher-student consistency constraint with simulated neural signal perturbations to learn robust representations invariant to neural drift. Then, we employ three dedicated CNDs under the Complementary-Disentangled Generalization (CDG) mechanism, which disentangles motor signals into velocity, direction, and speed with inspiration from neural preference theory. This disentangled learning enables SSCDL to capture invariant neural representations from diverse neural preference perspectives, significantly enhancing cross-day generalization. Extensive experimental results show that SSCDL delivers state-of-the-art decoding performance, exhibiting high robustness and cross-day stability. These capabilities underscore its strong potential for long-term interaction in human-centric robotic and fine-grained assistive applications.
Read the full abstract ▾
All authors Jiyu Wei, Di Hong, Zhanjie Zhang, Dazhong Rong, Qinming He, Yueming Wang
Signal Processing
Common-Transmitter Multipath-Aware TDoA Localization for Acoustic Backscatter-Enabled IoUT Networks
Ruhul Amin Khalil
Acoustic backscatter enables low-power Internet of
Underwater Things (IoUT) nodes, but localizing passive/semi-passive tags is difficult under weak returns and surface/seabed multipath. This letter proposes a clock-free common-transmitter time-difference-of-arrival (TDoA) framework, in which one anchor interrogates the tag and synchronized receivers measure the backscattered arrivals. Receiver-side differencing cancels the unknown interrogation time, forward anchor-to-tag delay, and tag switching delay. A weighted nonlinear least-squares
estimator combines signal-to-noise ratio (SNR), timing variance, and root-mean-square (RMS) delay spread. Simulations against received signal strength indicator (RSSI)-only, unweighted, SNR-weighted, and robust TDoA baselines show that at $20$ dB SNR, the root-mean-square error (RMSE) decreases from $4.71$ m to $4.13$ m, and the $5$ m success probability improves from $0.793$ to $0.835$.
Read the full abstract ▾
Systems and Control
Unlocking Directional Radiation in Pinching-Antenna Systems: Geometry-Aware Design and Experimental Verification
Haoyang Li, Weidong Liu, Zhongliang Li et al. (+3 more)
Pinching-
antenna systems (PASS) have recently attracted growing interest as a flexible architecture for creating "last-meter" line-of-sight wireless links through dielectric waveguides and reconfigurable radiation points. While modeling pinching
antennas (PAs) as isotropic point radiators has enabled tractable analyses and demonstrated the performance gains of PASS, their practical radiation characteristics remain underexplored. This article investigates PASS from the perspective of PA geometry. Starting from the physical coupling principle, we explain why PA shape affects the induced polarization current and incorporate directional gain into the channel model. The full-wave simulations are conducted to show how different PA geometries and orientations reshape the internal field distribution and far-field radiation pattern. A 60 GHz prototype
video transmission experiment is further presented to demonstrate the link-level impact of changing PA states. Finally, promising applications enabled by geometry-aware directional PASS are highlighted.
Read the full abstract ▾
All authors Haoyang Li, Weidong Liu, Zhongliang Li, Gaojie Chen, Zheng Yang, Zhiguo Ding
Signal Processing · Cross-listed from cs.IT
Beam Training for RIS-Aided ISAC Systems
Jinho Yang, Hyeongtaek Lee and Junil Choi
As a key technology for 6G, integrated sensing and communication (
ISAC) is receiving considerable
attention, and deploying a reconfigurable intelligent surface (RIS) can enhance both communication performance and sensing capability of
ISAC by providing additional degrees of freedom. In this paper, we investigate a beam training framework for RIS-aided
ISAC systems where beam
alignment for a communication user equipment (UE) is conducted while simultaneously detecting a single target through its echo signal. Using codebooks constructed according to the principles of the 5G standard, we propose a partial search procedure that achieves low training overhead and mathematically show that this strategy is sufficient to identify a suitable codeword combination to serve the UE. By applying the auxiliary beam pair method, the target's angle information from the perspectives of the base station and RIS is obtained. Then, a high-accuracy closed-form localization is proposed based on the angle estimates, and we further extend the proposed technique to multi-target localization scenarios. Numerical results highlight the advantages of the proposed technique in the
ISAC context, showing that the training procedure can effectively find a codeword combination and that the target localization technique outperforms the
benchmarks.
Read the full abstract ▾
Systems and Control
Understanding the role and value of battery storage in renewables-rich distribution grids under uncertainty and different market paradigms
Peer Valentin Brigger, Vineet Jagadeesan Nair and Anuradha Annaswamy
Economy-wide electrification is driving steep load growth, while renewables and batteries are increasing pressure at the distribution grid edge often requiring costly upgrades. Local Electricity
Markets (LEMs) can coordinate distributed energy resources (DERs) more efficiently than conventional designs. This paper extends the LEM framework by modeling photovoltaic (PV) inflow as an Ornstein-Uhlenbeck (OU) process to capture uncertainty and temporal correlation. Using rolling-horizon alternating optimal power flow on the IEEE 123-node feeder, we compare operator, customer, and system costs plus storage value under three
market designs: flat tariff (FT), real-time pricing (RTP), and LEM. Results show that the LEM achieves the lowest customer and system costs, under uncertainty, while RTP performs worst. It outperforms the other
markets across most storage and
solar penetration levels. However, it may worsen customer outcomes at very low PV penetration, indicating that LEM adoption should accompany PV expansion and precede storage expansion. We examine two approaches to estimating battery storage value. Storage value depends on storage penetration, imbalance between storage and PV penetrations, location, and whether batteries are colocated with renewables and/or load. Poorly placed units can worsen congestion, whereas well-sited units benefit owners and the system. Both the marginal value of new storage and the stored energy value of existing capacity decline as PV penetration rises. However, they respond oppositely to battery penetration: marginal value falls because of diminishing returns as the fleet grows, while stored energy value rises as more batteries compete for the same local PV generation. These metrics capture different dimensions of storage value under a LEM: one measures the benefit of expanding the fleet, while the other measures the scarcity premium on energy already stored.
Read the full abstract ▾
Audio and Speech Processing
Leveraging Gradient Reversal Loss and Multitask Learning for Datasets-Aware Audio Deepfake Detection
Mingrui Liang, Thomas Thebaud, Lukasz Wojciak et al. (+4 more)
Recent advances in
speech synthesis and voice conversion, which pose threats to security and
privacy, have underscored the need for deepfake detection technology. Although existing detection systems achieve strong performance on individual
datasets, they often fail to generalize across diverse
datasets. Prior methods for improving generalization, including data augmentation,
adversarial training on auxiliary factors such as
language or codec types, and Mixture-of-Experts (MoE), are limited by predefined augmentation coverage, difficulties in obtaining auxiliary factors, and substantial model complexity. In this work, we propose a practical
dataset-aware framework for deepfake detection. Our method targets heterogeneous
datasets for which auxiliary annotations such as
language, codec, or spoofing method may not be consistently available. We therefore rely only on
dataset identity as a naturally available supervisory signal for multitask (MT) and gradient reversal layer (GRL) training, allowing the model to investigate both
dataset-aware multitask supervision and
adversarial suppression of
dataset-specific information. We conduct experiments following the 2025
Speech DeepFake Arena
benchmark protocol, evaluating our model across multiple evaluation
datasets and reporting aggregate performance in terms of Equal Error Rate (EER), including Average EER and Pooled EER. Compared with the baseline, MT reduces Average EER by 13.14% relatively, while GRL reduces Pooled EER by 5.32% relatively. These results demonstrate that our method can improve aggregate detection performance across heterogeneous evaluation
datasets, offering a practical solution for deploying reliable deepfake detection systems on diverse and unseen real-world data.
Read the full abstract ▾
All authors Mingrui Liang, Thomas Thebaud, Lukasz Wojciak, Laureano Moro Velazquez, Yishay Carmiel, Jesus Villalba Lopez, Najim Dehak
Systems and Control
Reverse-Time Diffusion Processes for Discrete Time Linear and Nonlinear Systems with non-Gaussian Noise
Soura Dasgupta, Brian D. O. Anderson and Raghuraman Mudumbai
Generative AI relies on finding reverse time models for a discrete-time forward
diffusion with non-Gaussian initial state, but uses indirect approaches as there is no theory for direct reversal in discrete time. This paper develops a theory for directly finding reverse
diffusions for discrete time nonlinear processes with non-Gaussian states and process noise. We also give a necessary and sufficient condition for the reverse model to be input-affine when the forward process is linear and the process noise Gaussian, and show that for a wide variety of state densities an input-affine reverse
diffusion does not exist. This is among several differences between the reversal of
stochastic difference equations and their continuous time counterparts.
Read the full abstract ▾
Audio and Speech Processing
Qwen-Audio-3.0-TTS: Freely Controllable and Highly Robust Speech Synthesis with Multi-Stage Training Paradigm
Bajian Xiang, Cheng Wen, Han Zhao et al. (+12 more)
In this report, we present Qwen-Audio-3.0-TTS, a production-oriented
speech synthesis system that jointly advances content consistency, speaker similarity, prosodic naturalness, audio quality, controllability,
multilingual coverage, efficiency, and robustness. It combines a 12.5~Hz low-frame-rate
speech tokenizer for reduced inference latency with a five-stage progressive training paradigm for coordinated
language model (LM) and flow-matching model (FM) optimization. The model provides production-level control through free-style natural-
language instructions and fine-grained inline tags, while supporting 16
languages, 20 Chinese dialect regions, one-pass long-form synthesis up to 3 minutes, and robust generation from noisy, reverberant, or unclear reference
speech. Across SEED-TTS-Eval, CV3-Eval, instruction-following, long-form, and acoustic-robustness evaluations, Qwen-Audio-3.0-TTS achieves state-of-the-art performance on many reported dimensions or the strongest aggregate results. It also ranks first on the independent Artificial Analysis Text-to-
Speech Leaderboard. These results establish Qwen-Audio-3.0-TTS as a strong foundation for production-level
speech synthesis.
Read the full abstract ▾
All authors Bajian Xiang, Cheng Wen, Han Zhao, Hao Wang, Haoxu Wang, Jiawei Jin, Jiayan Cui, Jie Chen, Mengxi Nie, Tianyu Zhao, Weiqin Li, Xiang Lv, Xiangang Li, Yang Xiang, Yang Zhou
Systems and Control
Bridging Reinforcement Learning and Optimal Control via Feasible Action Mapping
Stefan Richter, Alberto Giammarino, Guillem Torrente et al. (+2 more)
Operating constrained dynamical systems requires controllers to efficiently solve complex tasks while enforcing recursive feasibility and physical constraints. To address these competing requirements, we present Feasible Action for Optimal Control (FAOC), a novel control framework integrating Reinforcement Learning (RL) and Optimal Control (OC). The core contribution is a computationally efficient, optimization-based mapping
algorithm that transforms the RL
agent's action from a static abstract set into a state-dependent feasible parameter set of the Optimal Control Problem (OCP), guaranteeing instantaneous parameter feasibility. When paired with invariant terminal sets, FAOC guarantees strict recursive feasibility and safe operation, effectively combining the predictable safety of OC with the behavioral flexibility of RL. Unlike prior work, the abstract action space does not require expert tuning, nor is the OCP formulation compromised by the inability of RL to guarantee feasibility. We evaluate FAOC on real-time motion planning for
robot table tennis, where simulated experiments demonstrate superior sample efficiency and closed-loop performance compared to state-of-the-art baselines. We open-source the used implementation of the mapping
algorithm and OCP for motion planning https://github.com/SonyResearch/feasible_action_for_optimal_control.
Read the full abstract ▾
All authors Stefan Richter, Alberto Giammarino, Guillem Torrente, Sam Blakeman, Peter Dürr
Signal Processing · Cross-listed from math.ST
The Phase Transition in Online PCA Depends on $n/d\log(d)$, not $n/d$
Apratim Dey
High dimensional statistical theory has established the importance of constant aspect ratio, when the number of dimensions ($d$) and samples ($n$) satisfy $n,d\to\infty$ with $n/d\to \gamma\in(0,\infty)$, in understanding the limits of canonical estimation problems. In particular, for estimating the top eigenvector of a $d\times d$ population covariance
matrix from $n$ iid samples, the BBP phase transition gives a precise threshold – a simple functional of the aspect ratio – such that the top sample principal component attains nonzero asymptotic correlation with the truth only when the leading population
eigenvalue exceeds it. In this paper, we show that for online / streaming
algorithms the story is very different, and constant aspect ratio is insufficient for nonzero overlap. We study Oja's
algorithm, the most popular method for online PCA. Let $\Sigma=\theta^2 v_0v_0^\top+I\in\mathbb{R}^{d\times d}$, and run Oja's
algorithm with step size $\delta/d$ on $n$ iid samples $X_k\sim\mathcal{N}(0,\Sigma)$, with output $\hat v_n$. Then, as $n,d\to\infty$ with $n/d\log d\to\gamma\in(0,\infty)$, we establish a phase transition: $|\langle\hat v_n,v_0\rangle|\to 0$ when $\gamma<\gamma_*$, and $\to\rho_*$ when $\gamma>\gamma_*$. Here $\rho_*=\rho_*(\theta,\delta)=\sqrt{(\theta^2-\delta/2)_+/\theta^2(1+\delta/2)}$ and $\gamma_*=\gamma_*(\theta,\delta)=1/2\delta(\theta^2-\delta/2)_+$. Further, at criticality, when $n=[\gamma_*d\log d+\eta d]$ and $d\to\infty$, $\eta\in\mathbb{R}$, the correlation is random: $|\langle\hat v_n,v_0\rangle|\stackrel{w}{\to}\rho_*|G|\exp(\eta/2\gamma_*)/\sqrt{\rho_*^4+G^2\exp(\eta/\gamma_*)}$ where $G\sim\mathcal{N}(0,1)$. This is in stark contrast to ordinary high dimensional PCA, where nonzero overlap is possible at constant $n/d$ and improves as $n/d$ increases.
Read the full abstract ▾