Papers

We introduce time-reversed imaging, a new paradigm that infers what just happened in a scene from fading multimodal traces. Instead of extrapolating or interpolating video frames, our goal is to infer past human-environment interactions from residual physical imprints observable in thermal, ultraviolet, and visible spectra. To study this problem, we present TRACE-HEI, the first proof-of-concept dataset for time-reversed imaging, containing synchronized tri-modal video sequences of actions such as sitting, touching, moving objects, and liquid spills, captured across diverse materials and recorded up to three minutes after contact. To establish the benchmark, we propose a multimodal inference approach that extracts structured textual descriptions of detected traces and uses them to constrain a vision-language-guided diffusion model for reconstructing plausible past frames. Experiments show that inferring recent events from fading traces is challenging but feasible when complementary modalities reduce solution ambiguity. This work defines the first computational and experimental foundation for time-reversed imaging, bridging vision, physics, and generative reasoning, and opening new directions for scene understanding beyond instantaneous observation.

Ill-posed inverse problems are encountered in numerous applications, possibly characterized by a highly non-linear forward model, both additive and multiplicative sources of noise, and censored data. In the absence of ground truth, uncertainty quantification is crucial to assess estimation reliability. This motivates the use of a Bayesian model and stochastic inference methods such as Markov Chain Monte Carlo algorithms. Problems combining all these challenges often lead to a complex and potentially multimodal posterior distribution, difficult to handle in practice. Approximate approaches have been proposed in the literature by either neglecting a source of noise or by using a tractable approximation of the likelihood function. These approaches either lead to an inaccurate model, or may require a complex calibration of the approximate likelihood. This paper proposes to tackle such problems with a general hierarchical Bayesian model and an efficient MCMC algorithm. The proposed formulation bypasses the need for calibrating the hyperparameters of an approximate model and is more versatile. The proposed method is assessed on a challenging scenario encountered in astronomy using synthetic data, in a variety of noise and censoring configurations. Comparisons are conducted against two baselines and a state-of-the-art method applicable in this context. The proposed approach is general and yields state-of-the-art results in terms of point-wise estimates and computing costs, with superior predictive performance. Results in the supplementary material further complete this comprehensive and rigorous model study. This work can serve as a guide for practitioners to select the best likelihood model according to their specific application.

This paper investigates physics-informed neural networks for modeling the full dynamic behavior of droop-controlled grid-forming converters. The approach is trained on synthetic data generated via numerical solvers and benchmarked against both traditional integration methods and a vanilla neural network. Results show higher predictive accuracy than the vanilla network using the same training data and substantially reduced runtime compared with numerical solvers.

The rapid expansion of offshore wind energy is central to the European Union's climate-neutrality targets, with High Voltage Direct Current-connected offshore wind power plants (HVDC-OWPPs) becoming increasingly important for integrating gigawatt-scale renewable generation over long distances. Simultaneously, the phase-out of conventional synchronous generators, together with the gradual withdrawal of renewable-energy subsidies, are creating growing demand for ancillary service provision from HVDC-OWPPs as alternatives. This paper presents a comprehensive review of fast frequency services from HVDC-OWPPs in the European context, focusing on inertia support, fast frequency reserve (FFR), and frequency containment reserve (FCR). The review covers both mandatory grid-code requirements and commercial service products, including the latest developments from ENTSO-E, ACER, NESO, TenneT, and Energinet. Furthermore, the paper discusses the key technical gaps and challenges in the practical implementation of fast frequency services from HVDC-OWPPs, highlighting the limited fast response capacity, complexity of stacking optimization, asymmetric provision of inertia, and overall coordination of an HVDC-OWPP system.

Modeling shared and subject-specific structure in multisubject spatiotemporal data remains challenging, particularly in neuroimaging, where both spatial and temporal patterns exhibit rich variability across subjects. Existing matrix and tensor decompositions provide interpretable factorizations, but rely on fixed multilinear structures or coupling schemes that may limit their flexibility in capturing complex variability. In this work, we introduce a spatiotemporal variational tensor decomposition (ST-VTD) framework that combines a tensor factorization generative model with structured priors to jointly represent spatial maps and temporal dynamics. Spatial factors are regularized to promote a low-rank structure inspired by the LL1 decomposition, while temporal factors are modeled using a learned Long short-term memory (LSTM)-based prior, enabling flexible and adaptive dynamics. Posterior inference is performed using an amortized variational formulation by unrolling iterations of an optimization algorithm, leading to an interpretable and parameter-efficient architecture. The proposed inference framework employs a warm-start strategy based on group independent component analysis, which we found to improve optimization performance. Experiments on a realistic synthetic functional MRI (fMRI) dataset demonstrate that the proposed approach significantly improves latent factor recovery compared with representative classical and probabilistic decomposition benchmarks.

Discrete-time Poisson channels with finite intersymbol interference provide a natural model for direct-detection optical links in which multipath memory and signal-dependent shot noise appear simultaneously. Under peak and average optical-intensity constraints, we study the independent and identically distributed (i.i.d.)-constrained capacity problem of such channels. We prove that every input distribution maximizing the stationary mutual information rate within the i.i.d. input class has finite support. The proof is carried out directly on the entropy rate of the continuous-state hidden Markov output process induced by the finite-memory channel. We first establish a filtering-forgetting estimate whose constants are uniform over all admissible i.i.d. input laws. We then derive the entropy-rate first variation, construct a holomorphic extension of the corresponding influence function, and combine the Karush-Kuhn-Tucker condition with a supralinear growth argument.

Indoor localization using Distributed Multiple-Input Multiple-Output (D-MIMO) and machine learning (ML) achieves sub-centimeter accuracy but faces midhaul capacity bottlenecks when transmitting raw Channel State Information (CSI) in Open Radio Access Networks (O-RAN) architectures. To address this, we propose a lightweight, distributed ML framework that shifts initial processing to the network edge. By deploying localized models as dApps on Distributed Units (DUs), each requiring just 1.39 MB of memory and 1.96 MFLOPs, the system performs CSI feature extraction and reduction on the edge. The reduced low-dimensional features are transmitted to the Central Unit (CU), where another dApp is deployed for location estimation. Evaluated on a high-density dataset, this framework reduces midhaul traffic by 100x while maintaining an average error of 8.5 mm, even with half the deployed Radio Units (RUs), providing a scalable blueprint for practical D-MIMO localization.

Implant medical wireless sensors for monitoring physiological parameters, automatic drug provision, and so on represent a new promising healthcare technology. Inherent characteristics of ultra-wideband (UWB) radio make this technology highly suitable for the wireless interface of implant sensors. A communication channel model is essential for developing these wireless systems. However, there are currently few models describing the radio propagation inside the human body. To address this problem, a statistical model is presented for UWB propagation channels inside the human chest in the 1-6 GHz frequency range. The proposed statistical model is developed from numerical simulations using a heterogeneous anatomical model that includes the frequency-dependent dielectric properties of different human tissues. Mathematical formulas for the computation of path loss, scattering and the statistical implementation of the channel impulse response at different depths inside the chest are described. Two typical depths for implanted sensors in the chest, namely 20 and 80 mm are analysed in detail. Average path loss of approximately 20 and 50 dB is observed in each case, respectively. Moreover, the channel exhibits little time dispersion with a root-mean-square delay spread below 1 ns in both cases. These results aim at facilitating the tasks associated with the design of in-body medical communication systems.

Acquiring dense high-frequency channel state information (CSI) is costly in multi-band low-altitude wireless systems because pilot resources are limited and channel dimensions change with the carrier frequency, bandwidth, and antenna array size. We address cross-band CSI reconstruction, which recovers dense target-band CSI from dense source-band CSI and sparse, noisy target-band pilots. We propose a foundation model that represents every band in a common power-angle-delay spectrum and uses radio-frequency metadata as auxiliary conditioning for an encoder-decoder. The model uses pilot-guided cross-attention to fuse source-band structure with target-band pilot, allowing one model to handle heterogeneous band pairs. It is trained by pilot-densification pretraining followed by supervised cross-band fine-tuning. On ray-tracing, 3GPP, and DeepMIMO datasets, the model lowers average normalized mean-square error by 6.1 dB over the state-of-the-art pair-specific baseline. It also transfers to two unseen pairs without paired fine-tuning, achieving 7.5 and 7.1 dB gains over fully supervised baselines, and remains effective when target pilots are noisy.

We introduce trajectory-regularized stochastic optimal control (TRSOC), which augments standard stochastic optimal control (SOC) with a Kullback–Leibler (KL) divergence between controlled and reference trajectory distributions. Using Girsanov's theorem, the trajectory KL reduces to a quadratic drift mismatch penalty, yielding a modified running cost that preserves the dynamic programming (DP) structure. We derive the corresponding Hamilton–Jacobi–Bellman (HJB) equation and characterize the optimal policy. In the linear-quadratic (LQ) setting, the formulation admits a closed-form solution with an augmented control cost. Experiments show that the regularization parameter induces a trade-off between performance-driven and reference-preserving behavior, including cases with reference dynamics learned from offline data.

A body area network is a radio communication protocol for short-range, low-power, and highly reliable wireless communication for use on the surface, inside, or in the peripheral proximity of the human body. Combined with various biomedical sensors, BANs enable realtime collection and monitoring of physiological signals. Therefore, it is regarded as an important technology for the treatment and prevention of chronic diseases, and health monitoring of the elderly. The IEEE 802 LAN/MAN Standards Committee approved Task Group TG15.6 in December 2007. As a result of more than four years of effort, in February 2012, TG15.6 published the first international standard for BANs, IEEE Std 802.15.6. Throughout the development of this standard, ample collaboration between the standardization group and the research community was required. In particular, understanding the radio propagation mechanisms for BANs demanded the most research effort. Technical challenges were magnified for the case of implant communication because of the impossibility of conducting in-body measurements with human subjects. Therefore, research in this field had to make use of intricate computer simulations. This article outlines some of the research that has been done to obtain accurate propagation models supporting the standardization of implant communication in BANs. Current research to enhance the channel models of IEEE Std 802.15.6 through the use of ultra wideband signals for implantable devices along with physical measurements in animals is also presented.

Reconfigurable intelligent surfaces (RISs) provide a flexible means of engineering wireless propagation channels through configurable multiport scattering networks. Among their architectures, group-connected beyond-diagonal RISs (BD-RISs) offer a practical tradeoff between scattering flexibility and hardware complexity by partitioning the surface into inter-connected groups. In reciprocal and lossless implementations, each group scattering block must be symmetric and unitary. In this paper, we investigate sum-rate maximization for reciprocal group-connected BD-RIS-assisted multiple-input multiple-output (MIMO) systems under such structural constraints. In particular, we build on a recent result where the feasible space determined by the symmetry and unitary constraints was characterized as a Riemannian manifold, extending it into a product manifold of symmetric-unitary block manifolds, thus expanding the approach to manifold phase optimization in the group-connected setting. As a consequence, the proposed group-extended framework performs tangent-space projections and Takagi retractions block-wise, while preserving the coupling among all groups through the common MIMO equivalent channel. The formulation naturally includes single-connected and fully connected reciprocal BD-RIS architectures as special cases and enables a flexible performance-complexity tradeoff through the group size.

Emerging edge intelligence systems increasingly rely on dense deployments of always-on sensors that must convey task-relevant information to a remote model under tight energy and spectral budgets. The deployment of event-driven neuromorphic sensing paired with spiking neural networks (SNNs) is attractive in this regime because it produces dynamically sparse representations, so that energy is spent on communication and computation only when informative events occur. Prior multiple-access protocols for remote inference using neuromorphic sensing and computing targeted collaborative settings, in which the server fuses information from all devices into a single decision. This paper instead addresses parallel remote inference, in which each device observes a distinct input, and requires its own classification decision. We propose NOMA-NC, a non-orthogonal multiple-access (NOMA) neuromorphic communication (NC) protocol built on the vector symbolic architecture (VSA) framework. In NOMA-NC, each device binds its sparse spike feature map with a device-specific permutation key, and all devices in a group transmit concurrently so that the over-the-air superposition directly realizes the VSA bundling operation. A shared decoding SNN, together with lightweight per-device learned unbinding, recovers all decisions in a single inference pass. Experiments on the N-MNIST and DVS128 Gesture datasets show that NOMA-NC yields goodput gains and savings in terms of receiver computing energy that are sub-proportional to the number of simultaneously active devices, without increasing the per-device transmission energy.

While LC technology facilitates the realization of energy-efficient and scalable RISs, their phase shift response is inherently temperature-dependent. Neglecting this thermal dependency can lead to performance degradation, which is particularly detrimental in secure wireless systems where phase-shift inaccuracies may result in unintended information leakage. To address this challenge, we investigate secure communication in LC-RIS-aided systems and develop a temperature-adaptive phase-shift design. Beyond thermal sensitivity, the massive number of elements at mmWave frequencies is required to compensate for high path loss. This large-scale deployment of LC-RISs can lead to significant overhead challenges due to the acquisition of CSI. To ensure practical feasibility, this work proposes a phase-shift design that does not rely on the full CSI; instead, it employs only the possible locations of legitimate users and potential eavesdroppers. By illuminating a spatial zone rather than a single target location, the proposed temperature-adaptive algorithm enhances robustness against both thermally induced phase errors and positioning inaccuracies. To solve the resulting optimization problem, we present an SDP-based approach to serve as a high-performance benchmark, as well as a low-complexity heuristic method. The latter demonstrates superior scalability as the number of RIS elements increases, which makes it highly effective for deploying extremely large surfaces in dynamic, real-time environments. Based on this scalable framework, we further design a temperature-robust algorithm that maintains high security without requiring real-time temperature data. Extensive simulation results confirm that our temperature-adaptive and temperature-robust approaches yield a superior secrecy rate compared to conventional designs that neglect temperature impacts.

This letter exploits near-field (NF) spherical wavefronts to derive geometry-dependent physical signatures for backscatter device (BD) authentication in extremely large-scale multiple-input multiple-output (XL-MIMO) backscatter communication systems. We propose a binary element-wise delay shift keying (EW-DSK) modulation scheme based on a multi-antenna BD architecture. The proposed scheme introduces a deterministic, bit-dependent delay that shifts the power delay profile (PDP) while preserving the aperture-dependent curvature induced by NF propagation. Using the dominant PDP component extracted across the receive antennas, we develop a compact delay-curvature signature extractor and a corresponding authentication test to mitigate BD impersonation attacks. Simulation results evaluate the proposed scheme in terms of curvature extraction accuracy, authentication accuracy, and spoofing success probability under different spoofer distances, signal-to-noise ratios, and effective bandwidths.

This letter proposes an innovation-domain decision-directed phase tracking (ID-DDPT) architecture for coherent detection over Rayleigh fading channels with temporally correlated phase evolution and Wiener phase noise. By reformulating phase tracking into the innovation domain, replacing the unbounded cumulative phase with its stationary increments, the proposed method converts a non-stationary estimation problem into a stable low-complexity filtering problem. A closed-form expression for the steady-state residual phase error variance is derived under the locked-regime assumption, along with an analytical optimal smoothing factor. Modeling the residual phase distortion as an effective signal-to-noise ratio (SNR) attenuation yields a tractable bit error rate (BER) approximation for BPSK over Rayleigh fading. A first-order error-propagation analysis further characterizes the impact of decision errors and provides insight into the onset of cycle slips. Simulation results demonstrate that ID-DDPT outperforms DBPSK and a complexity-equivalent scalar Kalman tracker (SKT), achieving near-coherent performance with $\mathcal{O}(1)$ per-symbol complexity and minimal pilot overhead.

Terahertz (THz) communication systems for sixth-generation (6G) networks are severely impaired by Wiener phase noise (WPN), whose innovation variance at sub-THz carriers is substantially larger than in millimeter-wave 5G systems. Conventional common-phase-error (CPE) compensation applies a single phase rotation per OFDM symbol and becomes inadequate when the phase trajectory varies significantly within the symbol duration. This letter proposes continuous phase trajectory reconstruction (CPTR), a closed-form intra-symbol phase noise tracking method that reconstructs the sample-level phase trajectory from pilot observations via least-squares polynomial fitting with $\mathcal{O}(N_p+N)$ complexity. We characterize the polynomial approximation error under WPN and derive the Cramér–Rao bound (CRB) for polynomial phase coefficient estimation, showing that CPTR is minimum-variance unbiased within the polynomial surrogate model. Simulations at 300~GHz with N~=~1024 and 16-QAM show that CPTR remains within 0.2~dB of the CRB across SNR~=~10–45~dB while achieving significantly lower complexity than Kalman-based tracking and substantial BER gains over CPE, linear interpolation, and cubic spline methods.

This letter proposes the Predictive Lightweight Multi-Agent Reinforcement Learning (PL-MARL) framework to ensure resilient coverage in bandwidth-constrained UAV swarms. To counter coordination collapse caused by sparse signaling and information aging, we introduce a Kinematic-Aware Inference Engine that proactively reconstructs neighbor trajectories via physical priors. This approach enables an efficient computation-for-communication trade-off, decoupling structural resilience from signaling frequency. Simulations confirm that PL-MARL maintains superior coverage and mission continuity under extreme signaling scarcity and node failure. Our results validate proactive inference as a scalable, low-latency solution for robust aerial coordination, effectively minimizing control overhead to preserve spectrum for payload services while ensuring resilience against interference.

Lightweight projectors are an established way to connect pre-trained speech encoders with large language models (LLMs), mapping acoustic features into token-level embeddings for tasks like ASR and spoken question answering. Existing systems, however, typically only support a few languages and are often limited to English. We introduce MEUSLI, the first open-science multilingual projector family that links a Whisper encoder with open-source multilingual LLMs, enabling fully open-source end-to-end ASR in 28 European languages. MEUSLI extends prior monolingual pipelines, delivering strong results across high- and low-resource languages. Using proper continual leaning techniques, MEUSLI can be easily extended to other languages not seen in training. We further demonstrate that the MEUSLI projector can be leveraged beyond ASR, enabling multilingual speech translation and topic identification with only a few hours of task specific supervision per language. Overall, MEUSLI provides a solid foundation for multilingual speech understanding tasks, supporting scalable and inclu- sive open-source SpeechLLM

The trade-off between noise averaging and temporal resolution fundamentally limits conventional zero-crossing frequency estimators under dynamic and noisy conditions. This paper presents an overlapped sampling-intervals zero-crossing fitting method (OS-ZFM), which introduces a structured overlapping regression framework that decouples noise averaging from temporal update rate. The method adopts a deterministic closed-form formulation, enabling a unified bias-variance analysis to characterize the statistical behavior of the estimator and clarify the role of structured data reuse. Numerical results under intensity and background noise at a signal-to-noise ratio (SNR) of 10 dB show that OS-ZFM reduces median estimation error by more than 60% compared to conventional zero-crossing fitting methods at the same temporal resolution. It further achieves up to 90% reduction relative to basic zero-crossing detection and consistently yields lower estimation errors than Hilbert-transform-based estimators. Experimental validation utilizing impact-induced transient motion measured by laser Doppler interferometry demonstrates that OS-ZFM reconstructs smooth and physically consistent trajectories with improved temporal fidelity. Owing to its low computational complexity and deterministic formulation, the proposed method enables accurate real-time frequency and acceleration tracking in resource-constrained measurement systems.

Single-pixel sensing encodes a scene as a short sequence of coded measurements, and image-free methods infer the task directly from that sequence. We show that removing image reconstruction relocates the central design problem to the lift: how 1D measurements become a 2D task representation. We organize this choice as a lift spectrum from a fixed-physics inverse, through a learned static projection, to content-adaptive retrieval. These are not interchangeable forms of reconstruction: the fixed-physics route reconstructs an image consumed at inference, whereas our spatiotemporal soft-fusion (STSF) network lifts measurements directly into task features, and task-prioritized loss scheduling (TPLS) uses a separate learned reconstruction branch only as scheduled training supervision. A probe-selected recurrent encoder and a parameter-matched lift ablation identify the STSF design. In simulation, STSF+TPLS exceeds the prior image-free baseline on three datasets at 3.13% sampling (+3.2 to +9.9 pp foreground mIoU) and remains competitive down to 0.39%. The strongest clean-trained reconstruct-then-segment baseline wins without measurement noise, but measurement noise reverses the ranking: the reconstructed task input carries a 20-70x larger normalized relative perturbation than the measurements themselves. Stressed to failure, the three lift regions exhibit distinct dominant signatures–collapse, imprinting, and coarsening. STSF+TPLS transfers without fine-tuning to a real single-pixel bench, where the reversal reappears as a proof of concept; inference takes about 14 ms per mask on an RTX 4090. Within the tested fixed-acquisition regime, measurement-to-space adaptivity therefore organizes both the clean-to-noisy operating envelope and the failure a system encounters. Code and pretrained weights: https://github.com/Hanyuyuan6/STSF-TPLS.

This paper addresses the modeling and synthesis of constrained robotic system behaviors using Petri nets (PNs). It investigates how to construct models in which all observable system states satisfy given logical constraints while remaining consistent with executable transition semantics. To answer this, we introduce the Hyper Petri Net (HyPN) approach, which synthesizes Petri nets from Boolean specifications while explicitly distinguishing between observable markings and underlying Petri net execution. The proposed method introduces an explicit execution semantics over observable states, induced by admissible (atomic) firing sequences, ensuring by construction that all observable markings satisfy the constraints and revealing a fundamental mismatch between logical feasibility and executable behavior. This is demonstrated in two scenarios inspired by a lunar rover system. These results are particularly relevant for the design of robotic and autonomous systems, as they provide a structured way to ensure correct system configurations while explicitly accounting for execution constraints. The proposed framework further suggests new research directions in execution abstraction, admissible transition systems, and policy selection for navigating between constraint-satisfying states.

Purpose: In Spiral Magnetic Resonance Spectroscopic Imaging (MRSI), achieving suitable spatio-temporal resolutions requires interleaving. Sampling inconsistencies occurring between temporally interleaved signals, -whether due to time-interleaved ADCs or phase/frequency mismatches in the excitation and demodulation stages-, result in artifacts in the recombined data. This paper proposes a mathematical description of these unavoidable artifacts and a postprocessing solution to circumvent the issues and mitigate them. Methods: The study proposed a detailed description and explanation of these artifacts using data acquired with MRSI interleaved sequence with gradients off, as well as standard interleaved FID sequence. The random signal mismatch between the interleaves was described as additional artifactual signals $\psi$(t) at each temporal interleave. A model was further proposed to match the acquired signal in relation to the ideal, artifact-free signal. Considering M interleaves, a correction method was derived relying on estimating 4 parameters in the $\psi$(t) model for each interleave by minimizing a multivariable cost function. This correction method was evaluated on spectra acquired from phantoms and in vivo acquisitions obtained in healthy volunteers. Results: The proposed correction method significantly reduced artifacts by a factor of 4 in the phantom and 1.8 in the healthy volunteers. This correction ensured accurate spectral quantification in regions where artifacts overlapped with the content of interest, such as lipids in subcutaneous fat. Additionally, it was demonstrated that appropriately selecting the number of temporal interleaves can shift the artifact away from the frequency of interest, leveraging the periodic and limited frequency span of the observed artifacts. Conclusion: In spiral-MRSI, temporal interleaves, while enhancing spectral bandwidth, introduced spurious content characterized by distinctive resonance frequencies and nonreproducible amplitudes and phases. The study proposes two solutions: manipulating the number of interleaves to control artifact frequency localization or using a retrospective correction method to approximate and attenuate artifacts.

Meta-learning for domain generalization (MLDG) improves out-of-distribution speech deepfake detection over empirical risk minimization (ERM) when both objectives train low-rank adapters on the same frozen self-supervised speech model. Because the architecture and adapter capacity are held fixed, this gap points to differences in how the training objective shapes the adapter, yet the field characterizes objectives through error rates rather than through the geometry of the solution they reach. We introduce a descriptive diagnostic for this question: holding architecture, rank, data, and seeds fixed and varying only the objective, we use the empirical Fisher on the finished adapter to compare the geometry that ERM and MLDG leave behind. We characterize each adapter with effective-rank diagnostics that separate where the adapter changes from where those changes matter to the loss, resolved by projection and by depth. Applied to ERM and MLDG, the diagnostic shows that the objective does not reshape all adapter projections alike: the loss-relevant update concentrates in the query and key projections while becoming more distributed in the output projection, consistently across six corpora and most strongly in the upper layers. The same contrast appears in the merged update independently of the low-rank factorization, indicating that it reflects the geometry of the effective update rather than the parameterization. These results show that the gap between ERM and MLDG is not only a difference in error rate, but a difference in how loss-relevant capacity is organized inside the adapter, and that loss-aware adapter geometry is a way to see it.

The low-rank Cholesky factor alternating direction implicit (LRCF-ADI) iteration method is an effective and efficient approach for computing low-rank solutions to large-scale Lyapunov equations in the form \(P\approx ZZ^\top\). This form is useful for balanced truncation, as the square-root algorithm requires computing the controllability and observability Gramians in this form. The standard LRCF-ADI method requires all ADI shifts to have negative real parts, which can be restrictive for applications like frequency-limited and data-driven balanced truncation, where purely imaginary ADI shifts are a more suitable choice. This paper proposes a new LRCF-ADI method where the ADI shifts can be located anywhere in the complex plane, including on the imaginary axis. The proposed generalized LRCF-ADI algorithm reduces to the standard LRCF-ADI algorithm as a special case. The new method is also extended to solve frequency-limited Lyapunov equations, time-limited Lyapunov equations, and Riccati equations. Approximations of matrix logarithm and matrix exponential products using the proposed method are also discussed. LRCF-ADI-based reduced models for balanced truncation can be constructed non-intrusively from transfer function samples at the mirror images of the ADI shifts, without accessing the state-space realization of the original model. Since the standard LRCF-ADI method requires all ADI shifts to have negative real parts, its non-intrusive implementation requires samples in the right half of the complex plane, which cannot be measured in an experimental setting. However, the ADI shifts in the proposed method can lie on the imaginary axis. Exploiting this property, we also propose a data-driven low-rank balanced truncation algorithm that requires only transfer function samples on the imaginary axis, which can be measured experimentally.

Establishing and maintaining 5G mmWave vehicular connectivity poses a challenge due to high user mobility, requiring the design of robust and efficient beam switching procedures. Unlike reactive beam switching based on channel state information (CSI) feedback received from vehicular users, proactive beam switching exploits CSI prediction to prepare in advance for upcoming beam switching decisions. In this paper, we develop a framework for autonomous and self-trainable CSI prediction for mmWave vehicular users. In the proposed framework, base stations (gNBs) collect and label data sets to train a CSI prediction model both independently and using federated learning (FL). The data set combines data extracted from the CSI feedback and cellular vehicle-to-everything (C-V2X) cooperative awareness messages (CAMs) of surrounding vehicles. The framework is placed in the context of machine learning and artificial intelligence (ML/AI)-based Open RAN (O-RAN) applications (rApps and xApps) fed by realistic real-world mobility and CSI data from the DeepMIMO simulator. Detailed evaluation results demonstrate feasibility, accuracy, and flexibility of the proposed CSI prediction framework

Unmanned aerial vehicle (UAV) communication is expected to support a wide range of low-altitude applications in 6G mobile networks. However, traditional statistical channel models provide limited accuracy in specific environments, while deterministic methods such as ray tracing usually rely on accurate three-dimensional environment models and involve high computational complexity. Existing multimodal channel prediction approaches mainly focus on large-scale metrics such as path loss, and remain insufficient for modeling small-scale parameters. To address these limitations, this paper proposes PanoLAMP, a Panoramic perception and vision-language model-based Low-Altitude Multipath Prediction framework. It adopts a pretrained vision-language model as the backbone and captures the propagation environment features through panoramic RGB-D observations collected at both the transmitter and receiver to predict the delay, power, azimuth angle, and zenith angle offset relative to the line-of-sight path. Experiments are conducted on a synthetic dataset containing 18,949 UAV-vehicle links across seven UAV altitudes. Experimental results show that the proposed method consistently outperforms representative baselines in both multipath parameters and statistical metrics, and demonstrates stronger generalization across different flight heights.

Efficient Non-Access Stratum (NAS) timer configuration is critical for reliable and energy-efficient Fifth Generation (5G) registration in Non-Terrestrial Network (NTN)-Terrestrial Network (TN) integrated systems, where Low Earth Orbit (LEO) satellite access introduces large registration bursts, heterogeneous propagation paths, and multi-hop satellite routing. Existing 3GPP NAS timers use fixed values, while prior closed-form timer models compute a global timer under network-level assumptions; both fail to capture user equipment (UE)-level differences in propagation delay, Access and Mobility Management Function (AMF) arrival position, and path reliability. In this paper, we propose a location-aware, UE-specific NAS timer optimization method for LEO NTN-TN integrated networks. The proposed method models the path-delay component using service-link geometry, ground-station distance, and Inter-Satellite Link (ISL) hop count, and adapts the endpoint-delay component according to each UE's expected AMF queue exposure and path reliability. Simulation results show that our method reduces registration latency, UE energy consumption, and avoidable registration attempts compared with fixed and global timer configurations, especially when timer over-provisioning causes unnecessary waiting.

Mixed-integer programming (MIP) precludes simultaneous charging and discharging in power dispatch of energy storage system (ESS), but is computationally intensive. Relaxation of this complementarity constraint improves computational efficiency, but may incur an impractical control SCD strategy. To bridge this gap, this letter proposes a necessary and sufficient condition for the exactness of such relaxation. We explicitly derive the relaxation gap and reveal that SCD does not necessarily imply a distinguished optimal value from the MIP model. In such cases, a feasible solution without SCD can be readily recovered without compromising the optimal value. Furthermore, a two-stage relaxation-and-recovery algorithm is proposed to improve the computational efficiency of ESS power dispatch optimization.

This paper proposes a novel baseband-free wideband MU-MIMO OFDM transmitter architecture enabled by two cascaded stacked intelligent metasurfaces (SIMs). Unlike conventional wireless transmitters, the proposed design shifts symbol loading, MU-MIMO precoding, and OFDM modulation from digital baseband to the wave domain. Specifically, in the first SIM (SIM$_1$), a programmable symbol-loading interface first loads one stacked virtual-subcarrier coefficient vector onto a common monochromatic carrier, while the remaining layers perform channel-adaptive MU-MIMO precoding from the data streams to the port-subcarrier domain. Then, the second SIM (SIM$_2$) realizes offline-configured wave-domain OFDM modulation by implementing the inverse discrete Fourier transform (IDFT) and cyclic-prefix (CP) insertion operator directly in the wave domain. Based on this architecture, we develop a wave-domain system model that explicitly characterizes the virtual-to-physical transition from block-level coefficients to radiated OFDM subcarriers. The designs of SIM$_1$ and SIM$_2$ are formulated as two operator-fitting problems, respectively targeting a block-diagonal wideband precoder and the ideal OFDM modulation operator. Under practical discrete phase constraints, we develop a quantization-aware training-based gradient descent (QAT-GD) framework for both SIM stages. Numerical results verify the effectiveness of the proposed optimization, the correctness of the synthesized wave-domain functionalities, and the strong end-to-end MU-MIMO OFDM performance of the resulting architecture.