Papers

We consider a multi-receiver status update system in which a transmitter monitors a finite-state semi-Markov source and decides whether to stay idle, unicast an update, or broadcast a common update. We formulate a risk-aware scheduling problem that minimizes the long-term average sum of the average Age of Incorrect Information (AoII), average risk ratio, and transmission cost. The risk state is defined by whether the AoII exceeds a prescribed threshold. We solve the problem using model-based and model-free policies and compare them with two baselines. Numerical results show that the proposed policies outperform the baselines, exploit both unicast and broadcast transmissions, and capture the effect of the dwell-time law on scheduling performance.

To meet the demands of 6G wireless systems operating in high-mobility scenarios, this paper presents a design of a random multiplexing (RM) communication system that is both storage-efficient and highly reliable. In principle, RM with cross-domain memory approximate message passing (CD-MAMP) can achieve replica maximum a posteriori (MAP)-optimal performance by constructing a fully dense equivalent channel matrix. However, its practical implementation is hindered by the large storage overhead of conventional interleavers and by performance degradation in severely ill-conditioned channels, which existing related work (focusing on interleaving and transform designs) fails to address simultaneously. To overcome these issues, we develop a storage-efficient and highly reliable system that integrates RM with CD-MAMP, referred to as RM-MAMP. Specifically, we propose a Logistic chaotic mapping interleaver with a quantitative parameter-selection criterion, and a dual-stage high-order permutation polynomial interleaver, both of which achieve nearly identical bit-error-rate (BER) as fully random interleavers while reducing the interleaver storage from O(N) to O(1) and significantly lowering interleaver signaling overhead. We further propose a highly reliable interleaved transform framework, comprising an interleaved phase perturbation transform and a multi-layer interleaved coupled transform, to enhance the incoherence and diversity of the equivalent channel matrix. Simulation results show that the proposed storage-efficient interleavers maintain BER performance comparable to fully random interleavers, while the highly reliable transforms provide over 4 dB gain in severely time-varying channels, confirming the dual benefits of reduced storage overhead and improved robustness for the enhanced RM-MAMP system.

This manuscript addresses a hierarchical control system designed to suppress traffic congestion. The lower-layered controllers, implemented in each controlled vehicle, monitor microscopic vehicle behaviors and assist human drivers to ensure sufficient spacing for following vehicles. This spacing logic is designed based on the Control Barrier Function. Meanwhile, the upper-layered controller monitors the macroscopic traffic flow and activates the necessary lower-layered controllers, using a data-driven approach for the activation logic design. Furthermore, the effectiveness of the proposed control system is evaluated in a traffic flow simulation environment constructed using real-world traffic data.

We investigate the impact of power amplifier (PA) nonlinearities on the sensing performance of affine filter bank modulation (AFBM). While AFBM offers several advantageous properties for integrated sensing and communications (ISAC) - including reduced out-of-band emission (OOBE), low peak-to-average power ratio (PAPR), and natural robustness to doubly-dispersive (DD) channel effects - mitigating waveform distortion typically requires highly linear PAs. This creates a fundamental contradiction with ISAC applications, which demand high transmit power for reliable sensing. Our analytical results reveal that the structure of the effective AFBM modulation matrix dictates how distortion propagates within the ambiguity function (AF). Furthermore, simulations demonstrate that both the AF and the overall sensing performance of AFBM remain remarkably insensitive to such nonlinearities. These findings highlight the robustness of AFBM, making it a highly viable candidate for practical ISAC deployments constrained by hardware impairments.

This paper investigates an analytical Koopman-based nonlinear model predictive control (K-NMPC) approach for tracking control of virtually coupled train systems. A nonlinear train movement model incorporating train dynamics, speed and control input limits, passenger comfort constraints, and collision avoidance is systematically lifted into a finite-dimensional Koopman space through closed-form observable functions. After freezing the affine parameter-varying lifted predictor along the shifted predicted trajectory, the online optimal control problem is solved as a quadratic program that can be solved efficiently. The proposed KNMPC is benchmarked against a time-discrete NMPC scheme, demonstrating comparable control performance with significantly reduced online computation time and strong potential for real-time implementation in practical virtually coupled train control systems.

Multi-channel mixed-SNR training improves out-of-distribution (OOD) generalisation of deep learning channel estimators for IEEE 802.11p vehicular communications, yet the internal mechanism responsible for this remains unexplained. This work presents REACH (Relevance-based Explanation and Architectural Compression for cHannel estimators), a gradient-based interpretability framework that operates at two levels. Input-level attribution identifies a subset of time-frequency features consistently relevant across all evaluated channel conditions, enabling input dimensionality reduction with minimal performance loss. Filter-level attribution reveals a near-universal internal representation, providing a representational account of the observed OOD generalisation. Guided by the resulting filter taxonomy, relevance-guided architecture compression substantially reduces both the number of parameters and the number of floating-point operations (FLOPs) with sub-1 dB normalised mean square error (NMSE) degradation, and OOD generalisation degrades more slowly than within-distribution accuracy under increasing compression.

This paper presents a novel data-free and training-free compression approach for speech foundation models using channelwise clustering via k-means. More fine-grained, mixed sparsity pruning by layer-level varying number of parameter clusters is also explored. Experiments conducted on the LibriSpeech dataset suggest that when operating with pruning sparsity of 50% on HuBERT-large, consistent WER reductions of 27.73%/18.61% absolute (34.37%/21.91% relative) over the magnitude-based pruning were obtained on the test-clean and test-other subsets before fine-tuning and 0.19%/0.79% absolute (3.36%/4.62% relative) after fine-tuning with only 3 epochs. Similar WER reductions of 2.86%/5.02% absolute (59.21%/55.29% relative) were observed against magnitudebased pruning on Whisper-large-v3 at 10% sparsity, all with no significant WER increase relative to the uncompressed baseline.

Due to high power consumption and hardware costs of fully digital arrays, hybrid beamformers are often considered as a more economic alternative. Furthermore, using high resolution analog to digital converters (ADCs) can also have prohibitive power consumption, which leads to lower resolution converters being considered for radio frequency (RF) front end design. The finite quantization resolution as well as the nonlinearities caused by the power amplifiers (PAs) and low noise amplifiers (LNAs) can have a substantial impact on system performance. While widely studied for communications, the impact of hardware impairments on sensing performance is considerably less explored. In this work, we study the interplay between hybrid beamforming architectures, hardware impairments, and sensing and communications performance. Additionally, we define the concept of double-isotropy for pilot-combiner pairs, formalizing the notion of a perfectly energy-fair beam sweep. The multiple start (MS) space alternating generalized expectation maximization algorithm (SAGE) is also introduced, aimed at addressing the optimization issues arising from parametric channel estimation (PCE) in hybrid beamformed systems. We then provide a set of numerical results assessing the impacts of beamformer architecture and ADC resolution on PCE, sensing, and communications performance. The results show that medium resolution ADCs lead to the most power efficient configurations, with the best tradeoff between power consumption and performance for the majority of beamforming architectures. Additionally, fully digital beamforming architectures with high resolution converters can often be substituted for a hybrid beamformer setup with medium resolution converters without significant performance loss at a lower power consumption and overall hardware cost.

Uncertainty in standalone microgrid operation usually originates from mismatches between power references and forecasts. These deviations are compensated by grid-forming controlled units, which distribute the required power contribution based on their droop gains. To introduce an additional degree of flexibility, it is possible to treat droop gains as decision variables to redistribute active-power contributions according to system-level objectives. However, directly applying updated droop gain references from a supervisory layer to the primary controllers can introduce power and frequency transients. This paper investigates transition mechanisms for applying scheduled active-power droop gain changes during operation. Hard switching, rate-limited transition, first-order IIR low-pass filtering, and cubic as well as quintic S-curve transitions are compared experimentally on two parallel 15 kW grid-forming inverter units. The results show that shaping the droop gain trajectory significantly reduces transient deviations compared to hard switching. In the considered case study, the S-curve transitions provide the strongest transient mitigation, reducing the active-power overshoot from 632.7 W to approximately 115 W and limiting the frequency overshoot to about 0.003 Hz.

Multi-talker conversational automatic speech recognition data are often used to train speaker diarization models. Because such data prioritize semantic continuity, pauses and boundary margins are included within speech segments, resulting in loose annotations. Models trained on such data tend to internalize mechanisms that reproduce this looseness, although tight speech intervals are sometimes preferable for downstream applications. In this paper, we address the novel task of enabling models to produce tight predictions using loose labels. Our method generates tighter pseudo labels using causal and anticausal models, which are inherently incapable of learning loosening behavior. We further propose a co-training scheme that iteratively tightens labels and updates both models for more progressive refinement. Experimental results show that the proposed method recovers about 70 % of the tightening effect achieved by ideal tight-label training and improves downstream performance.

Distilling a large speech foundation model (SFM) into an efficient student model has been successfully applied to low-resource environments. Although distillation reduces inference latency, it requires an additional student model training. However, the training efficiency of SFM distillation remains underexplored. In this work, we explore training acceleration of SFM distillation to speed up model deployment. We examine the potential of stacking, in which the model depth is progressively increased through training until the target model depth is reached. While existing stacking methods improve training speed, they suffer from performance degradation. To handle this limitation, we propose interleaved stacking, a novel stacking method that consistently preserves layer position throughout the stacking process. This property is particularly critical in SFMs, in which each layer encodes distinct layer-specific knowledge. We validate the effectiveness of the proposed method on SUPERB.

Power leaking directly from transmitting into receiving radio-frequency chains is a key challenge in the realization of monostatic sensing applications with multi-antenna communication front-ends, to which a promising solution is digitally precoding transmitted signals for improved leakage suppression. While digital transmit precodings perform well in theory, real-world deployments typically exhibit severely degraded leakage suppression. This work investigates quantization noise as a primary factor limiting the performance of such precoding schemes. A closed-form solution predicting the impact of quantization noise on the performance of arbitrary digital joint leakage estimation and leakage suppression precodings is derived, numerically analyzed, and validated in a hardware testbed.

The rapid adoption of electric vehicles (EVs) can cause severe voltage drops and line current overloads in distribution networks, creating an urgent need for scalable expansion planning methods. This paper proposes a computationally efficient violation-informed spatio-temporal adaptive targeting (STAT) framework for EV-driven distribution system expansion planning. The framework first identifies potential voltage and current violations through a violation analysis model, and then mitigates them through a joint optimal expansion planning model that co-optimizes investment decisions for line reconductoring, shunt capacitors, and battery energy storage systems. To reduce computational burden, the proposed STAT-temporal criticality assessment (STAT-TCA) method extracts primitive stress events from annual operating data, derives an initial set of candidate planning horizons from signature-consistent segments, and selects a final transferable critical horizon set through cross-horizon validation based on optimization feasibility and cost. Meanwhile, the proposed STAT-adaptive spatial targeting (STAT-AST) method constructs device-specific spatial features for BESS and SC siting to retain compact yet high-impact candidate bus sets. Case studies on 33-bus and 240-bus distribution systems demonstrate that the proposed STAT framework can substantially reduce the temporal and spatial planning dimensions while preserving planning fidelity. Full-year validation further confirms that the resulting investment plans can eliminate EV-induced voltage and thermal violations while maintaining feasible BESS operations.

Learning-based speech compression has achieved promising low-bitrate performance, but many neural speech codecs still describe quantized latents with preset-rate discrete symbols or apply entropy coding only after symbol generation. Such designs decouple representation learning from probability modeling, limiting their ability to exploit the non-uniform usage and temporal dependencies of learned speech latents. In this paper, we benchmark neural speech compression from a rate–distortion perspective and further investigate entropy-constrained coding for low-bitrate speech compression. We first formulate a unified learning-based speech coding pipeline and provide a benchmark-style analysis of recent neural speech codecs, showing that explicit probability modeling remains underexplored in learned speech compression. We then propose ECC, an Entropy-Constrained Codec that combines scalar quantization with a learned entropy model. ECC integrates hyperprior-based side information, channel-wise context modeling, latent residual prediction, and lightweight temporal modeling to estimate latent likelihoods for rate estimation during training and arithmetic coding during inference. To further improve low-bitrate efficiency, ECC introduces entropy skip, which omits highly predictable residual symbols using decoder-available scale estimates without transmitting additional skip masks. Extensive experiments show that ECC achieves a favorable low-bitrate rate–distortion trade-off over conventional and neural codec baselines, reducing BD-rate by 39.9% on ViSQOL and 76.3% on PESQ on average over two widely-used test sets. Ablation and diagnostic studies further validate the effectiveness of entropy modeling. Project Page: https://avery-xu.github.io/ECC-demo/

The Frequency Range 3 (FR3) band is attracting increasing attention due to limited lower-frequency spectrum and growing mobile communication demand. This study experimentally investigates channel characteristics in Urban Macro (UMa) scenarios at 8 GHz and 15 GHz using a large-scale MIMO platform with time-division multiplexing (TDM). Key parameters, including root mean square (RMS) delay spread (DS) and angular spread (AS), were extracted and compared with 3rd Generation Partnership Project (3GPP) TR 38.901. Results reveal clear frequency-dependent behaviors: RMS delay spread remains nearly constant under line of sight (LOS) but decreases from 8 GHz to 15 GHz in non-line of sight (NLOS), indicating reduced multipath dispersion at higher frequencies. Both azimuthal spreads (including ASA and ASD) and elevation spreads (including ESA and ESD) exhibit a corresponding decrease with increasing frequency, demonstrating a consistent trend towards more directional propagation across all angular domains. Capacity analysis indicates that the 15 GHz channel slightly outperforms 8 GHz in both LOS and NLOS scenarios due to more concentrated multipath energy and larger dominant singular values. Higher frequencies exhibit greater directionality, whereas lower frequencies provide broader multipath distributions and more stable performance, offering valuable guidance for multi-band MIMO modeling and 6G system design.

In this paper, we consider a class of networked systems comprising an interconnected set of linear subsystems, disturbance inputs, and performance outputs. Using dissipativity theory, we first propose a model-based hierarchical control design strategy to ensure the closed-loop networked system is dissipative from its disturbance inputs to performance outputs. This involves designing local controllers for each subsystem to enforce local dissipativity guarantees, which are then exploited to co-design distributed global controllers and the interconnection topology to enforce global dissipativity guarantees while optimizing interconnection topology costs. The overall design process requires only solving a sequence of linear matrix inequality (LMI) problems, thereby retaining compositionality and decentralizability while avoiding non-convex, iterative design processes that are inefficient and centralized. This model-based hierarchical control design strategy assumes the knowledge of the subsystem dynamics, which may not hold in many real-world networked systems. Motivated by this, we also propose a data-driven hierarchical control design strategy that assumes only the availability of rich input-state-output trajectory data from the subsystems. The proposed data-driven design process assumes that the unknown disturbances affecting the subsystem dynamics are bounded by a quadratic matrix inequality (relaxing conventional bounds) and accounts for this by using the matrix S-lemma. Finally, the effectiveness of the proposed model-based and data-driven hierarchical control designs is illustrated for a networked system representing a DC microgrid, with the aim of enforcing robust (dissipative) voltage regulation and current sharing.

The Circular Electron Positron Collider (CEPC) distributes a reference clock distributed to 192 control nodes along its 100~km underground tunnel. The required synchronization precision is 30~ps (standard deviation). We present an enhanced White Rabbit (WR)-based clock synchronization system designed to meet this requirement. A noise-budget analysis of the standard WR slave loop identifies the analog actuation chain (DAC + VCXO + multiplier PLL) and restart-induced timing uncertainty as the dominant limitations. In our redesigned node, the DAC+VCXO chain is replaced by a Si5345A DSPLL clock generator with DCO-based phase control, removing the board-level analog tuning stage. GTX transceiver phase alignment and manual byte-alignment fixing reduce restart uncertainty from 88.8~ps to 12~ps peak-to-peak. For multi-node operation, we introduce a cascaded global-control architecture with PC-side PID auto-tuned by TD3 reinforcement learning, on-chip-temperature feed-forward calibrated to $-0.76\,\mathrm{ps}/^\circ\mathrm{C}$. The measured point-to-point synchronization precision is 3.38~ps over 1~m fiber and 3.92~ps over 50~km. In a 12-level cascade, the end-node precision reaches 6.66~ps at constant temperature and 7.30~ps under a 13$\,^\circ$C temperature swing. Synchronized-clock TIE jitter stays below 1~ps regardless of cascade depth. Restart uncertainty is 2.82~ps (std.\ dev.). A 4-level cascade operated stably for 25 hours of continuous monitoring. All measured metrics fall well within the CEPC 30~ps budget.

Large Language Models (LLMs) have rapidly emerged as tools of interest across engineering disciplines, and Process Systems Engineering (PSE) is no exception. This survey provides a systematic review of LLM applications in PSE, organizing the literature into seven categories: (1) process design and engineering, (2) molecular design and synthesis, (3) process modeling and simulation, (4) time-series forecasting, (5) optimization and scheduling, (6) process control, and (7) fault detection and diagnosis. For each category, we summarize the state of the art, identify common methodological approaches, and critically assess demonstrated capabilities versus aspirational claims. We find that LLMs show genuine promise for tasks involving natural language, including querying documentation, synthesizing unstructured knowledge, and enabling flexible human-machine interaction. However, applications requiring real-time execution, constraint satisfaction, or formal safety guarantees remain challenging. We conclude by identifying open problems and productive research directions for the PSE community.

Pixel antennas enable antenna coding, a technique that can provide more degrees of freedom in wave manipulation, to enhance wireless communications. However, acquiring full channel state information (CSI) at the transmitter incurs prohibitive overhead due to the unique hardware constraints from pixel antennas. This paper thus proposes a limited feedback multi-input multi-output (MIMO) system using pixel antennas, where the antenna coder and digital precoder are designed based on pre-defined codebooks and efficient index feedbacks. We first derive the optimal digital precoder under practical power constraints that provides insights on simplifying the joint codebook construction for antenna coder and digital precoder. We then develop a low-complexity offline codebook construction algorithm that enables subsequent codebook designs for the antenna coder and digital precoder. Simulation results demonstrate that the proposed scheme significantly outperforms unconstrained MIMO systems using conventional antennas with fixed configurations.

Evaluating generative spatial audio for First-Order Ambisonics (FOA) remains challenging due to a limited understanding of how metrics respond to changes in spatial parameters such as azimuth and elevation. We propose a framework to analyze metric sensitivity along continuous spatial trajectories, drawing on principles of sensitivity analysis in parametric sound synthesis. Using controlled FOA scenes with increasing scene complexity, we define three desiderata for metric behavior: Responsiveness, Smoothness, and Symmetry. We assess standard distribution-based and sample-based metrics, including Fréchet Audio Distance (FAD), intensity vectors, and acoustic maps. Our findings show that FAD using localization-specific embeddings and acoustic maps yield high Responsiveness and robust Smoothness and Symmetry across conditions, while intensity vectors degrade with increasing scene complexity. This is the first step towards investigating the sensitivity of metrics for generative spatial audio.

This paper presents a two-dimensional finite element method (FEM) solver for computing modal field distributions and dispersion characteristics of hollow metallic waveguides. To solve the waveguide problem, the source-free frequency-domain Maxwell equations are reduced to scalar Helmholtz eigenvalue formulations evaluated over the waveguide's transverse cross section. The computational method determines both transverse electric (TE) and transverse magnetic (TM) mode families by enforcing perfectly electrically conducting (PEC) boundary conditions. The framework is initially validated against analytical benchmarks using empty rectangular and circular waveguides, demonstrating high accuracy in computing cutoff wavenumbers, dispersion curves, and field maps for the first three unique modes. After validation, the solver is applied to analyze single-ridged and double-ridged waveguides. The numerical results demonstrate that introducing metallic ridges successfully redistributes the modal fields and significantly lowers the cutoff frequency of the dominant mode relative to empty rectangular guides. Ultimately, this work confirms that the generalized eigenvalue FEM formulation is a robust and adaptable tool for analyzing complex waveguide geometries where exact analytical solutions are unavailable.

The success of large-scale deep learning models in neuroscience is fundamentally constrained by severe data heterogeneity. Native fMRI data aggregated from diverse sources exhibit substantial variation in both spatial and temporal resolutions. Consequently, most existing frameworks rely on lengthy, rigid preprocessing pipelines that enforce uniformity across datasets. This practice introduces two critical limitations: (1) potential degradation of subject-specific anatomical information; (2) significant computational overhead, often requiring hours of processing per subject. Here, we propose FlexiBrain, a resolution-agnostic voxel-level encoding framework for native fMRI based on Mamba-JEPA. FlexiBrain defines patch sizes in real-world physical units and employs a dynamic patch resizing, thereby bypassing destructive spatial standardization while enabling direct ingestion of data in native space. We instantiate the framework using an efficient Mamba-JEPA backbone to model high-dimensional 4D fMRI signals. Across five diverse downstream neuroscience tasks, FlexiBrain consistently outperforms recent state-of-the-art methods, achieving gains of up to 12 percentage points without external data augmentation. Importantly, FlexiBrain functions as a seamless plug-in module, substantially reducing preprocessing costs and accelerating the development of robust voxel-level fMRI foundation models. Code is available at https://github.com/OneMore1/FlexiBrain.

Indoor localization from wireless measurements remains challenging in large-scale deployments due to substantial variation in building geometry, the set of detectable access points (APs), and the heterogeneity of received signals. Existing learning-based methods often perform well only in limited settings and degrade under environmental shifts, making robust anchor-free localization across diverse indoor environments notoriously difficult. In this paper, we present OmniLoc, an environment-interactive foundation model for anchor-free user equipment localization across diverse indoor environments. To the best of our knowledge, OmniLoc is the first foundation-model-based approach built directly on wireless measurements for this task. OmniLoc is built on three key designs. First, a unified input tokenization module converts heterogeneous wireless measurements into a common representation that is more amenable to learning. Second, a geometry-aware Transformer performs AP-aware feature extraction by emphasizing dominant APs while aggregating complementary evidence from supporting APs. Third, a geometry-aware location estimation module conditions regression on geometric embeddings to produce geometrically consistent location predictions. We evaluate OmniLoc on both a large-scale in-house dataset and a public benchmark dataset. Results show that OmniLoc significantly outperforms existing methods, consistently improves existing backbones when its design components are integrated, and demonstrates strong generalization in cross-environment evaluations.

In this paper, we study Mahalanobis-guided latent out-of-distribution (OOD) detection for test-time RL controller switching in nonlinear time-varying systems. RL controllers can quickly control high-dimensional systems within the training distribution, but their performance can degrade when time-varying dynamics produce unseen observations. We consider a combined ES–DRL controller, where RL provides fast in-distribution actions and bounded extremum seeking (ES) provides robust model-independent control under OOD operation. The key challenge is deciding when to switch. We train a variational autoencoder (VAE) on in-distribution beam-profile observations and use Mahalanobis distance in the VAE latent space to detect OOD beam profiles at test time. This OOD decision sets a binary switch that selects either the RL controller or the ES controller. We evaluate the approach in safety-critical particle accelerator control. In this setting, spatial magnet motion creates OOD beam profiles that were not seen during RL training. Visualization of the VAE latent space shows that the proposed method identifies this OOD scenario and provides an interpretable signal for switching between RL and ES in the combined controller.

This paper investigates coherent multiband orthogonal frequency division multiplexing (OFDM) sensing within an integrated sensing and communication (ISAC) framework. We consider an intra-band configuration in which two sensing subbands of equal width are allocated symmetrically within the same OFDM channel, while the central portion remains available for communication. We address the reconstruction of missing frequency-domain samples induced by the spectral gap and the suppression of the resulting grating lobes in the delay profile. To this end, we propose a low-complexity iterative reconstruction method consisting of an initial delay-domain equalization stage and an iterative apodization-based operator with data-consistency enforcement. Performance results for multi-target scenarios show that the proposed approach remains close to the full-band reference for moderate gap sizes and degrades only for larger gaps because of residual grating lobes. Compared with the compressed-sensing-based orthogonal matching pursuit (OMP) baseline, it exhibits a more favorable performance trend as the number of targets increases, especially in the practically relevant low-signal-to-noise ratio (SNR) regime, while offering a complexity scaling that is independent of the estimated number of targets.

The cumulative distribution transform (CDT) is a quantile-based transport representation that exactly linearizes one-dimensional translations of positive densities. We study how this structure behaves under additive perturbations and how it can be exploited for shift recovery. Under a local nondegeneracy condition, we derive a first-order expansion showing that additive noise in physical space induces a nonlocal perturbation in CDT space through the primitive of the noise, weighted by the reciprocal density. This yields an explicit description of transform-domain sensitivity and shows, in particular, that perturbations are amplified in low-density regions. When the physical-space perturbation is modeled as a centered Gaussian random field, the induced first-order CDT perturbation is again Gaussian, with an explicit covariance kernel. We then use this structure to study recovery in CDT coordinates. In the known-template setting, the transport shift is obtained by projection onto the constant mode, giving an explicit estimator together with exactness in the noiseless case and a stability bound under perturbations. In the unknown-template setting, multiple observations permit joint recovery of the shifts and a common template up to the natural constant-mode gauge, leading to a simple de-shift–and–average procedure. We also consider a signed-signal analogue based on the signed cumulative distribution transform (SCDT), where shifts are estimated numerically by feature matching and unknown templates are recovered by alternating alignment and averaging. Numerical experiments validate the perturbation analysis and illustrate effective recovery for both density-valued and signed signals.

Speech foundation models often struggle in low-resource domains due to domain mismatch and data scarcity. We propose Gumbel-BEARD, a domain adaptation framework that automates Whisper encoder layer selection via an end-to-end trainable hard Gumbel-Softmax selector. It enables self-supervised adaptation with a BEST-RQ objective that dynamically adapts to target acoustic characteristics without manual tuning. Experiments on the MyST child speech corpus demonstrate efficiency and scalability: with 10 h of labeled data for fine-tuning, our method matches a fully supervised baseline trained on the complete 133 h labeled set. We establish new state-of-the-art word error rates (WERs) of 8.21% using Whisper-medium on MyST and 11.06% using Whisper-small on the OGI Spontaneous dataset. Evaluation on CORAAL further confirms robustness to adult dialectal domain shifts, with up to 6% relative WER reduction, highlighting the generalizability of our approach to diverse low-resource conditions.

Large Audio-Language Models (LALMs) excel at audio understanding but expose little about where in an audio signal they attend. We introduce instruction-based vector steering, which constructs a steering vector by contrasting activations from differently instructed prompts while keeping the audio fixed. Through a systematic probe of LALM attention, we find that - unlike standard prompting or audio-based steering - this intervention significantly redistributes the temporal attention allocated to audio tokens, concentrating it on acoustically relevant regions. We then show that this attention shift is behaviorally meaningful: in a controlled three-event setting, reading out the temporal position of maximal steering-induced attention change recovers the location of a queried sound event without any training, attaining 60.87% and 68.72% overlap with ground-truth intervals on Qwen2-Audio and Audio Flamingo 3, far above direct prompting (31.84%, 46.75%) and random baselines (27.74%). Our results characterize a mechanistic property of instruction-based steering in LALMs and provide a training-free probe for the latent temporal structure these models encode.

Full-duplex spoken language models (FD-SLMs) enable seamless speech interaction by allowing models to listen and speak simultaneously, yet the internal mechanism by which they coordinate listening and speaking remains underexplored. We analyze the predictive behavior encoded in FD-SLM hidden representations and find that they exhibit stream-specific predictive patterns: during listening, they preferentially predict the incoming user stream, whereas during speaking, they preferentially predict the model output stream. Building on this observation, we show that FD-SLMs dynamically modulate their internal predictive focus between two states: a generative state aligned with model output generation and a perceptive state aligned with incoming user input. However, this modulation can lag behind abrupt changes in conversational context. During user interruptions, the model remains transiently biased toward the generative state before transitioning into the perceptive state, causing it to miss the beginning of the incoming input. We term this delayed internal transition state inertia. To quantify its downstream impact, we introduce the Zero-Buffer Benchmark (ZBB), a diagnostic benchmark for evaluating immediate interruption comprehension when user speech begins abruptly. We evaluate this setting using response correctness and initial-word occurrence rate (IWOR). Finally, we mitigate state inertia through activation steering with a perception vector, a training-free intervention with little additional computational overhead. Across multiple state-of-the-art FD-SLMs, activation steering substantially improves interruption handling; for example, on PersonaPlex, it improves correctness from 28% to 45% and IWOR from 40% to 72% without any fine-tuning.

Although symmetricity in the converter controller is desirable for robust stability margins, a direct link between system-level asymmetricity and instability has yet to be clearly established. Converter control introduces three-phase asymmetricity through loops such as DC-link voltage control, a phase-locked loop , and a power synchronization loop. Furthermore, the inherently asymmetric topology of the two-level voltage-source converter, which converts a DC voltage into a three-phase balanced set, acts as the underlying origin of the asymmetries that propagate into the control structure. Consequently, establishing a direct relationship between system asymmetricity (rather than control asymmetricity alone) and the stability margin is essential for understanding the underlying instability mechanisms. In this work, asymmetricity is quantified using the Asymmetricity Quantification Index (AQI), derived from the sequence-domain representation of the interconnected converter-grid impedance. Within this domain, symmetricity is identified through the definition of symmetrical matrices, which serve as the benchmark against which asymmetricity is measured. A robust and generalized analysis correlates AQI with the stability margin, including both grid-following and grid-forming control structures connected to the power grid. It is found that instability arises from increased asymmetricity in the combined converter-grid system, which is dominated by asymmetric control loops and operating points. Thus, reducing asymmetricity without compromising controller functionality can improve stability margins. The analysis is validated in both control-hardware-in-the-loop and power-hardware-in-the-loop environments.