arXiv subjects

Zhonghao Lyu

Publications and source records attributed to Zhonghao Lyu.

At least 19 recordsLinked to original sources

arXiv2026

SWRouter: Similarity-Contractive Window Routing for Multi-Turn Large Language Model Conversations

Large language models exhibit complementary strengths, motivating routing methods that dispatch each query to the most suitable model. Although existing routers are effective in single-turn settings, they do not directly transfer to multi-turn dialogue, where routing performance critically depends on how historical context is segmented, retained, and incorporated into the current prompt. This introduces two fundamental challenges: preventing information loss and information confusion during context construction, and evaluating routing quality without conflating model selection with prompt construction quality. In this paper, we propose SWRouter, a Similarity-Contractive Window Router for multi-turn large language model routing. SWRouter combines a similarity-based context segmentation mechanism for prompt construction with a dual-metric evaluation framework that decouples construction accuracy from router performance. Experiments on multi-turn dialogue benchmarks demonstrate that SWRouter consistently surpasses strong baselines, achieving a 16.26% improvement in evaluation accuracy over the best individual large language model and an additional 8.22% gain over the Conv-ID Context baseline. Our results highlight that multi-turn large language model routing requires a joint design of context construction and evaluation, rather than a direct extension of single-turn routing methods.

cs.CL↗

arXiv2026

Pruning-Aware Multi-Cluster Co-Inference for Large AI Models in AI-RANs

The increasing scale and computational demands of large artificial intelligence models (LAIMs) present significant challenges for efficient inference in resource-constrained distributed environments. In this paper, we propose a multi-cluster LAIM co-inference framework, where an edge server equipped with multiple graphics processing units (GPUs) coordinates multiple user clusters to execute inference tasks collaboratively. Within each cluster, devices capture data from diverse perspectives and employ lightweight on-device LAIMs to extract local features. These features are then transmitted to the edge server, where they are aggregated and fused to generate a more accurate inference outcome. To reveal the fundamental trade-off between model pruning and collaborative inference performance, we develop a theoretical framework that characterizes the impact of pruning ratios and device contributions using rate-distortion theory and partial information decomposition. Based on this analysis, we formulate a joint optimization problem that determines the model pruning ratio, the task scheduling strategy, the bandwidth allocation, and the transmission power, with the goal of minimizing the inference distortion while satisfying the constraints of latency, energy consumption, and server capacity. Extensive simulation results demonstrate that the proposed framework significantly outperforms existing benchmark schemes, achieving superior inference accuracy and resource efficiency in multi-cluster edge intelligence networks.

cs.DC↗

arXiv2026

Sense Smarter, Think Better: A Survey on Edge Perception for Next-Generation Networks

Edge perception has emerged as a foundational capability for future wireless networks, enabling the network edge to proactively sense, interpret, and interact with the physical environment in a task-oriented and resource-aware manner. This survey provides a comprehensive and structured overview of edge perception. We first review representative sensing modalities and edge artificial intelligence (AI) techniques as the fundamental building blocks. We then examine their synergistic interactions. We systematically analyze how edge AI enhances sensing capabilities, encompassing both in-band and out-of-band modalities, as well as multi-modal sensor data fusion. Moreover, we discuss the role of task-driven sensing in facilitating edge AI, including integrated sensing-communication-computation designs, and active perception frameworks that dynamically adapt sensing strategies for downstream applications. Finally, we identify key challenges and open issues. By consolidating fragmented research across sensing, communication, and edge AI, this survey provides forward-looking insights for the design and implementation of edge perception systems for sixth-generation (6G) networks.

eess.SP↗

arXiv2026

Quantization-Aware Collaborative Inference for Large Embodied AI Models

Large artificial intelligence models (LAIMs) are increasingly regarded as a core intelligence engine for embodied AI applications. However, the massive parameter scale and computational demands of LAIMs pose significant challenges for resource-limited embodied agents. To address this issue, we investigate quantization-aware collaborative inference (co-inference) for embodied AI systems. First, we develop a tractable approximation for quantization-induced inference distortion. Based on this approximation, we derive lower and upper bounds on the quantization rate-inference distortion function, characterizing its dependence on LAIM statistics, including the quantization bit-width. Next, we formulate a joint quantization bit-width and computation frequency design problem under delay and energy constraints, aiming to minimize the distortion upper bound while ensuring tightness through the corresponding lower bound. Extensive evaluations validate the proposed distortion approximation, the derived rate-distortion bounds, and the effectiveness of the proposed joint design. Particularly, simulations and real-world testbed experiments demonstrate the effectiveness of the proposed joint design in balancing inference quality, latency, and energy consumption in edge embodied AI systems.

cs.LG↗

arXiv2026

FlexSpec: Frozen Drafts Meet Evolving Targets in Edge-Cloud Collaborative LLM Speculative Decoding

Deploying large language models (LLMs) in mobile and edge computing environments is constrained by limited on-device resources, scarce wireless bandwidth, and frequent model evolution. Although edge-cloud collaborative inference with speculative decoding (SD) can reduce end-to-end latency by executing a lightweight draft model at the edge and verifying it with a cloud-side target model, existing frameworks fundamentally rely on tight coupling between the two models. Consequently, repeated model synchronization introduces excessive communication overhead, increasing end-to-end latency, and ultimately limiting the scalability of SD in edge environments. To address these limitations, we propose FlexSpec, a communication-efficient collaborative inference framework tailored for evolving edge-cloud systems. The core design of FlexSpec is a shared-backbone architecture that allows a single and static edge-side draft model to remain compatible with a large family of evolving cloud-side target models. By decoupling edge deployment from cloud-side model updates, FlexSpec eliminates the need for edge-side retraining or repeated model downloads, substantially reducing communication and maintenance costs. Furthermore, to accommodate time-varying wireless conditions and heterogeneous device constraints, we develop a channel-aware adaptive speculation mechanism that dynamically adjusts the speculative draft length based on real-time channel state information and device energy budgets. Extensive experiments demonstrate that FlexSpec achieves superior performance compared to conventional SD approaches in terms of inference efficiency.

cs.DC↗

arXiv2025

Closing the Generalization Gap in Parameter-efficient Federated Edge Learning

Federated edge learning (FEEL) provides a promising foundation for edge artificial intelligence (AI) by enabling collaborative model training while preserving data privacy. However, limited and heterogeneous local datasets, as well as resource-constrained deployment, severely degrade both model generalization and resource utilization, leading to a compromised learning performance. Therefore, we propose a parameter-efficient FEEL framework that jointly leverages model pruning and client selection to tackle such challenges. First, we derive an information-theoretic generalization statement that characterizes the discrepancy between training and testing function losses and embed it into the convergence analysis. It reveals that a larger local generalization statement can undermine the global convergence. Then, we formulate a generalization-aware average squared gradient norm bound minimization problem, by jointly optimizing the pruning ratios, client selection, and communication-computation resources under energy and delay constraints. Despite its non-convexity, the resulting mixed-integer problem is efficiently solved via an alternating optimization algorithm. Extensive experiments demonstrate that the proposed design achieves superior learning performance than state-of-the-art baselines, validating the effectiveness of coupling generalization-aware analysis with system-level optimization for efficient FEEL.

cs.LG↗

arXiv2025

Robust Multi-modal Task-oriented Communications with Redundancy-aware Representations

Semantic communications for multi-modal data can transmit task-relevant information efficiently over noisy and bandwidth-limited channels. However, a key challenge is to simultaneously compress inter-modal redundancy and improve semantic reliability under channel distortion. To address the challenge, we propose a robust and efficient multi-modal task-oriented communication framework that integrates a two-stage variational information bottleneck (VIB) with mutual information (MI) redundancy minimization. In the first stage, we apply uni-modal VIB to compress each modality separately, i.e., text, audio, and video, while preserving task-specific features. To enhance efficiency, an MI minimization module with adversarial training is then used to suppress cross-modal dependencies and to promote complementarity rather than redundancy. In the second stage, a multi-modal VIB is further used to compress the fused representation and to enhance robustness against channel distortion. Experimental results on multi-modal emotion recognition tasks demonstrate that the proposed framework significantly outperforms existing baselines in accuracy and reliability, particularly under low signal-to-noise ratio regimes. Our work provides a principled framework that jointly optimizes modality-specific compression, inter-modal redundancy, and communication reliability.

eess.IV↗

arXiv2025

Vision-Aided ISAC in Low-Altitude Economy Networks via De-Diffused Visual Priors

Emerging low-altitude economy networks (LAENets) require agile and privacy-preserving resource control under dynamic agent mobility and limited infrastructure support. To meet these challenges, we propose a vision-aided integrated sensing and communication (ISAC) framework for UAV-assisted access systems, where onboard masked De-Diffusion models extract compact semantic tokens, including agent type, activity class, and heading orientation, while explicitly suppressing sensitive visual content. These tokens are fused with mmWave radar measurements to construct a semantic risk heatmap reflecting motion density, occlusion, and scene complexity, which guides access technology selection and resource scheduling. We formulate a multi-objective optimization problem to jointly maximize weighted energy and perception efficiency via radio access technology (RAT) assignment, power control, and beamforming, subject to agent-specific QoS constraints. To solve this, we develop De-Diffusion-driven vision-aided risk-aware resource optimization algorithm DeDiff-VARARO, a novel two-stage cross-modal control algorithm: the first stage reconstructs visual scenes from tokens via De-Diffusion model for semantic parsing, while the second stage employs a deep deterministic policy gradient (DDPG)-based policy to adapt RAT selection, power control, and beam assignment based on fused radar-visual states. Simulation results show that DeDiff-VARARO consistently outperforms baselines in reward convergence, link robustness, and semantic fidelity, achieving within $4\%$ of the performance of a raw-image upper bound while preserving user privacy and scalability in dense environments.

eess.SY↗

arXiv2025

MIMO Pinching-Antenna-Aided SWIPT

Pinching-antenna systems (PASS) have recently emerged as a promising technology for improving wireless communications by establishing or strengthening reliable line-of-sight (LoS) links by adjusting the positions of pinching antennas (PAs). Motivated by these benefits, we propose a novel PASS-aided multi-input multi-output (MIMO) system for simultaneous wireless information and power transfer (SWIPT), where the PASS are equipped with multiple waveguides to provide information transmission and wireless power transfer (WPT) for several multiple antenna information decoding receivers (IDRs), and energy harvesting receivers (EHRs), respectively. Based on the system, we consider maximizing the sum-rate of all IDRs while guaranteeing the minimum harvested energy of each EHR by jointly optimizing the pinching beamforming and the PA positions. To solve this highly non-convex problem, we iteratively optimize the pinching beamforming based on a weighted minimum mean-squared-error (WMMSE) method and update the PA positions with a Gauss-Seidel-based approach in an alternating optimization (AO) framework. Numerical results verify the significant superiority of the PASS compared with conventional designs.

cs.IT↗

arXiv2025

Empowering Intelligent Low-altitude Economy with Large AI Model Deployment

Low-altitude economy (LAE) represents an emerging economic paradigm that redefines commercial and social aerial activities. Large artificial intelligence models (LAIMs) offer transformative potential to further enhance the intelligence of LAE services. However, deploying LAIMs in LAE poses several challenges, including the significant gap between their computational/storage demands and the limited onboard resources of LAE entities, the mismatch between lab-trained LAIMs and dynamic physical environments, and the inefficiencies of traditional decoupled designs for sensing, communication, and computation. To address these issues, we first propose a hierarchical system architecture tailored for LAIM deployment and present representative LAE application scenarios. Next, we explore key enabling techniques that facilitate the mutual co-evolution of LAIMs and low-altitude systems, and introduce a task-oriented execution pipeline for scalable and adaptive service delivery. Then, the proposed framework is validated through real-world case studies. Finally, we outline open challenges to inspire future research.

eess.SP↗

arXiv2025

The Larger the Merrier? Efficient Large AI Model Inference in Wireless Edge Networks

The growing demand for large artificial intelligence model (LAIM) services is driving a paradigm shift from traditional cloud-based inference to edge-based inference for low-latency, privacy-preserving applications. In particular, edge-device co-inference, which partitions LAIMs between edge devices and servers, has emerged as a promising strategy for resource-efficient LAIM execution in wireless networks. In this paper, we investigate a pruning-aware LAIM co-inference scheme, where a pre-trained LAIM is pruned and partitioned into on-device and on-server sub-models for deployment. For analysis, we first prove that the LAIM output distortion is upper bounded by its parameter distortion. Then, we derive a lower bound on parameter distortion via rate-distortion theory, analytically capturing the relationship between pruning ratio and co-inference performance. Next, based on the analytical results, we formulate an LAIM co-inference distortion bound minimization problem by jointly optimizing the pruning ratio, transmit power, and computation frequency under system latency, energy, and available resource constraints. Moreover, we propose an efficient algorithm to tackle the considered highly non-convex problem. Finally, extensive simulations demonstrate the effectiveness of the proposed design. In particular, model parameter distortion is shown to provide a reliable bound on output distortion. Also, the proposed joint pruning ratio and resource management design achieves superior performance in balancing trade-offs among inference performance, system latency, and energy consumption compared with benchmark schemes, such as fully on-device and on-server inference. Moreover, the split point is shown to play a critical role in system performance optimization under heterogeneous and resource-limited edge environments.

cs.LG↗

arXiv2025

Pinching-Antenna Systems (PASS) Aided Over-the-air Computation

Over-the-air computation (AirComp) enables fast data aggregation for edge intelligence applications. However the performance of AirComp can be severely degraded by channel misalignments. Pinching antenna systems (PASS) have recently emerged as a promising solution for physically reshaping favorable wireless channels to reduce misalignments and thus AirComp errors, via low-cost, fully passive, and highly reconfigurable antenna deployment. Motivated by these benefits, we propose a novel PASS-aided AirComp system that introduces new design degrees of freedom through flexible pinching antenna (PA) placement. To improve performance, we consider a mean squared error (MSE) minimization problem by jointly optimizing the PA position, transmit power, and decoding vector. To solve this highly non-convex problem, we propose an alternating optimization based framework with Gauss-Seidel based PA position updates. Simulation results show that our proposed joint PA position and communication design significantly outperforms various benchmark schemes in AirComp accuracy.

cs.IT↗

arXiv2024

CSI-BERT2: A BERT-inspired Framework for Efficient CSI Prediction and Classification in Wireless Communication and Sensing

Channel state information (CSI) is a fundamental component in both wireless communication and sensing systems, enabling critical functions such as radio resource optimization and environmental perception. In wireless sensing, data scarcity and packet loss hinder efficient model training, while in wireless communication, high-dimensional CSI matrices and short coherent times caused by high mobility present challenges in CSI estimation. To address these issues, we propose a unified framework named CSI-BERT2 for CSI prediction and classification tasks, built on CSI-BERT, which adapts BERT to capture the complex relationships among CSI sequences through a bidirectional self-attention mechanism. We introduce a two-stage training method that first uses a mask language model (MLM) to enable the model to learn general feature extraction from scarce datasets in an unsupervised manner, followed by fine-tuning for specific downstream tasks. Specifically, we extend MLM into a mask prediction model (MPM), which efficiently addresses the CSI prediction task. To further enhance the representation capacity of CSI data, we modify the structure of the original CSI-BERT. We introduce an adaptive re-weighting layer (ARL) to enhance subcarrier representation and a multi-layer perceptron (MLP)-based temporal embedding module to mitigate temporal information loss problem inherent in the original Transformer. Extensive experiments on both real-world collected and simulated datasets demonstrate that CSI-BERT2 achieves state-of-the-art performance across all tasks. Our results further show that CSI-BERT2 generalizes effectively across varying sampling rates and robustly handles discontinuous CSI sequences caused by packet loss-challenges that conventional methods fail to address. The dataset and code are publicly available at https://github.com/RS2002/CSI-BERT2 .

cs.LG↗

arXiv2024

Semantic Communications for 3D Human Face Transmission with Neural Radiance Fields

This paper investigates the transmission of three-dimensional (3D) human face content for immersive communication over a rate-constrained transmitter-receiver link. We propose a new framework named NeRF-SeCom, which leverages neural radiance fields (NeRF) and semantic communications to improve the quality of 3D visualizations while minimizing the communication overhead. In the NeRF-SeCom framework, we first train a NeRF face model based on the NeRFBlendShape method, which is pre-shared between the transmitter and receiver as the semantic knowledge base to facilitate the real-time transmission. Next, with knowledge base, the transmitter extracts and sends only the essential semantic features for the receiver to reconstruct 3D face in real time. To optimize the transmission efficiency, we classify the expression features into static and dynamic types. Over each video chunk, static features are transmitted once for all frames, whereas dynamic features are transmitted over a portion of frames to adhere to rate constraints. Additionally, we propose a feature prediction mechanism, which allows the receiver to predict the dynamic features for frames that are not transmitted. Experiments show that our proposed NeRF-SeCom framework significantly outperforms benchmark methods in delivering high-quality 3D visualizations of human faces.

eess.IV↗

arXiv2024

Networked ISAC for Low-Altitude Economy: Coordinated Transmit Beamforming and UAV Trajectory Design

This paper exploits the networked integrated sensing and communications (ISAC) to support low-altitude economy (LAE), in which a set of networked ground base stations (GBSs) cooperatively transmit joint information and sensing signals to communicate with multiple authorized unmanned aerial vehicles (UAVs) and concurrently detect unauthorized objects over the interested region in the three-dimensional (3D) space. We assume that each GBS is equipped with uniform linear array (ULA) antennas, which are deployed either horizontally or vertically to the ground. We also consider two types of UAV receivers, which have and do not have the capability of canceling the interference caused by dedicated sensing signals, respectively. Under each setup, we jointly design the coordinated transmit beamforming at multiple GBSs together with the authorized UAVs' trajectory control and their GBS associations, for enhancing the authorized UAVs' communication performance while ensuring the sensing requirements. In particular, we aim to maximize the average sum rate of authorized UAVs over a given flight period, subject to the minimum illumination power constraints toward the interested 3D sensing region, the maximum transmit power constraints at individual GBSs, and the flight constraints of UAVs. These problems are highly non-convex and challenging to solve, due to the involvement of binary UAV-GBS association variables as well as the coupling of beamforming and trajectory variables. To solve these non-convex problems, we propose efficient algorithms by using the techniques of alternating optimization, successive convex approximation, and semi-definite relaxation. Numerical results show that the proposed joint coordinated transmit beamforming and UAV trajectory designs efficiently balance the sensing-communication performance tradeoffs and significantly outperform various benchmarks.

eess.SP↗

arXiv2024

Embracing Radiance Field Rendering in 6G: Over-the-Air Training and Inference with 3D Contents

The efficient representation, transmission, and reconstruction of three-dimensional (3D) contents are becoming increasingly important for sixth-generation (6G) networks that aim to merge virtual and physical worlds for offering immersive communication experiences. Neural radiance field (NeRF) and 3D Gaussian splatting (3D-GS) have recently emerged as two promising 3D representation techniques based on radiance field rendering, which are able to provide photorealistic rendering results for complex scenes. Therefore, embracing NeRF and 3D-GS in 6G networks is envisioned to be a prominent solution to support emerging 3D applications with enhanced quality of experience. This paper provides a comprehensive overview on the integration of NeRF and 3D-GS in 6G. First, we review the basics of the radiance field rendering techniques, and highlight their applications and implementation challenges over wireless networks. Next, we consider the over-the-air training of NeRF and 3D-GS models over wireless networks by presenting various learning techniques. We particularly focus on the federated learning design over a hierarchical device-edge-cloud architecture, which is suitable for exploiting distributed data and computing resources over 6G networks to train large models representing large-scale scenes. Then, we consider the over-the-air rendering of NeRF and 3D-GS models at wireless network edge. We present three practical rendering architectures, namely local, remote, and co-rendering, respectively, and provide model compression approaches to facilitate the transmission of radiance field models for rendering. We also present rendering acceleration approaches and joint computation and communication designs to enhance the rendering efficiency. In a case study, we propose a new semantic communication enabled 3D content transmission design.

cs.IT↗

arXiv2024

Networked ISAC for Low-Altitude Economy: Transmit Beamforming and UAV Trajectory Design

This paper studies the exploitation of networked integrated sensing and communications (ISAC) to support low-altitude economy (LAE), in which a set of networked ground base stations (GBSs) transmit wireless signals to cooperatively communicate with multiple authorized unmanned aerial vehicles (UAVs) and concurrently use the echo signals to detect the invasion of unauthorized objects in interested airspace. Under this setup, we jointly design the cooperative transmit beamforming at multiple GBSs together with the trajectory control of authorized UAVs and their GBS associations, for enhancing the authorized UAVs' communication performance while ensuring the sensing requirements for airspace monitoring. In particular, our objective is to maximize the average sum rate of authorized UAVs over a particular flight period, subject to the minimum illumination power constraints for sensing over the interested airspace, the maximum transmit power constraints at individual GBSs, and the flight constraints at UAVs. This problem is non-convex and challenging to solve, due to the involvement of integer variables and the coupling of optimization variables. To solve this non-convex problem, we propose an efficient algorithm by using the techniques of alternating optimization (AO), successive convex approximation (SCA), and semi-definite relaxation (SDR). Numerical results show that the obtained transmit beamforming and UAV trajectory designs in the proposed algorithm efficiently balance the tradeoff between the sensing and communication performances, thus significantly outperforming various benchmarks.

eess.SP↗

arXiv2024

Intelligent Reflecting Surface Aided AirComp: Multi-Timescale Design and Performance Analysis

The integration of intelligent reflecting surface (IRS) into over-the-air computation (AirComp) is an effective solution for reducing the computational mean squared error (MSE) via its high passive beamforming gain. Prior works on IRS aided AirComp generally rely on the full instantaneous channel state information (I-CSI), which is not applicable to large-scale systems due to its heavy signalling overhead. To address this issue, we propose a novel multi-timescale transmission protocol. In particular, the receive beamforming at the access point (AP) is pre-determined based on the static angle information and the IRS phase-shifts are optimized relying on the long-term statistical CSI. With the obtained AP receive beamforming and IRS phase-shifts, the effective low-dimensional I-CSI is exploited to determine devices' transmit power in each coherence block, thus substantially reducing the signalling overhead. Theoretical analysis unveils that the achievable MSE scales on the order of ${\cal O}\left( {K/\left( {{N^2}M} \right)} \right)$, where $M$, $N$, and $K$ are the number of AP antennas, IRS elements, and devices, respectively. We also prove that the channel-inversion power control is asymptotically optimal for large $N$, which reveals that the full power transmission policy is not needed for lowering the power consumption of energy-limited devices.

cs.IT↗