arXiv subjects

Merouane Debbah

Publications and source records attributed to Merouane Debbah.

At least 73 records · Page 4Linked to original sources

arXiv2026

One-Step Generative Channel Estimation via Average Velocity Field

Generative models have shown immense potential for wireless communication by learning complex channel data distributions. However, the iterative denoising process associated with these models imposes a significant challenge in latency-sensitive wireless communication scenarios, particularly in channel estimation. To address this challenge, we propose a novel solution for one-step generative channel estimation. Our approach bypasses the time-consuming iterative steps of conventional models by directly learning the average velocity field. Through extensive simulations, we validate the effectiveness of our proposed method over existing state-of-the-art diffusion-based approach. Specifically, our scheme achieves a normalized mean squared error up to 2.65 dB lower than the diffusion method and reduces latency by around 90%, demonstrating the potential of our method to enhance channel estimation performance.

cs.IT↗

arXiv2026

Reasoning Beyond Limits: Advances and Open Problems for LLMs

Recent breakthroughs in generative reasoning have fundamentally reshaped how large language models (LLMs) address complex tasks, enabling them to dynamically retrieve, refine, and organize information into coherent multi-step reasoning chains. Techniques such as inference-time scaling, reinforcement learning, supervised fine-tuning, and distillation have been effectively applied to state-of-the-art models, including DeepSeek-R1, OpenAI o1 and o3, GPT-4o, Qwen-32B, and various Llama variants, significantly enhancing their reasoning capabilities. In this paper, we present a comprehensive review of the top 27 LLMs released between 2023 and 2025, such as Mistral AI Small 3 24B, DeepSeek-R1, Search-o1, QwQ-32B, and Phi-4, and analyze their core innovations and performance improvements. We also provide a detailed overview of recent advancements in multilingual large language models (MLLMs), emphasizing methods that improve cross-lingual reasoning and address the limitations of English-centric training. In parallel, we present a comprehensive review of progress in state space model (SSM)-based architectures, including models such as Mamba, which demonstrate improved efficiency for long-context processing compared to transformer-based approaches. Our analysis covers training strategies including general optimization techniques, mixture-of-experts (MoE) configurations, retrieval-augmented generation (RAG), chain-of-thought prompting, self-improvement methods, and test-time compute scaling and distillation frameworks. Finally, we identify key challenges for future research, including enabling multi-step reasoning without human supervision, improving robustness in chained task execution, balancing structured prompting with generative flexibility, and enhancing the integration of long-context retrieval and external tools.

cs.LG↗

arXiv2026

$α^3$-Bench: A Unified Benchmark of Safety, Robustness, and Efficiency for LLM-Based UAV Agents over 6G Networks

Large Language Models (LLMs) are increasingly used as high level controllers for autonomous Unmanned Aerial Vehicle (UAV) missions. However, existing evaluations rarely assess whether such agents remain safe, protocol compliant, and effective under realistic next generation networking constraints. This paper introduces $α^3$-Bench, a benchmark for evaluating LLM driven UAV autonomy as a multi turn conversational reasoning and control problem operating under dynamic 6G conditions. Each mission is formulated as a language mediated control loop between an LLM based UAV agent and a human operator, where decisions must satisfy strict schema validity, mission policies, speaker alternation, and safety constraints while adapting to fluctuating network slices, latency, jitter, packet loss, throughput, and edge load variations. To reflect modern agentic workflows, $α^3$-Bench integrates a dual action layer supporting both tool calls and agent to agent coordination, enabling evaluation of tool use consistency and multi agent interactions. We construct a large scale corpus of 113k conversational UAV episodes grounded in UAVBench scenarios and evaluate 17 state of the art LLMs using a fixed subset of 50 episodes per scenario under deterministic decoding. We propose a composite $α^3$ metric that unifies six pillars: Task Outcome, Safety Policy, Tool Consistency, Interaction Quality, Network Robustness, and Communication Cost, with efficiency normalized scores per second and per thousand tokens. Results show that while several models achieve high mission success and safety compliance, robustness and efficiency vary significantly under degraded 6G conditions, highlighting the need for network aware and resource efficient LLM based UAV agents. The dataset is publicly available on GitHub : https://github.com/maferrag/AlphaBench

eess.SY↗

arXiv2025

Context Video Semantic Transmission with Variable Length and Rate Coding over MIMO Channels

The evolution of semantic communications has profoundly impacted wireless video transmission, whose applications dominate driver of modern bandwidth consumption. However, most existing schemes are predominantly optimized for simple additive white Gaussian noise or Rayleigh fading channels, neglecting the ubiquitous multiple-input multiple-output (MIMO) environments that critically hinder practical deployment. To bridge this gap, we propose the context video semantic transmission (CVST) framework under MIMO channels. Building upon an efficient contextual video transmission backbone, CVST effectively learns a context-channel correlation map to explicitly formulate the relationships between feature groups and MIMO subchannels. Leveraging these channel-aware features, we design a multi-reference entropy coding mechanism, enabling channel state-aware variable length coding. Furthermore, CVST incorporates a checkerboard-based feature modulation strategy to achieve multiple rate points within a single trained model, thereby enhancing deployment flexibility. These innovations constitute our multi-reference variable length and rate coding (MR-VLRC) scheme. By integrating contextual transmission with MR-VLRC, CVST demonstrates substantial performance gains over various standardized separated coding methods and recent wireless video semantic communication approaches. The code is available at https://github.com/xie233333/CVST.

cs.IT↗

arXiv2025

From Prompt Injections to Protocol Exploits: Threats in LLM-Powered AI Agents Workflows

Autonomous AI agents powered by large language models (LLMs) with structured function-calling interfaces enable real-time data retrieval, computation, and multi-step orchestration. However, the rapid growth of plugins, connectors, and inter-agent protocols has outpaced security practices, leading to brittle integrations that rely on ad-hoc authentication, inconsistent schemas, and weak validation. This survey introduces a unified end-to-end threat model for LLM-agent ecosystems, covering host-to-tool and agent-to-agent communications. We systematically categorize more than thirty attack techniques spanning input manipulation, model compromise, system and privacy attacks, and protocol-level vulnerabilities. For each category, we provide a formal threat formulation defining attacker capabilities, objectives, and affected system layers. Representative examples include Prompt-to-SQL injections and the Toxic Agent Flow exploit in GitHub MCP servers. We analyze attack feasibility, review existing defenses, and discuss mitigation strategies such as dynamic trust management, cryptographic provenance tracking, and sandboxed agent interfaces. The framework is validated through expert review and cross-mapping with real-world incidents and public vulnerability repositories, including CVE and NIST NVD. Compared to prior surveys, this work presents the first integrated taxonomy bridging input-level exploits and protocol-layer vulnerabilities in LLM-agent ecosystems, offering actionable guidance for designing secure and resilient agentic AI systems.

cs.CR↗

arXiv2025

Lightweight Foundation Model for Wireless Time Series Downstream Tasks on Edge Devices

While machine learning is widely used to optimize wireless networks, training a separate model for each task in communication and localization is becoming increasingly unsustainable due to the significant costs associated with training and deployment. Foundation models offer a more scalable alternative by enabling a single model to be adapted across multiple tasks through fine-tuning with limited samples. However, current foundation models mostly rely on large-scale Transformer architectures, resulting in computationally intensive models unsuitable for deployment on typical edge devices. This paper presents a lightweight foundation model based on simple Multi-Layer-Perceptron (MLP) encoders that independently process input patches. Our model supports 4 types of downstream tasks (long-range technology recognition, short-range technology recognition, modulation recognition and line-of-sight-detection) from multiple input types (IQ and CIR) and different sampling rates. We show that, unlike Transformers, which can exhibit performance drops as downstream tasks are added, our MLP model maintains robust generalization performance, achieving over 97% accurate fine-tuning results for previously unseen data classes. These results are achieved despite having only 21K trainable parameters, allowing an inference time of 0.33 ms on common edge devices, making the model suitable for constrained real-time deployments.

eess.SP↗

arXiv2025

UAVBench: An Open Benchmark Dataset for Autonomous and Agentic AI UAV Systems via LLM-Generated Flight Scenarios

Autonomous aerial systems increasingly rely on large language models (LLMs) for mission planning, perception, and decision-making, yet the lack of standardized and physically grounded benchmarks limits systematic evaluation of their reasoning capabilities. To address this gap, we introduce UAVBench, an open benchmark dataset comprising 50,000 validated UAV flight scenarios generated through taxonomy-guided LLM prompting and multi-stage safety validation. Each scenario is encoded in a structured JSON schema that includes mission objectives, vehicle configuration, environmental conditions, and quantitative risk labels, providing a unified representation of UAV operations across diverse domains. Building on this foundation, we present UAVBench_MCQ, a reasoning-oriented extension containing 50,000 multiple-choice questions spanning ten cognitive and ethical reasoning styles, ranging from aerodynamics and navigation to multi-agent coordination and integrated reasoning. This framework enables interpretable and machine-checkable assessment of UAV-specific cognition under realistic operational contexts. We evaluate 32 state-of-the-art LLMs, including GPT-5, ChatGPT-4o, Gemini 2.5 Flash, DeepSeek V3, Qwen3 235B, and ERNIE 4.5 300B, and find strong performance in perception and policy reasoning but persistent challenges in ethics-aware and resource-constrained decision-making. UAVBench establishes a reproducible and physically grounded foundation for benchmarking agentic AI in autonomous aerial systems and advancing next-generation UAV reasoning intelligence. To support open science and reproducibility, we release the UAVBench dataset, the UAVBench_MCQ benchmark, evaluation scripts, and all related materials on GitHub at https://github.com/maferrag/UAVBench

cs.AI↗

arXiv2025

Federated Attention: A Distributed Paradigm for Collaborative LLM Inference over Edge Networks

Large language models (LLMs) are proliferating rapidly at the edge, delivering intelligent capabilities across diverse application scenarios. However, their practical deployment in collaborative scenarios confronts fundamental challenges: privacy vulnerabilities, communication overhead, and computational bottlenecks. To address these, we propose Federated Attention (FedAttn), which integrates the federated paradigm into the self-attention mechanism, creating a new distributed LLM inference framework that simultaneously achieves privacy protection, communication efficiency, and computational efficiency. FedAttn enables participants to perform local self-attention over their own token representations while periodically exchanging and aggregating Key-Value (KV) matrices across multiple Transformer blocks, collaboratively generating LLM responses without exposing private prompts. Further, we identify a structural duality between contextual representation refinement in FedAttn and parameter optimization in FL across private data, local computation, and global aggregation. This key insight provides a principled foundation for systematically porting federated optimization techniques to collaborative LLM inference. Building on this framework, we theoretically analyze how local self-attention computation within participants and heterogeneous token relevance among participants shape error propagation dynamics across Transformer blocks. Moreover, we characterize the fundamental trade-off between response quality and communication/computation efficiency, which is governed by the synchronization interval and the number of participants. Experimental results validate our theoretical analysis, and reveal significant optimization opportunities through sparse attention and adaptive KV aggregation, highlighting FedAttn's potential to deliver scalability and efficiency in real-world edge deployments.

cs.DC↗

arXiv2025

Stacked Intelligent Metasurfaces for 6G Wireless Networks: Principles, Applications, and Research Directions

The sixth-generation (6G) wireless networks are expected to deliver ubiquitous connectivity, resilient coverage, and intelligence-driven services in highly dynamic environments. To achieve these goals, distributed wireless architectures such as cell-free massive multiple-input multiple-output (MIMO) have attracted significant attention due to their scalability and fairness. Recently, stacked intelligent metasurfaces (SIMs) have emerged as a promising evolution of reconfigurable intelligent surfaces, offering multi-layer electromagnetic domain processing with enhanced controllability and spatial degrees of freedom. By integrating SIMs into distributed wireless networks, advanced wave-domain operations can be realized, enabling efficient interference management, improved energy and spectral efficiency, and robust physical-layer security. This article provides a comprehensive overview of SIM-aided distributed wireless networks, including their application scenarios, classification, and system architectures. Key signal processing challenges, such as hierarchical frameworks, user association, and joint precoding, are discussed, followed by case studies demonstrating significant performance gains. Finally, future research directions in hardware design, energy consumption modeling, algorithm development, and artificial intelligence integration are highlighted, aiming to pave the way for scalable and intelligent 6G distributed wireless networks.

cs.IT↗

arXiv2025

FlowMoE: A Scalable Pipeline Scheduling Framework for Distributed Mixture-of-Experts Training

The parameter size of modern large language models (LLMs) can be scaled up via the sparsely-activated Mixture-of-Experts (MoE) technique to avoid excessive increase of the computational costs. To further improve training efficiency, pipelining computation and communication has become a promising solution for distributed MoE training. However, existing work primarily focuses on scheduling tasks within the MoE layer, such as expert computing and all-to-all (A2A) communication, while neglecting other key operations including multi-head attention (MHA) computing, gating, and all-reduce communication. In this paper, we propose FlowMoE, a scalable framework for scheduling multi-type task pipelines. First, FlowMoE constructs a unified pipeline to consistently scheduling MHA computing, gating, expert computing, and A2A communication. Second, FlowMoE introduces a tensor chunk-based priority scheduling mechanism to overlap the all-reduce communication with all computing tasks. We implement FlowMoE as an adaptive and generic framework atop PyTorch. Extensive experiments with 675 typical MoE layers and four real-world MoE models across two GPU clusters demonstrate that our proposed FlowMoE framework outperforms state-of-the-art MoE training frameworks, reducing training time by 13%-57%, energy consumption by 10%-39%, and memory usage by 7%-32%.

cs.DC↗

arXiv2025

Advanced Architectures Integrated with Agentic AI for Next-Generation Wireless Networks

This paper investigates a range of cutting-edge technologies and architectural innovations aimed at simplifying network operations, reducing operational expenditure (OpEx), and enabling the deployment of new service models. The focus is on (i) Proposing novel, more efficient 6G architectures, with both Control and User planes enabling the seamless expansion of services, while addressing long-term 6G network evolution. (ii) Exploring advanced techniques for constrained artificial intelligence (AI) operations, particularly the design of AI agents for real-time learning, optimizing energy consumption, and the allocation of computational resources. (iii) Identifying technologies and architectures that support the orchestration of backend services using serverless computing models across multiple domains, particularly for vertical industries. (iv) Introducing optically-based, ultra-high-speed, low-latency network architectures, with fast optical switching and real-time control, replacing conventional electronic switching to reduce power consumption by an order of magnitude.

cs.NI↗

arXiv2025

A Unified Foundation Model for Wireless Technology Recognition and Localization

Wireless Technology Recognition (WTR) and localization are essential in modern communication systems, enabling efficient spectrum management, seamless coexistence of diverse technologies, and accurate positioning in dynamic environments. In real-world conditions, solutions must handle signals from various resources with different sampling rates, capturing devices, frequency bands, and propagation conditions. Traditional methods, such as energy detection and conventional Deep Learning (DL) models like Convolutional Neural Networks (CNNs), often lack the robustness to generalize across unseen technologies, environments, or tasks. In this work, we introduce a Transformer-based foundation model for both WTR and localization, pre-trained in a self-supervised manner on large-scale, unlabeled datasets of In-phase and Quadrature (IQ) and Channel Impulse Response (CIR) timeseries. The model leverages input patching for computational efficiency and employs a two-stage pipeline: self-supervised pre-training to learn general-purpose representations, followed by lightweight fine-tuning for task-specific adaptation. This enables the model to generalize to new wireless technologies and unseen environments using minimal labeled samples. Evaluations across short-range and long-range datasets show superior accuracy in WTR (up to 99.99%), Line-Of-Sight (LOS) detection (up to 100%), and ranging error correction (reducing Mean Absolute Error (MAE) by up to 50%), all while maintaining low computational complexity. These results underscore the potential of a reusable wireless foundation model for multi-task applications with minimal retraining.

eess.SP↗

arXiv2025

Introducing Meta-Fiber into Stacked Intelligent Metasurfaces for MIMO Communications: A Low-Complexity Design with only Two Layers

Stacked intelligent metasurfaces (SIMs), which integrate multiple programmable metasurface layers, have recently emerged as a promising technology for advanced wave-domain signal processing. SIMs benefit from flexible spatial degree-of-freedom (DoF) while reducing the requirement for costly radio-frequency (RF) chains. However, current state-of-the-art SIM designs face challenges such as complex phase shift optimization and energy attenuation from multiple layers. To address these aspects, we propose incorporating meta-fibers into SIMs, with the aim of reducing the number of layers and enhancing the energy efficiency. First, we introduce a meta-fiber-connected 2-layer SIM that exhibits the same flexible signal processing capabilities as conventional multi-layer structures, and explains the operating principle. Subsequently, we formulate and solve the optimization problem of minimizing the mean square error (MSE) between the SIM channel and the desired channel matrices. Specifically, by designing the phase shifts of the meta-atoms associated with the transmitting-SIM and receiving-SIM, a non-interference system with parallel subchannels is established. In order to reduce the computational complexity, a closed-form expression for each phase shift at each iteration of an alternating optimization (AO) algorithm is proposed. We show that the proposed algorithm is applicable to conventional multi-layer SIMs. The channel capacity bound and computational complexity are analyzed to provide design insights. Finally, numerical results are illustrated, demonstrating that the proposed two-layer SIM with meta-fiber achieves over a 25% improvement in channel capacity while reducing the total number of meta-atoms by 59% as compared with a conventional seven-layer SIM.

cs.IT↗

arXiv2025

Cooperative Target Detection with AUVs: A Dual-Timescale Hierarchical MARDL Approach

Autonomous Underwater Vehicles (AUVs) have shown great potential for cooperative detection and reconnaissance. However, collaborative AUV communications introduce risks of exposure. In adversarial environments, achieving efficient collaboration while ensuring covert operations becomes a key challenge for underwater cooperative missions. In this paper, we propose a novel dual time-scale Hierarchical Multi-Agent Proximal Policy Optimization (H-MAPPO) framework. The high-level component determines the individuals participating in the task based on a central AUV, while the low-level component reduces exposure probabilities through power and trajectory control by the participating AUVs. Simulation results show that the proposed framework achieves rapid convergence, outperforms benchmark algorithms in terms of performance, and maximizes long-term cooperative efficiency while ensuring covert operations.

cs.RO↗

arXiv2025

Per-antenna power constraints: constructing Pareto-optimal precoders with cubic complexity under non-negligible noise conditions

Precoding matrix construction is a key element of the wireless signal processing using the multiple-input and multiple-output model. It is established that the problem of global throughput optimization under per-antenna power constraints belongs, in general, to the class of monotonic optimization problems, and is unsolvable in real-time. The most widely used real-time baseline is the suboptimal solution of Zero-Forcing, which achieves a cubic complexity by discarding the background noise coefficients. This baseline, however, is not readily adapted to per-antenna power constraints, and performs poorly if background noise coefficients are not negligible. In this paper, we are going to present a computational algorithm which constructs a precoder that is SINR multiobjective Pareto-optimal under per-antenna power constraints - with a complexity that differs from that of Zero-Forcing only by a constant factor. The algorithm has a set of input parameters, changing which skews the importance of particular user throughputs: these parameters make up an efficient parameterization of the entire Pareto boundary.

math.NA↗

arXiv2025

Port-LLM: A Port Prediction Method for Fluid Antenna based on Large Language Models

The objective of this study is to address the mobility challenges faced by user equipment (UE) through the implementation of fluid antenna (FA) on the UE side. This approach aims to maintain the time-varying channel in a relatively stable state by strategically relocating the FA to an appropriate port. To the best of our knowledge, this paper introduces, for the first time, the application of large language models (LLMs) in the prediction of FA ports, presenting a novel model termed Port-LLM. Our proposed method for predicting the moving port of the FA is a two-step prediction method. To enhance the learning efficacy of our proposed Port-LLM model, we integrate low-rank adaptation (LoRA) fine-tuning technology. Additionally, to further exploit the natural language processing capabilities of pre-trained LLMs, we propose a framework named Prompt-Port-LLM, which is constructed upon the Port-LLM architecture and incorporates prompt fine-tuning techniques along with a specialized prompt encoder module. The simulation results show that our proposed models all exhibit strong generalization ability and robustness under different numbers of base station antennas and medium-to-high mobility speeds of UE. In comparison to existing methods, the performance of the port predicted by our models demonstrates superior efficacy. Moreover, both of our proposed models achieve millimeter-level inference speed.

eess.SP↗

arXiv2025

Fluid Antenna Port Prediction based on Large Language Models

This study seeks to utilize large language models (LLMs) to forecast the moving ports of fluid antenna (FA). By repositioning the antenna to the locations identified by our proposed model, we intend to address the mobility challenges faced by user equipment (UE). To the best of our knowledge, this paper introduces, for the first time, the application of LLMs in the prediction of FA ports, presenting a novel model termed Port-LLM. The architecture of our model is based on the pre-trained GPT-2 framework. We designed specialized data preprocessing, input embedding, and output projection modules to effectively bridge the disparities between the wireless communication data and the data format utilized by the pre-trained LLM. Simulation results demonstrate that our model exhibits superior predictive performance under different numbers of base station (BS) antennas and varying UE speeds, indicating strong generalization and robustness ability. Furthermore, the spectral efficiency (SE) attained by our model surpasses that achieved by traditional methods in both medium and high-speed mobile environments.

eess.SP↗

arXiv2025

DeepTelecom: A Digital-Twin Deep Learning Dataset for Channel and MIMO Applications

Domain-specific datasets are the foundation for unleashing artificial intelligence (AI)-driven wireless innovation. Yet existing wireless AI corpora are slow to produce, offer limited modeling fidelity, and cover only narrow scenario types. To address the challenges, we create DeepTelecom, a three-dimension (3D) digital-twin channel dataset. Specifically, a large language model (LLM)-assisted pipeline first builds the third level of details (LoD3) outdoor and indoor scenes with segmentable material-parameterizable surfaces. Then, DeepTelecom simulates full radio-wave propagation effects based on Sionna's ray-tracing engine. Leveraging GPU acceleration, DeepTelecom streams ray-path trajectories and real-time signal-strength heat maps, compiles them into high-frame-rate videos, and simultaneously outputs synchronized multi-view images, channel tensors, and multi-scale fading traces. By efficiently streaming large-scale, high-fidelity, and multimodal channel data, DeepTelecom not only furnishes a unified benchmark for wireless AI research but also supplies the domain-rich training substrate that enables foundation models to tightly fuse large model intelligence with future communication systems.

cs.IT↗