arXiv subjects

Falko Dressler

Publications and source records attributed to Falko Dressler.

At least 19 recordsLinked to original sources

arXiv2026

CRISP: Co-designing Pruning and Scheduling for Efficient MoE Inference on Edge Servers

Mixture-of-Experts (MoE) inference on memory-constrained edge platforms is often dominated by expert-weight movement rather than computation. Although each token activates only a small subset of experts, routing is input-dependent, so all experts must remain accessible and memory movement dominates latency. Existing approaches such as uniform pruning and expert dropping overlook two properties of MoE execution: expert combinations are highly skewed in frequency, and experts contribute asymmetrically within each combination. We present CRISP, a combination-aware framework that co-designs pruning and runtime scheduling for efficient MoE serving on memory-constrained edge platforms. CRISP reorders neurons by importance so one stored expert can support multiple execution widths, assigns differentiated execution to hot, warm, and cold combinations based on frequency and expert role, and exploits heterogeneous widths to overlap expert loading with computation. Across Mixtral-8x7B and DeepSeek-MoE-16B, CRISP improves throughput by up to $2.1\times$ at the same memory budget while limiting accuracy loss to 3.6\% relative to full-width inference. Our results highlight that efficient edge MoE serving requires jointly optimizing which expert capacity is preserved and how expert execution is scheduled.

cs.LG↗

arXiv2026

Efficient Exact Unlearning for Federated Zeroth-Order Optimization

Federated zeroth-order optimization (FedZOO) reduces the memory and communication costs of collaborative training for edge intelligence by avoiding backpropagation while communicating only random seeds and scalar coefficients. However, existing methods lack efficient mechanisms for exact unlearning to fully remove the influence of training data when users request deletion. A key insight is that total variation stability of the learning algorithm creates overlap between model distributions before and after data deletion, allowing exact unlearning to reuse computation through coupling. Based on this observation, we jointly design two stable zeroth-order learning algorithms and their corresponding exact-unlearning procedures: \textsc{FedZOO-S} obtains stability through client and mini-batch subsampling, enabling \textsc{FedZOUL-S} to perform efficient unlearning by preserving earlier updates and restarting from the first affected step. \textsc{FedZOO-NS} further combines subsampling with Gaussian noise to increase overlap between update distributions; \textsc{FedZOUL-NS} exploits this overlap through rejection--reflection coupling to reuse affected updates when the coupling is accepted. We prove exact unlearning for both algorithms: the model and saved state have the same joint distribution as retraining with the corresponding learner on retained data. We further establish stability and convergence guarantees for the learning algorithms and derive upper bounds on expected unlearning cost. Experiments demonstrate effective unlearning with competitive model utility and up to $56.1\%$ less cumulative unlearning time than re-optimization. Compared with first-order baselines, our methods reduce communication overhead by up to approximately four orders of magnitude.

cs.LG↗

arXiv2026

One-Shot Federated Model Editing for Device Dynamics at the Edge

Standard federated learning (FL) requires persistent device connectivity across multiple communication rounds, making it unsuitable for dynamic environments such as vehicular networks, where devices frequently join and leave. One-shot FL offers a promising alternative by completing training in a single communication round. However, one-shot FL suffers from model solidification and cannot readily adapt to subsequent client dynamics, including new client contributions and the withdrawal of previous contributions. We therefore consider exact model editing, which removes withdrawn contributions through unlearning or incorporates newly contributed ones through incremental learning without retraining from scratch. We identify that efficient editing fundamentally hinges on learning stability. To this end, we propose F-SOSA, the first stable one-shot FL algorithm that enables rapid and exact model editing. F-SOSA leverages sub-sampling of local models with carefully tuned rate for stable aggregation and performs curvature-aware model fusion to preserve utility. We prove that F-SOSA achieves vanishing training error and reasonably small generalization error bounds. Building on this stability, we further develop two single-round editing algorithms: F-SOSA-U for unlearning and F-SOSA-I for incremental learning. They ensure that the edited model, along with all intermediate algorithmic states, is statistically indistinguishable from retraining while incurring only an expected $O(ρ)$ fraction of its cost, where $ρ$ quantifies stability. Experiments demonstrate that our approach matches retraining performance while reducing cumulative time and communication overhead by at least 36% and 22%, respectively, under realistic urban mobility simulations.

cs.LG↗

arXiv2026

Learning When to Listen: Physics-Informed Synchronization in Molecular Channels

Recent advances in chemical computing create new opportunities to realize biocompatible communication devices for application in precision medicine. Implanted microscopic transmitters and receivers can exchange information using molecules that diffuse through a biological medium, enabling them to form a network and jointly coordinate sensing and actuation operations. However, randomness in molecular propagation causes severe signal distortion, and receivers have limited computational capacity to compensate. The receiver design must therefore combine robustness to diffusion-induced distortion with computational architectures amenable to implementation in the chemical domain. To meet these requirements, we propose a physics-informed reinforcement learning (RL) framework implemented with the reservoir computing (RC) architecture. We study the task of synchronizing the receiver's clock with a sequence transmitted through a molecular communication (MC) channel. The physics part approximates the maximum-likelihood function synchronization objective, while the other learns a timing policy using this objective as its reward. Simulation results yield clock misalignments above half a second, compatible with molecular transmissions operating on the timescale of seconds.

eess.SP↗

arXiv2026

Byzantine-Resilient Federated Large Language Model via Hypergraph Message Passing

Federated Large Language Models (FedLLMs) are promising to deliver privacy-preserving, collaborative intelligence. The decentralized training process is vulnerable to Byzantine attacks that corrupt semantic consistency and factual resilience through malicious model updates. This paper presents a new resilience strategy that prevents semantically corrupted knowledge from propagating during federated aggregation. A defense problem is formulated to jointly maximize global accuracy and suppress manipulated model updates, enabling trustworthy knowledge aggregation without accessing private local data. To strengthen the resilience of FedLLMs, we develop a novel Hypergraph Message Passing Graph Autoencoder (HMP-GAE) framework, where low-rank model updates are represented as nodes in a hypergraph and high-order semantic relationships among multiple agents are exploited through iterative message passing. By leveraging historical checkpoints and cross-agent relational reasoning, HMP-GAE identifies and suppresses semantically inconsistent updates while preserving knowledge alignment among agents throughout federated training. HMP-GAE is implemented using Qwen-2.5 and Llama-3.2 as pretrained FedLLM backbones and evaluated on AG News and Yahoo! Answers datasets. Compared to existing defenses, HMP-GAE improves global accuracy by $1.21$\%, reduces final-model classification semantic entropy by $44.08$\%, and lowers perplexity by $13.34$\%. HMP-GAE outperforms the GAE-based defense by improving the accuracy $1.39$\%, reducing the semantic entropy by $54.5$\%, and lowering perplexity by $35.5$\%.

cs.LG↗

arXiv2026

Toward Reliable and Accurate Predictive ISAC in Mobile mmWave Networks

Integrated sensing and communications (ISAC) systems offer a promising framework for beam tracking, which enhances channel awareness and improves communication reliability in mobile millimeter wave (mmWave) networks. Motivated by this, we propose a low-complexity, model-driven channel state information (CSI) framework with integrated channel prediction capability in ISAC-enabled systems. To reduce the overhead of CSI acquisition and uplink feedback, the proposed framework leverages predicted target sensing parameters to construct partial CSI from both line-of-sight (LoS) and detected non-line-of-sight (NLoS) components. Exploiting this partial CSI, we formulate a beamforming optimization problem that minimizes the worst targets' Cramér-Rao bound (CRB) under per-user communication constraints. This problem is non-convex and highly non-linear; therefore, we develop an efficient suboptimal solution using successive convex approximation (SCA) and Schur-complement-based decomposition techniques, and we evaluate its complexity in comparison with other approaches. Moreover, since obtaining the absolute phase of the channel paths may require additional pilot resources, we formulate a robust beamforming design that does not assume this knowledge, and propose a solution based on sample averaging. Numerical results demonstrate that the ISAC-based robust channel prediction and localization framework significantly improves prediction accuracy and link reliability compared with conventional schemes, such as communication-based localization and tracking and separated sensing and communication. These results highlight the potential of ISAC for reliable, adaptive, and high-frequency, blockage-prone wireless networks. Furthermore, the performance of the proposed framework is evaluated under severe blockage scenarios, which further demonstrates enhanced network reliability.

eess.SP↗

arXiv2026

A Novel Pulse Radar Framework for ISAC-FD-enabled URLLC Systems

We propose a novel pulse radar framework for integrated sensing and communication (ISAC) systems supporting full-duplex (FD) ultra-reliable low-latency communication (URLLC). Motivated by the stringent delay and reliability requirements of sixth-generation (6G) URLLC services, our framework integrates pulse-based radar sensing with sporadic packet transmissions for URLLC users while accounting for residual self-interference (SI) inherent in FD systems. We adopt a time-structured approach in which the silent intervals of the radar pulse are opportunistically exploited for communication. A key focus is minimizing the total energy consumption of the base station, which is critical for sustainable radar-assisted wireless networks. To this end, we jointly optimize the sensing pulse width, beamforming vectors, and URLLC scheduling while ensuring communication and sensing quality-of-service (QoS) constraints. We formulate a non-convex mixed-integer optimization problem addressing sensing and communication requirements jointly. Sensing performance, quantified via the Cramer-Rao bound (CRB), is reformulated using the Schur complement. Finite blocklength capacity approximations capture the URLLC constraint, while probabilistic constraints are modeled using cumulative distribution functions. The original non-convex problem is decomposed into two sub-problems via an alternating optimization algorithm. Simulation results reveal important trade-offs between sensing time, energy consumption, and residual SI. Our proposed method improves system performance while explicitly accounting for SI, and demonstrates improved URLLC reliability with reduced energy consumption. The results show that proper design of the sensing duration and FD scheduling yields a flexible ISAC framework capable of robust operation even under strong interference.

eess.SP↗

arXiv2026

FedSubMuon: Communication-Efficient Federated LLM Fine-Tuning via Structured Subspace Muon

Federated fine-tuning adapts large language models (LLMs) to decentralized client data, but its scalability in cross-device training is often limited by the high communication cost. Muon is an optimizer that improves optimization performance by orthogonalizing momentum for matrix-valued parameters. Existing federated Muon methods demonstrate the benefit of matrix-aware optimization in federated learning, but still require transmitting full layer-size updates and optimizer state. A natural way to reduce communication is to directly apply Muon to LoRA factors, but this changes the optimized object and weakens Muon's matrix-aware update geometry. We propose FedSubMuon, a communication-efficient federated Muon fine-tuning method that optimizes compact coefficient matrices within shared structured subspaces. This design keeps Muon on a single matrix-valued trainable object, while reducing the client upload to compact coefficient matrices. We further introduce FedSubMuon-GT, an accuracy-oriented extension that uses projected gradients to adapt tracked subspace bases toward task-relevant gradient directions. Experiments on instruction tuning and mathematical reasoning show that FedSubMuon-GT achieves the best overall accuracy on four of five dataset-model pairs, while FedSubMuon performs best under all matched communication budgets. On Dolly-15K, the closest communication baseline requires 5.5 times and 1.4 times more total communication on Llama-1B and Qwen-4B, respectively.

cs.LG↗

arXiv2026

FedLNS: Leverage LayerNorm Signature Modeling to Mitigate Adversarial Manipulation in Federated LLMs

Federated training enables language models to learn from distributed private text, but the server cannot directly verify the local supervision or optimization process that produces each client update. A malicious client can therefore train on corrupted targets, introduce incorrect context-token associations, and degrade the global model through repeated aggregation. Such degradation can also increase the risk of unreliable or hallucinatory generation. We propose Federated Learning with Normalization Signatures (FedLNS), a server-side framework for lightweight malicious-update screening. FedLNS represents each client update through changes in trainable normalization-layer parameters and screens suspicious updates against a robust, history-aware cross-client reference. Because the signatures are extracted at the server from the returned local models, FedLNS requires no additional client-to-server parameter or metadata exchange compared to standard federated learning (FL) methods. After screening, the retained full-model updates can be aggregated using standard FL or another compatible aggregation rule. FedLNS requires no raw client data, trusted server dataset, labeled attack examples, or separately trained detector. Experiments on GPT-style, BERT-style, and LLaMA-style models trained from scratch with 200 clients show that, under 40% population-level target manipulation, FedLNS achieves lower test perplexity than the strongest of six baselines for all three architectures under both IID (independently and identically distributed) and non-IID data partitions.

cs.LG↗

arXiv2026

VaporISAC: Integrated Sensing and Communication via Molecular Signals

Conventional electromagnetic (EM)-based integrated sensing and communication (ISAC) systems degrade in cluttered, obstructed, and radio-frequency-hostile environments, while macroscopic molecular communication (MC) remains largely unexplored as an ISAC medium. This article introduces VaporISAC, a molecular ISAC framework in which chemical vapor pulses simultaneously convey information and probe the propagation environment, enabling a one signal, two outputs paradigm. The same received waveform is jointly processed to recover transmitted information and infer environmental properties such as airflow, turbulence, smoke, and chemical conditions. Rather than replacing conventional EM-based ISAC, VaporISAC complements existing approaches in chemically dynamic, infrastructure-limited, and EM-challenged environments. The sensing principles, system architecture, proof-of-concept demonstrations, emerging applications, and open research challenges of VaporISAC are presented, positioning it as a promising new paradigm for resilient communication and environmental sensing.

cs.ET↗

arXiv2026

Graph Representation Learning Augmented Model Manipulation on Federated Fine-Tuning of LLMs

Federated fine-tuning (FFT) has emerged as a privacy-preserving paradigm for collaboratively adapting large language models (LLMs). Built upon federated learning, FFT enables distributed agents to jointly refine a shared pretrained LLM by aggregating local LLM updates without sharing local raw data. However, FFT-based LLMs remain vulnerable to model manipulation threats, in which adversarial participants upload manipulated LLM updates that corrupt the aggregation process and degrade the performance of the global LLM. In this paper, we propose an Augmented Model maniPulation (AugMP) strategy against FFT-based LLMs. Specifically, we design a novel graph representation learning framework that captures feature correlations among benign LLM updates to guide the generation of malicious updates. To enhance manipulation effectiveness and stealthiness, we develop an iterative manipulation algorithm based on an augmented Lagrangian dual formulation. Through this formulation, malicious updates are optimized to embed adversarial objectives while preserving benign-like parameter characteristics. Experimental results across multiple LLM backbones demonstrate that the AugMP strategy achieves the strongest manipulation performance among all competing baselines, reducing the global LLM accuracy by up to $26\%$ and degrading the average accuracy of local LLM agents by up to $22\%$. Meanwhile, AugMP maintains high statistical and geometric consistency with benign updates, enabling it to evade conventional distance- and similarity-based defense methods.

cs.LG↗

arXiv2026

Communicating Smartly in Molecular Communication Environments: Neural Networks in the Internet of Bio-Nano Things

Recent developments in the Internet of Bio-Nano-Things (IoBNT) are laying the foundation for innovative healthcare applications that envision a network of remotely coordinated nanodevices within the human body to monitor and actuate over potential diseases. However, interconnecting such nanodevices requires communication strategies that can cope with molecular communication (MC) channels, whose complex, stochastic, and dynamic behavior often makes accurate physical modeling infeasible. To explore the limits of nanodevice interconnectivity under these conditions, this survey focuses on data-driven communication strategies for MC systems, with particular emphasis on machine learning (ML) methods and neural network (NN) architectures for a robust and adaptive communication scheme at the nanoscale. Research on NN-enabled MC spans several aspects covered in this survey, including NNs for communication in IoBNT networks, the feasibility of biocompatible NN realization, explainable approaches, and the generation of training datasets. We also include open-source code examples to support reproducible research across key MC scenarios. Finally, we identify emerging challenges, including the need for robust NN architectures, biologically integrated NN modules, and scalable training strategies.

eess.SP↗

arXiv2026

ARIADNE: AI-RAN Informed Link Adaptation in Digital Twin Network Environments

Artificial Intelligence (AI)-powered Radio Access Network (RAN) networks have attracted significant attention from both industry and academia. Meanwhile, Digital Twins offer a safe playground for experimenting with AI/Machine Learning (ML)-based solutions for advanced AI-RAN research. By enabling the testing of online algorithms before deployment on the RAN, they reduce costs and safety risks associated with physical field testing. In this article, we propose ARIADNE, an online Reinforcement Learning (RL)-based module that seamlessly integrates with SIONNA and is tasked with performing link adaptation. We explore different design choices and demonstrate how ARIADNE can surpass industry-standard and state-of-the-art methods by achieving up to 11% and 20% improvements in Spectral Efficiency, respectively. Finally, we show that RL learns a Modulation and Coding Scheme (MCS) selection strategy that diverges from Outer Loop Link Adaptation (OLLA), exhibiting either more conservative or more aggressive behavior depending on the configuration, a trend further corroborated by training offline on 5th generation (5G) over-the-air (OTA) measurements.

cs.NI↗

arXiv2026

Physics-Informed Transformer for Multi-Band Channel Frequency Response Reconstruction

Wideband channel frequency response (CFR) estimation is challenging in multi-band wireless systems, especially when one or more sub-bands are temporarily blocked by co-channel interference. We present a physics-informed complex Transformer that reconstructs the full wideband CFR from such fragmented, partially observed spectrum snapshots. The interference pattern in each sub-band is modeled as an independent two-state discrete-time Markov chain, capturing realistic bursty occupancy behavior. Our model operates on the joint time-frequency grid of $T$ snapshots and $F$ frequency bins and uses a factored self-attention mechanism that separately attends along both axes, reducing the computational complexity to $O(TF^2 + FT^2)$. Complex-valued inputs and outputs are processed through a holomorphic linear layer that preserves phase relationships. Training uses a composite physics-informed loss combining spectral fidelity, power delay profile (PDP) reconstruction, channel impulse response (CIR) sparsity, and temporal smoothness. Mobility effects are incorporated through per-sample velocity randomization, enabling generalization across different mobility regimes. Evaluation against three classical baselines, namely, last-observation-carry-forward, zero-fill, and cubic-spline interpolation, shows that our approach achieves the highest PDP similarity with respect to the ground truth, reaching $ρ\geq 0.82$ compared to $ρ\geq 0.62$ for the best baseline at interference occupancy levels up to 50%. Furthermore, the model degrades smoothly across the full velocity range, consistently outperforming all other baselines.

cs.NI↗

arXiv2026

AoI-FusionNet: Age-Aware Tightly Coupled Fusion of UWB-IMU under Sparse Ranging Conditions

Accurate motion tracking of snow particles in avalanche events requires robust localization in global navigation satellite system (GNSS)-denied outdoor environments. This paper introduces AoI-FusionNet, a tightly coupled deep learning-based fusion framework that directly combines raw ultra-wideband (UWB) time-of-flight (ToF) measurements with inertial measurement unit (IMU) data for 3D trajectory estimation. Unlike loose-coupled pipelines based on intermediate trilateration, the proposed approach operates directly on heterogeneous sensor inputs, enabling localization even under insufficient ranging availability. The framework integrates an Age-of-Information (AoI)-aware decay module to reduce the influence of stale UWB ranging measurements and a learned attention gating mechanism that adaptively balances the contribution of UWB and IMU modalities based on measurement availability and temporal freshness. To evaluate robustness under limited data and measurement variability, we apply a diffusion-based residual augmentation strategy during training, producing an augmented variant termed AoI-FusionNet-DGAN. We assess the performance of the proposed model using offline post-processing of real-world measurement data collected in an alpine environment and benchmark it against UWB multilateration and loose-coupled fusion baselines. The results demonstrate that AoI-FusionNet substantially reduces mean and tail localization errors under intermittent and degraded sensing conditions.

eess.SP↗

arXiv2026

BRAIN: Bayesian Reasoning via Active Inference for Agentic and Embodied Intelligence in Mobile Networks

Future sixth-generation (6G) mobile networks will demand artificial intelligence (AI) agents that are not only autonomous and efficient, but also capable of real-time adaptation in dynamic environments and transparent in their decisionmaking. However, prevailing agentic AI approaches in networking, exhibit significant shortcomings in this regard. Conventional deep reinforcement learning (DRL)-based agents lack explainability and often suffer from brittle adaptation, including catastrophic forgetting of past knowledge under non-stationary conditions. In this paper, we propose an alternative solution for these challenges: Bayesian reasoning via Active Inference (BRAIN) agent. BRAIN harnesses a deep generative model of the network environment and minimizes variational free energy to unify perception and action in a single closed-loop paradigm. We implement BRAIN as O-RAN eXtended application (xApp) on GPU-accelerated testbed and demonstrate its advantages over standard DRL baselines. In our experiments, BRAIN exhibits (i) robust causal reasoning for dynamic radio resource allocation, maintaining slice-specific quality of service (QoS) targets (throughput, latency, reliability) under varying traffic loads, (ii) superior adaptability with up to 28.3% higher robustness to sudden traffic shifts versus benchmarks (achieved without any retraining), and (iii) real-time interpretability of its decisions through human-interpretable belief state diagnostics.

cs.IT↗

arXiv2026

Second-Order Convergence in Private Stochastic Non-Convex Optimization

We investigate the problem of finding second-order stationary points (SOSP) in differentially private (DP) stochastic non-convex optimization. Existing methods suffer from two key limitations: (i) inaccurate convergence error rate due to overlooking gradient variance in the saddle point escape analysis, and (ii) dependence on auxiliary private model selection procedures for identifying DP-SOSP, which can significantly impair utility, particularly in distributed settings. To address these issues, we propose a generic perturbed stochastic gradient descent (PSGD) framework built upon Gaussian noise injection and general gradient oracles. A core innovation of our framework is using model drift distance to determine whether PSGD escapes saddle points, ensuring convergence to approximate local minima without relying on second-order information or additional DP-SOSP identification. By leveraging the adaptive DP-SPIDER estimator as a specific gradient oracle, we develop a new DP algorithm that rectifies the convergence error rates reported in prior work. We further extend this algorithm to distributed learning with heterogeneous data, providing the first formal guarantees for finding DP-SOSP in such settings. Our analysis also highlights the detrimental impacts of private selection procedures in distributed learning under high-dimensional models, underscoring the practical benefits of our design. Numerical experiments on real-world datasets validate the efficacy of our approach.

cs.LG↗

arXiv2026

Certified Unlearning in Decentralized Federated Learning

Driven by the right to be forgotten (RTBF), machine unlearning has become an essential requirement for privacy-preserving machine learning. However, its realization in decentralized federated learning (DFL) remains largely unexplored. In DFL, clients exchange local updates only with neighbors, causing model information to propagate and mix across the network. As a result, when a client requests data deletion, its influence is implicitly embedded throughout the system, making removal difficult without centralized coordination. We propose a novel certified unlearning framework for DFL based on Newton-style updates. Our approach first quantifies how a client's data influence propagates during training. Leveraging curvature information of the loss with respect to the target data, we then construct corrective updates using Newton-style approximations. To ensure scalability, we approximate second-order information via Fisher information matrices. The resulting updates are perturbed with calibrated noise and broadcast through the network to eliminate residual influence across clients. We theoretically prove that our approach satisfies the formal definition of certified unlearning, ensuring that the unlearned model is difficult to distinguish from a retrained model without the deleted data. We also establish utility bounds showing that the unlearned model remains close to retraining from scratch. Extensive experiments across diverse decentralized settings demonstrate the effectiveness and efficiency of our framework.

cs.LG↗