Papers

In-network load balancing outperforms traditional software load balancing while costing less. For instance, programmable switch ASICs can use hashing to select the backend server for the initial packet of each flow at the line rate. However, when the pool of available servers changes, ensuring that the subsequent flow packets are mapped to the same server is challenging due to the data plane's limited memory resources and performance requirements. With the emergence of the QUIC transport protocol, several works show how Connection ID fields (CIDs) can embed the server identifier for all non-initial packets. This approach requires modifications on the server side and violates the QUIC specification, which mandates that CIDs remain unlinkable. In this work, we show that stateless QUIC load balancing can be implemented inside the data plane with no changes to CIDs. Moreover, QUIC packets, except the initial client packet, can bypass the load balancer. We also investigate and mitigate attacks on QUIC in this scenario, including full load balancer bypass and 0-RTT IP spoofing.

Wi-Fi networks are moving beyond random channel access toward tightly coordinated operation across access points (APs), a shift reflected in Wi-Fi 8's multi-AP coordination (MAPC). However, the current MAPC specification restricts cooperation to AP pairs, fundamentally limiting the gains achievable in dense deployments and calling for scalable, network-wide coordination in beyond Wi-Fi 8 systems. We target coordinated spatial reuse (Co-SR), where APs transmit concurrently at reduced power. Effective Co-SR demands joint selection and configuration of AP-station transmissions, yet existing approaches simply do not scale: they rely on heavy signaling, slow convergence, unrealistic assumptions, and often require computation time that explodes with network size. We introduce FM4WiFi, a generative ML pipeline that addresses these limitations by producing high-quality Co-SR configurations in a single inference step. FM4WiFi integrates (i) an autoencoder that learns compact latent representations of network states, (ii) a flow-matching generative model that synthesizes feasible Co-SR configurations (including rate control, absent from prior work), and (iii) a surrogate rate predictor that allows rapid, large-scale Co-SR candidate evaluation without dependence on a live system or digital twin. Across extensive evaluations (including experimental validation), FM4WiFi matches or exceeds state-of-the-art baselines at medium-to-large scales and scales to 30+ APs with sub-second inference. Extensive ablation studies validate each modeling and optimization choice.

Virtualization in 5G and beyond networks enables the creation of virtual networks (i.e., network slices) tailored to the needs of different applications. To maximize revenue under limited infrastructure resources, InPs must decide in real time whether to admit incoming slice requests (SRs) based on their resource demands and offered values, while accounting for the opportunity cost of consuming scarce resources. To address this challenge, we introduce Online Pricing-based Slice Admission Control and Resource Allocation (OPA) framework. This framework dynamically assigns pseudo-prices to resources that capture long-term scarcity and anticipated inter-temporal opportunity costs. The short-term admission and resource allocation decisions for each SR are then guided by these prices. Additionally, we design an exponential pricing strategy that guarantees bounded worst-case performance. To improve practical performance, we further develop a data-driven exponential pricing approach that learns from historical data. Evaluations on a real-world network topology show that it improves mean revenue by 32.2% and 26.7% over state-of-the-art DRL and optimization-based approaches, respectively, while reducing computational cost by an order of magnitude relative to the latter.

This paper proposes FedCritic-MIMO, a communication-efficient serverless federated multi-agent reinforcement learning framework for AI-native resource control across independently deployable cell-level controllers in open and disaggregated 6G RANs. Controllers share no trainer, retain local actors and personalized critic components, and exchange only compatible shared critic parameters. FedCritic-MIMO targets reuse-$1$ multi-cell massive-MIMO OFDMA deployments, where RAN controllers jointly manage user scheduling, per-stream power allocation, beamforming, interference, and long-term QoS with limited inter-controller signaling. Each base station locally executes its actor without centralized training or actor federation, while critic knowledge is exchanged peer-to-peer over an interference-aware graph. It enables this collaboration through wireless-aware event triggering, adaptive layer-wise top-$k$ sparse critic exchange with error feedback, and balanced interference-aware fusion. We establish conditional finite-time stationarity and consensus guarantees for the balanced, compressed peer-to-peer critic recursion under a fixed-policy, frozen-target critic-regression model. In strongly interference-coupled reuse-$1$ simulations, FedCritic-MIMO achieves the best performance-communication tradeoff among heuristic, independent-learning, centralized-training, and communication-ablation baselines. It achieves the highest held-out throughput, improves user-rate distribution and mean SINR, increases QoS satisfaction, and attains the lowest interference cost per delivered bit among learning baselines. It reduces critic-communication overhead by $76\%$ relative to uncompressed distributed critic exchange. These results demonstrate that serverless exchange of compatible shared critic parameters can coordinate RAN controllers without centralized trajectory collection or parameter-server aggregation.

Indoor industrial UAV uplink networks face serious blockage and shadowing from shelves, metal equipment, and production facilities. UAVs are also often clustered and fly along similar straight inspection routes at fixed heights. These features make traditional small-cell deployment less suitable, especially when high reliability, continuous coverage, and good service for weak UAVs are required. Cell-free networks can improve robustness through distributed access points (APs) and UAV?centric communications. Reconfigurable holographic surface (RHS)-enabled APs provide programmable analog receive beams and generate scalar post-RHS observations, which are jointly processed at the CPU for distributed uplink MIMO detection at relatively low hardware cost. Conventional AP association relies on distance, large-scale fading, or post-combining SINR obtained with user-specific digital combiners. Here, however, all UAVs served by a single-feed RHS AP share one amplitude?constrained receive pattern and one scalar AP output. We therefore derive an SINR-like score from this physical output and, under weak inter-AP disturbance correlation, approximate the CPU-side log-det objective by an additive per-AP surrogate, yielding a low-complexity ranking rule. The results show that the nearest AP is not always the best choice, the AP-UAV height difference may have an optimal value, and larger serving clusters bring diminishing returns. Simulations show that the proposed method improves the minimum UAV data rate, average spectral efficiency, fairness, and energy efficiency compared with benchmark schemes.

The growing demand for mobile video streaming requires edge delivery systems that adapt efficiently to rapid network fluctuations and diverse user preferences. Existing approaches such as FlyCache rely on reactive ABR heuristics and loosely coupled cache policies, limiting their responsiveness and coordination under real-world wireless dynamics. We propose ProCAVE (Proactive Caching with Adaptive Video Experience), a self-adaptive DRL-based framework that unifies predictive bandwidth modeling, proactive bitrate selection, and preference-aware cache control. ProCAVE employs: (i) a lightweight Transformer for short-term throughput forecasting; (ii) a PPO-driven ABR agent; and (iii) a DDPG-based continuous cache controller operating on a high-dimensional global state. Experiments using MovieLens preference traces and Ghent 4G bandwidth measurements show that ProCAVE improves byte hit rate, reduces backhaul load, and enhances QoE compared with FlyCache and other baselines. These results highlight the benefits of predictive, DRL-coordinated control for efficient and user-centric edge video delivery.

Software-defined wide-area networking concentrates operational authority in controllers, orchestrators, and Internet-facing edges, but severity-only remediation queues do not represent current exposure or evidence quality. This paper specifies Predictive Exposure and Cryptographic Readiness (PECR), a decision-support method that combines nine normalized severity, threat, reachability, and consequence factors while reporting confidence and score intervals separately. Cryptographic dependencies enter an independent migration queue. The evaluation uses 100 role- and zone-conditioned synthetic vulnerability-asset records over 62 role-consistent assets, plus 30 independent generator replications. Against the operational PECR order, CVSS-only, EPSS-only, KEV-first, and CVSS x EPSS queues yield Kendall $\tau$ values of 0.196, 0.221, 0.301, and 0.234; a stronger five-factor context-lite comparator yields 0.755. Across 30,000 joint Dirichlet weight draws, median $\tau$ is 0.836, 0.893, and 0.924 under broad, moderate, and narrow perturbations. Masking 10% of factor cells labels 47 records as evidence-limited because their intervals cross a decision boundary. The study establishes executable specification, synthetic queue differentiation, and stress-test robustness. It does not establish predictive accuracy, avoided loss, or production effectiveness because no real outcome labels are used.

Large LLMs are powerful tools for network automation, but they are expensive, slow to serve, hard to audit, poorly tailored to individual networks, and create long-term dependencies on a small number of AI providers. Existing alternatives fall short: small open-source models are cheaper but unreliable, while deterministic scripts and verification are controllable but hard to build and maintain. We propose a middle ground. An offline loop composed of fuzzing and validation discovers the recurring mistakes small models make, then resulting networking knowledge is expressed as explicit logic rules. In production, these rules guide a small model on tasks such as configuration translation, yielding automation that is cheaper, lower-latency, auditable, and easier to adapt to a specific network. More critically, network knowledge and operational experience stay and are accumulated where they belong, the network itself, not a rented service.

We analyze the timeliness of status-update delivery in a low Earth orbit (LEO) satellite-assisted energy-harvesting Internet of Things network using the Age of Information (AoI) metric. A ground source harvests ambient energy and sends status updates to a remote destination through LEO satellites. Because of satellite mobility, source-to-satellite connectivity alternates between on and off periods whose durations depend on the satellite-ground geometry. The source does not know the connectivity state a priori and therefore employs a probe-before-transmission mechanism: it first expends one energy unit to sense satellite availability and transmits an update only after a successful probe. We combine spherical stochastic geometry with semi-Markov analysis to characterize the coupled evolution of satellite connectivity and the source energy buffer, and derive an analytical expression for the time-average AoI. We then develop a lower-complexity approximation by replacing the instantaneous connectivity state in the energy process with the long-term on-state probability. The resulting approximation is accurate when the energy constraint is weak or satellite connectivity is highly intermittent. Numerical results show that probing can substantially reduce AoI relative to blind transmission by preventing energy expenditure during off periods, particularly under sparse satellite deployment, stringent decoding requirements, or limited energy harvesting.

Mobility robustness optimization (MRO) requires an appropriate selection of handover (HO) parameters such as the time-to-trigger (TTT) and offset margin to balance HO failures and ping-pong HOs. Existing stochastic geometry-based analyses for MRO have treated the angular position of the target base station (BS) as uniformly distributed over a feasible region. However, this treatment does not explicitly capture the spatial distribution of the target BS dynamically selected as a user equipment (UE) moves through the network. In this paper, we develop a stochastic geometry-based analytical framework for MRO in sub-6 GHz cellular networks. We derive the distribution of the HO triggering time and the spatial distribution of the dynamically selected target BS under straight-line UE mobility. Based on these distributions, we formulate too-late HO and ping-pong HO events as mutually exclusive events and analytically derive their probabilities. Numerical results validate the analysis, demonstrate improved accuracy over the conventional uniform-angle model, and reveal the tradeoff between the two HO events and the dependence of the optimal TTT on BS density.

Machine learning is significantly transforming algorithmic trading, yet the requirement for rapid execution speeds persists. While both aspects aim to boost profitability, embedding advanced machine-learning techniques with reduced trading latency presents a notable challenge. Adopting in-network machine learning, which involves offloading inference to programmable network devices, offers a delicate equilibrium in this trade-off. In this paper, we present LOBIN, a solution that utilizes machine learning within the network for market prediction based on high-frequency market data feeds. LOBIN is adept at constructing limit order books and performing inference directly within programmable switches. When compared to server-based benchmarks, LOBIN not only predicts future stock price movements with higher throughput but also maintains robust machine learning performance. It achieves over a 10% reduction in latency compared to the NASDAQ order-matching server benchmark and delivers microsecond-level latency. Furthermore, the machine learning performance of LOBIN can be further enhanced through the adoption of a hybrid deployment approach that integrates both the switch and the servers. Our evaluation demonstrates that among all data feeds of evaluated stocks, the application of hybrid deployment results in approximately 45% of the traffic and 38% of the total potential transaction value being processed within switches without server intervention, reducing latency while ensuring that the average change in error rate of predictions remains at around 3% relative to benchmarks based solely on server use.

LLM deployment is migrating from data centers to edge devices, where Mixture-of-Experts (MoE) models offer a promising path: sparse expert activation allows the model to be spread across multiple low-cost edge nodes. Distributed MoE inference repeatedly dispatches embeddings from one main node to many workers - a one-to-many pattern poorly served by the sequential unicasts of mainstream stacks (NCCL, TCP), yet naturally matched by UDP broadcast. We propose a UDP broadcast method for collaborative edge MoE inference, augmented with timeout-driven retransmission exploiting near deterministic latency in distributed MoE for reliability and unordered result gathering for robustness to expert mispredictions, yielding a consistent 1.4x speedup over NCCL and TCP on a wired 8-node cluster. In wireless settings, however, we uncover a deeper, long-forgotten bottleneck: IEEE 802.11 caps broadcast rates at 54 Mbps regardless of physical-layer capacity - a legacy policy built for sparse control traffic, not edge AI. NS-3 simulations at distances 1m, 2m and 5m show that the optimal rates are much higher (64x, 43x, and 32x, respectively) than the 54 Mbps cap applied in standard. Thus, we argue that broadcast is no longer a control-plane relic: it is time for Wi-Fi standards to treat it as a high-throughput data-plane citizen.

Massive connectivity, a key building block of 5G, is expected to play an important role in the next generation of wireless systems, and its expected requirements are being revolutionized through the modeling of the information dynamics related to the vast numbers of Internet of things (IoT) devices. Motivated by this, the present paper introduces a model that captures the spatio-temporal nature of freshness of information sent via random access channel policies from an extremely large set of IoT devices via simple scalar parameters, i.e., the probability of success and accuracy of received updates. There are many information freshness metrics, starting from the age of information (AoI), all of which are proxies for the actual application performance, characterized over the temporal dimension. Our model adds the spatial dimension to this picture, observing that sensors distributed over the same area may have a strong correlation, and information from multiple close-by sensors may improve the overall accuracy of the receiver. We focus on characterizing the uncertainty of the receiver, expressed through the conditional entropy, considering a network of partially reliable, spatially distributed sensors observing the same process and reporting their measurements over a slotted ALOHA channel. We consider a simple forgetful receiver and a more complete model which accounts for the full history of past observations, deriving their performance, and optimizing the transmission probability of nodes to minimize the expected uncertainty.

The modern transport protocol QUIC is designed to enhance network performance and security, but it remains vulnerable to handshake flooding attacks. Such attacks exhaust CPU resources by forcing the server to perform expensive cryptographic operations via a large number of handshaking requests. QUIC provides a built-in defense mechanism, the Retry mechanism, to mitigate these attacks. However, our experiments reveal that it can still become a performance bottleneck under large-scale QUIC handshake floods due to substantial computational overhead. In this paper, we design and implement TurboRetry, a split design, that offloads the Retry mechanism onto DPUs to efficiently mitigate QUIC handshake floods. TurboRetry partitions the tasks of the Retry into two categories, and then assigns them to the DPUs and the host, respectively. To preserve QUIC semantics and reduce the coordination overhead, TurboRetry designs an extended Retry token format and an efficient cooperation scheme. In addition, TurboRetry offloads the connection authorization task to the on-path DPA to further improve both performance and security. Our evaluation shows that TurboRetry outperforms the host-side implementation by a wide margin, improving throughput by 10-20$\times$.

In Robot Operating System 2 (ROS 2), Data Distribution Service (DDS) participants must discover one another before exchanging data. In wireless environments, delayed or lost discovery messages cause reliability timers to expire, triggering retransmissions that intensify channel contention and further delay the delivery of discovery messages. This self-reinforcing feedback can escalate into a discovery storm. Existing models characterize discovery demand under fixed delivery conditions, but do not capture how shared-channel delay changes protocol state and generates further traffic. To address this issue, we present the first closed-loop analytical model of ROS 2 discovery that characterizes how delay-induced feedback amplifies retransmission overhead and leads to severe discovery storms. Our model represents channel contention as a shared service process, coupling message-delivery latency with receiver states and reliability timers. The model predicts both discovery completion time and per-class message counts. We validate the model through 1,350 experimental runs across 90 topology configurations. An open-loop airtime baseline captures only a fraction of the high-load completion time. The closed-loop model reproduces this rise and conservatively upper-bounds the observed high-load range. Guided by insights from the model, we further design a response-aware discovery policy that reduces mean discovery completion time by 25.3% to 39.7%.

Post-quantum cryptography (PQC) is becoming an important component of long-term trust in Internet-facing infrastructure. Publicly observable PQC support provides evidence of externally visible deployment, but does not necessarily reflect the overall progress of an organisation's post-quantum migration. This distinction matters when observable deployment is used as an indicator of organisational readiness or progress towards migration deadlines. We present a measurement study of observable PQC deployment across 4,665 UK organisations spanning ten sectors. We measure post-quantum key-exchange support across HTTPS and SMTP STARTTLS endpoints, attribute reachable endpoints to their underlying infrastructure providers where possible, and statistically examine protocol-, sector-, and provider-level deployment patterns. Among reachable endpoints, 44.0% of HTTPS services supported at least one evaluated PQC key-exchange group, compared with 6.4% of SMTP services. Among organisations reachable over both protocols, HTTPS support was significantly more common than SMTP support (matched odds ratio 16.89). Although deployment varied across sectors, infrastructure provider identity was substantially more predictive than organisational sector, and observable deployment was highly concentrated among a small number of providers. Only 144 organisations supported PQC across both web and email infrastructure, highlighting an uneven and fragmented migration landscape. No post-quantum certificate signatures were observed across the measured endpoints. These findings show that observable PQC deployment is currently shaped predominantly by infrastructure-provider deployment decisions and should not be interpreted as a complete measure of organisational migration readiness.

Evaluating congestion control is inherently challenging because performance depends on the interaction between the congestion-control algorithm, transport stack, application behaviour, measurement process, and network dynamics. This challenge is growing as state-of-the-art protocols incorporate pacing, selective loss recovery, model-based control, and, more recently, reinforcement learning. Low Earth Orbit (LEO) satellite networks are a particularly demanding setting: rapidly changing paths, handovers, non-congestive loss, RTT variation, and transient hotspots all affect transport behaviour. This paper reports the lessons learned from an extensive evaluation campaign across both simulation and emulation for LEO satellite congestion control. We compare multiple classes of congestion-control algorithms, including Cubic, BBR variants, LEO-specific protocols, and reinforcement-learning-based control, using comparable implementations across OMNeT++/INET simulation and Mininet-based emulation with the Linux transport stack. This gives us a rare opportunity to examine not only protocol performance, but also the methodological strengths and limitations of each experimental environment. Our findings show that simulation is indispensable for constellation-scale exploration, controlled parameter sweeps, and future deployment studies, but can miss behaviours caused by production transport-stack mechanisms such as pacing, SACK, RACK, kernel timing, and rate sampling. Emulation exposes these implementation-dependent effects and provides a necessary validation step, but is harder to scale and less exactly repeatable. We distil these experiences into practical lessons for combining simulation and emulation to obtain results that are scalable, reproducible, and deployment-relevant.

The Open Radio Access Network (O-RAN) architecture introduces flexible functional splits and open interfaces that enable distributed and centralized deployment of baseband processing. While this flexibility offers opportunities for improved resource utilization, it also introduces fundamental trade-offs between energy efficiency and latency. In this paper, we develop a throughput-based end-to-end energy consumption model for O-RAN and extend it by incorporating detailed latency modeling and application-specific Artificial Intelligence/Machine Learning inference costs. The proposed end-to-end modeling framework provides a general representation of processing, transport, and inference-related energy and delay across the access, metro, and long-haul network segments. Building on this general model, we formulate an optimization problem that selects the placement of baseband processing and AI inference tasks across candidate O-RAN configurations to analyze energy-latency tradeoffs under network load, server frequency, and energy-budget constraints. Using representative hardware platforms and realistic traffic assumptions, we evaluate multiple baseband processing placements corresponding to different O-RAN functional configurations. Our results reveal how user quality of service requirements and network load conditions jointly determine the optimal placement of baseband processing and AI inference tasks, highlighting the inherent trade-off between energy efficiency and latency. The analysis provides practical insights for latency-aware and energy-efficient O-RAN deployments supporting emerging AI-driven services.

Recent years have seen the evolution of the traditional Radio Access Network (RAN) toward more open, programmable, disaggregated, and intelligent architectures, known as an Open RAN. Future Next Generation (NextG) networks are envisioned to be AI-native, enabling data-driven closed-loop optimization of Base Station resources, while Reconfigurable Intelligent Surfaces (RIS) emerge as key enablers for wireless propagation and spectral efficiency toward 6G and beyond. This dissertation focuses on the design, optimization, and experimental evaluation of NextG RANs integrating Open RAN principles, data-driven control loops, and intelligent resource allocation. The work emphasizes cross-layer optimization, including energy-efficient power control, and explores AI-driven network slicing, scheduling, and link adaptation, demonstrating NextG RANs reconfigurable in real time to meet 6G requirements, first analyzing architectural enablers and modeling frameworks, then prototyping and evaluating solutions on experimental platforms and Digital Twins. Main contributions include: (i) Deep Reinforcement Learning (DRL) solutions for network slicing and scheduling; (ii) PandORA, a framework for automatic design, training, and deployment of DRL-based Open RAN applications on the Colosseum wireless network emulator; (iii) physical-layer RIS channel modeling and optimized resource allocation across spectrum bands; (iv) system-level evaluation of RIS-assisted channels for eMBB and URLLC traffic; (v) integration of RIS within Open RAN; (vi) online RL solutions for link adaptation; and (vii) spectrum sharing between cellular and Non-Terrestrial Network links via power control and beamforming. This work provides algorithmic designs, frameworks, and validation from simulation and hardware-in-the-loop emulation to over-the-air 5G testbed experiments, addressing industry and academic needs for wireless research.

This paper investigates collaborative mobile edge computing (MEC) servers for large language model (LLM) inference under soft deadline constraints. In this system, to improve the quality of service, computations are expected to be completed within their deadlines. However, due to dependencies among tasks or subtasks, any missed deadline can lead to catastrophic consequences for the entire request. In this context, this work proposes an extended deadline mechanism with constrained flexibility. The main challenges lie in handling large-scale computations under strict latency constraints while limiting the number of allowable deadline extensions, especially in the presence of task dependencies within each request. To tackle these challenges, we develop a transformer-enhanced proximal policy optimization (PPO) framework that enables efficient collaboration among MEC servers. The proposed approach aims to maximize the number of tasks completed within their deadlines while minimizing the use of deadline extensions. By capturing temporal dependencies and cross-server interactions, the transformer improves decision-making for task migration. Simulation results demonstrate that the proposed method significantly outperforms conventional PPO and heuristic-based approaches in terms of task completion rate and overall system efficiency.

The deployment of Vision-Language Models (VLMs) in autonomous driving (AD) systems is constrained by on-board computing power, restricting vehicles to small VLMs (SVLMs) with limited perception and reasoning capabilities. Infrastructure-assisted AD alleviates this resource constraint by enabling collaboration with large VLMs (LVLMs) at edge servers. However, in dynamic vehicular environments, the utility of sensory data for downstream tasks decays rapidly, making timeliness of information a critical concern. To balance the accuracy gains of LVLMs with their latency-induced timeliness degradation, we develop a Timeliness-Aware Large-Small VLM Collaboration (TALSC) framework. Specifically, we first model the Age of Information (AoI) evolution for VLM inference and characterize the coupling among AoI, token length, and task performance to formulate a general timeliness metric. Building on this, we propose the TALSC online scheduling algorithm. Since scheduling decisions have a delayed impact on future timeliness metric and the output token number is unknown at scheduling time, we design a Lyapunov drift-plus-estimated-penalty algorithm and provides a guaranteed performance. In simulation, we first conduct a case study to derive a fitted timeliness metric based on nuScenes dataset, and further show that TALSC outperforms baselines under various communication and computing settings, achieving up to a 12.6% normalized improvement in Micro-F1 score compared with the best-performing baseline.

SD-WAN teams often use static severity scores to decide what to fix first. These scores do not show live exploitation, network exposure, attack paths, business impact, or cryptographic migration risk. This study asks whether a vendor-neutral PECR framework can produce a different and more traceable ranking than CVSS alone. We reviewed 19 DOI-verified publications from 2020-2026 and five current NIST documents. The review supported ten normalized factors, one weighted score, and a separate confidence measure. We then compared equal-weight PECR with CVSS in five synthetic cases. The tests used Kendall's $\tau_b$, mean absolute rank shift (MARS), top-three Jaccard overlap ($J_3$), factor removal, and bounded weight changes. CVSS ranked the cases A-D-B-E-C. PECR ranked them A-B-E-C-D. The results were $\tau_b = 0.40$, MARS = 1.2, and $J_3 = 0.50$. The exact $\tau$ test was not significant ($p = 0.483$) because the sample had only five cases. The PECR order remained unchanged in 87.5% of 1,024 weight combinations. It also remained unchanged in eight of ten single-factor removal tests. The evidence supports separate measures for severity, exploitation, and organizational context. Evidence for live SD-WAN attack paths and daily PQC triage is still limited. PECR can produce a different and auditable ranking. This synthetic test does not prove better operational results or better human understanding.

Modern computing and networking infrastructure emits telemetry continuously, yet operators convert it into decisions with a separate predictor per task, entity, and horizon. One generative model, pretrained once over an operator's own event streams, could replace this fleet, an approach that already scales to high-cardinality streams in recommendation systems. However, point-forecast error on operational telemetry saturates near simple last-value baselines, so lower error alone need not improve the decisions it feeds. To close this gap, we present \sys, a generative foundation model for network and systems telemetry, evaluated by replaying the decisions its calibrated conditional quantiles drive. \sys treats heterogeneous telemetry as event-driven, irregularly timed entity streams and serves flexible forecast horizons in a single pass, requiring no future timestamps. To our knowledge, \sys is the first pretrained telemetry model to improve both HPC scheduling and network provisioning decisions under replay, its runtime estimator transferring zero-shot across months and its pretrained weights across domains from hours of target data. Extensive experiments on HPC job logs and network traffic confirm that \sys lowers the mean bounded slowdown of backfilling by up to approximately $77\%$ over deployed user estimates and roughly halves the deployed rule's violation rate. Our code is available at https://github.com/ZzZTripleZzZ/all-in-one.

Modern machine learning training workloads run on large-scale networks of compute accelerators. The networks commonly deployed in these systems are typically variations of two basic topologies: the fat-tree Clos and the torus. In this paper, we derive analytical results the elucidate how the choice of topology shapes achievable performance for the small set of collective communication operations that underlies modern machine learning workloads. We also consider how these results change when we include additional factors such as network failures and job placement strategies. Overall, we find that one topology does not dominate in all cases, but that the Clos achieves better collective completion time in most cases and provides benefits in resilience and flexibility.

The integration of Low Earth Orbit (LEO) Non-Terrestrial Networks (NTNs) into 5G and upcoming 6G architectures introduces various challenges, including severe propagation delays, ultra-high base station mobility, and channel non-stationarity, complicating radio resource management of heterogeneous network slices. In this paper, we propose a deep reinforcement learning (DRL) meta-scheduler for twin-timescale resource allocation. Our solution adopts a decoupled Open Radio Access Network (RAN) architecture, in which a strategic 100 ms meta-scheduler selects scheduling policies for the different network slices using stale telemetry, while a fast-timescale MAC packet scheduler processes per-TTI user requests. The resulting Markov Decision Process captures non-stationary orbital dynamics and heterogeneous SLAs constraints via a TD3 agent. Simulation results under varying traffic load show that, unlike other solutions, the proposed meta-scheduler explicitly trades a statistically insignificant 1% capacity fraction (p > 0.05) to strictly bound the variance and overall magnitude of RLC-layer queuing delay for Mission-Critical (MC) traffic. Crucially, it enforces this isolation without inducing the broadband slice starvation characteristic of standard maximum-CQI heuristics, establishing a robust foundation for 6G O-RAN NTN resource allocation.

Routing in Low Earth Orbit (LEO) satellite networks is challenging due to highly dynamic topologies and spatio-temporal network conditions. Reinforcement Learning (RL) has emerged as a promising approach for adaptive routing; however, its performance critically depends on reward function design, which must balance objectives such as goodput and end-to-end delay. In practice, reward design remains a complex manual process requiring significant domain expertise and extensive trial-and-error. Recent works have explored Large Language Models (LLMs) for automated reward design, but their application to highly dynamic systems such as LEO satellite networks remains largely unexplored. We propose LARGE, a framework that automates reward design for RL-based routing by combining LLM- driven generation with iterative simulator-in-the-loop evaluation. LARGE generates an initial reward from LLM prior knowledge and iteratively refines it using simulation feedback. This loop enables exploration of diverse reward formulations while aligning them with network objectives. Results show that LARGE improves reward quality within a few iterations through feedback-driven refinement. Across different backbones, the framework achieves performance comparable to an expert-designed baseline, with the best-performing configuration reaching goodput within approximately 3% of the baseline and slightly lower end-to-end delay, without manual reward engineering. These results indicate that effectiveness emerges from the iterative feedback-driven process enabled by LARGE, highlighting the potential of framework-driven LLM-in-the-loop optimization for RL-based routing in dynamic satellite networks.

In vehicle-to-infrastructure (V2I) communication the setting of IEEE 802.11 Distributed Coordination Function (DCF) parameters has a decisive bearing on performance, yet the literature seldom pins down how much each parameter actually matters once traffic, MAC and queueing are modelled together. Treating a previously validated analytical framework as a fixed deterministic input-output map, we rank the DCF and traffic parameters that shape throughput, collision probability, delay, packet delivery ratio and Age of Information in a single-AP V2I network. A local one-factor-at-a-time analysis, cast in dimensionless elasticities so that parameters of different units become comparable, is paired with a variance-based global analysis built on first-order and total-effect Sobol indices. Two clean groups emerge: collision probability is set by the contending-vehicle population – itself governed by vehicle velocity and density – together with the minimum contention window, whereas delay is driven by the channel rate, the offered load and the packet size, and carries strong interaction effects that no local reading can expose. We then derive the closed-form structure of these sensitivities from the model relations, which explains the rankings, forces certain parameters into equal-magnitude elasticities, and locates where the local ranking reverses. Finally the collision-sensitivity structure is turned into a design output rather than a ranking: a closed-form contention-window control law, linear in the contending population and closed with a Greenshields density model, that a roadside access point can evaluate online from measured density or velocity. The fixed IEEE 802.11 default is recovered as the single population at which this law is optimal; away from it the throughput gain grows with density and is largest in the dense, safety-critical regime.