TNSM Searchable Index

Author(s)	Title	Year	Publication	Keywords
Shuang Zheng, Xing Zhang, Michael Sheng, Haixu Wang, Wenbo Wang	Beam Hopping Low Earth Orbit Satellite Resource Allocation for Differentiated Services and Robustness Analysis under Model Attacks	2026	Early Access	Beams Satellites Resource management	Beam hopping (BH)-enabled Low Earth Orbit (LEO) satellites play a pivotal role in next-generation communication networks, providing global coverage, improving spectrum efficiency, and supporting flexible adaptation to heterogeneous service demands. To fully exploit these capabilities, artificial intelligence (AI) techniques are increasingly employed for dynamic resource allocation and power management. However, limited onboard resources and potential adversarial perturbations pose challenges to both efficiency and robustness. To address these issues, we leverage digital twin technology to accurately capture the spatio-temporal dynamics of user–satellite visibility, providing precise state information for decision-making. Building on this, we formulate a joint optimization framework for BH scheduling and power allocation as a Markov Decision Process and propose the BRIDGE—BH with Reinforcement learning incorporating Integrated Dirichlet and Gumbel-TopK Exploration—which integrates a quality of service (QoS)-driven subchannel scheduling mechanism to ensure efficient and differentiated resource allocation. The model’s robustness is systematically evaluated under three classical adversarial attacks. Simulation results demonstrate that our approach achieves superior energy efficiency, service throughput, and fairness, while the robustness analysis shows stable performance under the considered bounded adversarial perturbations.	10.1109/TNSM.2026.3710750
David Kule Mukuhi, Leo Mendiboure, Rami Langar, Rodrigue Fargeon, Sylvain Cherrier, Marion Berbineau, Pierre-Yves Petton	Application-aware Slicing for FRMCS: A Deep Reinforcement Learning Approach	2026	Early Access		The Future Railway Mobile Communication System (FRMCS) will replace GSM-R to support safety-critical and high-throughput applications over a limited 5–10 MHz spectrum. Railway services range from ultra-reliable train control, such as European Train Control System and Automatic Train Operation, to bandwidth-intensive video surveillance and best-effort passenger Wi-Fi, each with distinct requirements. Existing network slicing solutions designed for public 5G networks focus on aggregate slice-level guarantees, neglecting heterogeneous application requirements and the strong channel fluctuations induced by high-speed train mobility. To overcome this limitation, we propose in this paper an Application-Driven Slice Scheduling (ADSS) approach tailored for railway communications. ADSS leverages Deep Reinforcement Learning combined with channel-aware resource allocation to dynamically assign Resource Blocks, ensuring application-level Service Level Agreement (SLA) fulfillment. Evaluations on real Signal-to-Noise Ratio traces from trains traveling at speeds up to 350 km/h, demonstrate that ADSS achieves superior application-level SLA satisfaction, reduces violation gaps, and improves spectral efficiency compared to heuristic and state-of-the-art schedulers.	10.1109/TNSM.2026.3710830
Lion Steger, Liming Kuang, Johannes Zirngibl, Georg Carle, Oliver Gasser	Still on Target? An Evaluation of IPv6 Target Generation Algorithms	2026	Early Access		Internet measurements are a crucial foundation of IPv6-related research. Due to the infeasibility of full address space scans for IPv6 however, those measurements rely on collections of reliably responsive, unbiased addresses, as provided e.g., by the IPv6 Hitlist service. Although used for various use cases, the hitlist provides an unfiltered list of responsive addresses, the hosts behind which can come from a range of different networks and devices, such as web servers, customer-premises equipment (CPE) devices, and Internet infrastructure. In this paper, we demonstrate the importance of tailoring hitlists in accordance with the research goal in question. By using PeeringDB we classify hitlist addresses into six different network categories, uncovering that 42% of hitlist addresses are in ISP networks. Moreover, we show the different behavior of those addresses depending on their respective category, e.g., ISP addresses exhibiting a relatively low lifetime. Furthermore, we analyze different Target Generation Algorithms (TGAs), which are used to increase the coverage of IPv6 measurements by generating new responsive targets for scans. We use seed sets, e.g., based on the categorized Hitlist. We evaluate the performance of TGAs under various conditions and find generated addresses to show vastly differing responsiveness levels for different TGAs. Furthermore, we evaluate of algorithm run times and differences between multiple TGA runs.	10.1109/TNSM.2026.3705935
Matheus Dória, Ricardo Silva, Carlos Lima, Daniel Luna, Allan Martins, Paulo Eduardo, Augusto Neto, Vicente Sousa	Open RAN KPI Monitoring: a Dynamic Energy-Aware Framework	2026	Early Access		Open RAN emerged to break out of vendor lock-in constraints, enabling specialized control-plane applications with well-defined open and standardized interfaces. In a monitoring-oriented Real-Time Radio Access Network Intelligent Controller, xApps typically collect Key Performance Indicators (KPIs) at fixed intervals, which can overload the system when very frequent reporting is required. This paper introduces a new Open RAN KPI monitoring framework that outperforms fixed-interval solutions by dynamically adjusting the KPI collection timescale at runtime, using Risk Analysis of monitored KPI accuracy loss to balance monitoring needs and resource usage. A prototype built with OpenAirInterface and FlexRIC shows reduced E2 signaling overhead and significant energy savings. Extrapolating to large-scale scenarios, the framework demonstrates potential savings of up to 608.48 kWh per day (18,254.4 kWh per month), equivalent to an estimated reduction of 0.99 tons of CO2 emissions.	10.1109/TNSM.2026.3713811
Tong Li, Shicheng Wei, Wencheng Yang, Yan Li	HotPatchCaps: A Capsule Network with Runtime Hot Patching for Zero-Day API Attack Detections	2026	Early Access		Modern services are awash in Application Programming Interfaces (APIs), yet most security pipelines end at predeployment testing using fuzzers and scanners. This leaves a runtime gap where payload obfuscation and other evolving request-visible misuse patterns outpace static rules and slow retraining cycles. We present HotPatchCaps, an expert-in-the-loop runtime framework that closes this gap by hot patching expert knowledge into a capsule architecture without retraining. HotPatchCaps fuses Term Frequency–Inverse Document Frequency (TF–IDF) statistics on request tokens with security cues such as parameter names, encodings, and payload substrings, and employs slot-controlled routing to amplify semantically relevant evidence into interpretable capsule activations. New rules arrive as lightweight runtime patches that can be injected on the fly, aligning with operational practice while preserving the generalization of learned models. We evaluated the CSIC 2010 dataset and the ATRDF 2023 dataset in both in-distribution and zero-day settings against classical machine learning (ML) and deep baselines. Experimental results demonstrate that HotPatchCaps consistently improves accuracy and recall at competitive precision and remains robust under label noise and schema drift. By turning expert knowledge into patchable capsule priors, HotPatchCaps provides a practical path from testing to on-call defence for API-centric systems.	10.1109/TNSM.2026.3713465
Soonbeom Kwon, Yusu Noh, Youngwoo Jang, Illyoung Choi, Byungchul Tak, In-geol Chun, Young-Kyoon Suh	Scalable and Robust Resource Provisioning via Adaptive Task Scheduling for Edge Devices	2026	Early Access	Schedules Scheduling Cloning	Edge devices, such as wearables, drones, and CCTV systems, are vital for real-time data collection in urban intelligence. However, their limited computational and storage capacities pose significant challenges. While offloading to public clouds offers scalability, it often incurs high latency and operational costs. Conversely, centralizing workloads on edge servers may result in the underutilization of high-performance edge devices. To address these limitations, we introduce ERPF, a Kubernetes-based Edge Resource Provisioning Framework that augments the capabilities of heterogeneous edge environments. ERPF orchestrates dynamic volume provisioning, GPU-aware resource allocation, execution context migration, and adaptive task distribution to improve system flexibility and efficiency. Building on this, we propose a novel adaptive task scheduling technique, termed eATS, composed of three key mechanisms: (i) Partition Smoothing Scheme for stable task granularity control, (ii) Resilient Edge Reintegration for failure detection and task reassignment, and (iii) Competitive Task Cloning for speculative execution with fastest-result commitment. The proposed eATS scheme reduces task execution time by up to 27.6%, lowers partition size variability by 8.7×, and improves scheduling robustness across heterogeneous edge devices over the baseline.	10.1109/TNSM.2026.3694238
Weilin Wang, Xiaojing Fan, Huachun Zhou, Jingfu Yan, Aoran Huang	A Collaborative Mechanism for Edge-Offloading and Intelligent Intrusion Detection Services	2026	Early Access	Algorithms Security Training	Mobile edge computing (MEC) is a promising technology for supporting computing-intensive and delay-sensitive applications. The network operator can enhance users’ personalized service experiences by implementing advanced offloading solutions. However, existing schemes often overlook security risks posed by malicious users, and struggle to balance quality of service (QoS) and security capabilities. To this end, we propose a collaborative mechanism for edge offloading and intelligent intrusion detection services to optimize personalized service experiences for normal users at the task level. First, we introduce a new optimization model, Collaboration of Edge Offloading and Intelligent Intrusion Detection Services (CEOI2DS), tailored for MEC environments with malicious users, considering security decisions, resource allocation, and function placement decision-making steps. It aims to maximize the operator’s average long-term revenue while meeting QoS requirements and resource constraints, encouraging the operator to deliver optimal security capabilities while ensuring personalized QoS for users. Then, to tackle this problem, we design a Collaborative Three-Agent Deep Reinforcement Learning (CTADRL) algorithm. Three agents conduct collaborative training and decision-making by interacting with the MEC environment. They comprehensively analyze user requirements, risk probabilities, and network resource status to formulate optimal service policies, enhancing the overall experience for normal users. Experimental results demonstrate that under different user risk probabilities and computing resources, the proposed mechanism and algorithm exhibit better adaptability and stability regarding processing success rate and revenue.	10.1109/TNSM.2026.3713143
Jing-Yang Voon, Yao Chiang, Hung-Yu Wei	Resource Allocation and Container Scaling for Microservices in Multi-Cluster Edge Computing System	2026	Early Access	Resource management Optimization Delays	With the advent of the 6G era and the evolution of distributed systems, edge computing has become a pivotal architecture for deploying latency-sensitive, resource-efficient applications. In particular, the microservice architecture, characterized by modular and loosely coupled components, has gained significant traction for building scalable and maintainable applications at the network edge. However, deploying microservice-based applications in heterogeneous and geographically distributed Multi-Cluster Edge Computing (MCEC) environments presents critical challenges, especially in achieving efficient and scalable resource management. Although existing research has explored resource allocation and container scaling for microservice-based systems, most prior works consider container efficiency in isolation or within single-cluster or cloud-centric environments, without jointly addressing container-level efficiency, inter-cluster task offloading, and resource allocation in MCEC scenarios. To address this gap, we propose RACCOON, a request-offloading cascaded resource allocation algorithm tailored for microservice-oriented deployments in MCEC settings. RACCOON aims to minimize user-perceived service latency while optimizing overall resource utilization. Complementing this, we introduce RAS-CAL, a reinforcement learning (RL)-based container scaling mechanism that dynamically adjusts resource provisioning at the container level to further enhance system performance. Experimental evaluation shows that our approach consistently outperforms methods that address only resource allocation, only task offloading, or only container scaling, by jointly optimizing these dimensions to reduce end-to-end user-perceived latency and computational overhead.	10.1109/TNSM.2026.3713212
David Segura, Emil J. Khatib, Raquel Barco	Evaluation of Mobile Network Slicing in a Logistics Distribution Center	2026	Early Access	5G mobile communication Massive machine type communications Timing	Logistics is a key economic sector where any optimization that reduces costs or improves service has a great impact on society at large. Network Slicing (NS) is a technique that allows the creation of different independent networks with different dedicated resources on a shared physical infrastructure. This is particularly useful in scenarios where different applications with different requirements coexist. In this paper, an open-source simulator based on NS-3 with several implemented extensions has been developed, with a realistic representation of a distribution center scenario, including the logistics activities that take place there. Under this developed simulator, the role of two 5G NS strategies in Smart Logistics is studied: the use of a static balanced division and the use of a dynamic allocation of resources between slices. These strategies have been evaluated in terms of Quality of Service (QoS) for different traffic profiles via simulations. Results show that a dynamic slice allocation makes a more efficient usage of the network resources, improving the QoS for the different traffic profiles, even when there is a traffic peak. This improvement ranges from 6.48% to 95.65%, depending on the specific traffic profile and the evaluated metric.	10.1109/TNSM.2026.3712270
Zihan Jia, Chen Chen, Alia Asheralieva, Lin Guan, Ziren Xiao	HAN: Adaptive DRL-based Congestion Control via Model Uncertainty	2026	Early Access	Modeling Algorithms Uncertainty	Congestion control (CC) algorithms are commonly categorized as rule-based, learning-based, and hybrid approaches. Rule-based methods such as Cubic and BBR provide predictable behavior and low overhead, but they often adapt poorly to dynamic or previously unseen network conditions. Deep reinforcement learning (DRL)-based CC algorithms can improve adaptability and average performance, yet they still suffer from limited generalization, unsafe exploration, instability, and long-tail latency in unfamiliar environments. Hybrid designs attempt to combine the strengths of both families by guiding DRL agents with expert policies. However, training-time expert guidance can restrict exploration, impose a performance ceiling, and, in some designs, unintentionally reward fallback-triggering actions. We propose HAN (Hybrid with Adaptive Uncertainty-based Navigation), a hybrid CC framework that switches between a learned DRL policy and a traditional CC algorithm according to model uncertainty at inference time. Rather than constraining the learning process with expert rules, HAN estimates the confidence of DRL decisions and selectively defers to a traditional CC algorithm when uncertainty is high. This uncertainty-aware mechanism preserves the adaptability of DRL while improving stability and robustness across diverse network scenarios. We compare HAN with state-of-the-art CC algorithms under varying bandwidth, buffer size, random loss, and heterogeneous network environments. Empirical results show that HAN achieves up to 83% higher throughput and reduces average latency by 36% compared with existing CC methods. These findings highlight the potential of uncertainty-aware decision-making for future CC algorithm design.	10.1109/TNSM.2026.3711195
Zhizhou He, Mohammad Shojafar, Rahim Tafazolli	Service-oriented Attention HAPPO for xApps Coordination in Digital Twin Enabled AI-RAN	2026	Early Access	Modeling Open RAN Training	Coordinating multiple heterogeneous xApps in Open Radio Access Networks (O-RAN) is challenging because real-time inter-xApp synchronization introduces communication overhead, latency, and scalability bottlenecks under live deployments. This paper proposes an attention-based Heterogeneous-Agent Proximal Policy Optimization (HAPPO) framework that coordinates four heterogeneous xApps—power control, resource-block allocation, access control, and beam selection—through a Digital Twin (DT)-enabled training pipeline. The framework combines (i) hybrid actor heads supporting mixed continuous/discrete actions, (ii) task-aware self-attention for implicit coordination without predefined communication graphs, and (iii) DT-based asynchronous experience collection with delay modelling, online correction, and safe rollback. We further validate that the learned attention aligns with the physical coupling between xApps (Pearson ρ = 0.78±0.04), establishing an interpretable link between the policy and underlying network dependencies. On a VIAVI O-RAN RIC emulator across 10 random seeds, the proposed scheme attains a 91.2% aggregate QoS score and 1.05 ms inference latency, improves per-slice SLA satisfaction by up to 6.4% over strong MARL baselines (FACMAC, MAPPO, CommNet), and reaches the 80%-of-final performance level 1.49× faster than the attention-free counter-part. Improvements are statistically significant (p < 0.05, paired two-sided Welch t-tests with Holm–Bonferroni correction across baselines; under fluctuating UE demands and diverse service-level requirements).	10.1109/TNSM.2026.3710685
Yongqiang Dong, Jiangnan Sun, Jiawen Li, Yongbo Liu	Learning to Configure Like Engineers: Manual Guided Network Configuration Sketch Generation	2026	Early Access	Modeling Syntactics Large language models	Network configuration automation is a key component of intelligent network operations aiming to transform user intents into executable device configurations. Most existing approaches take a paradigm of parameter filling within predefined sketches, where the sketches have to be crafted manually by engineers and user intents are expressed in a specific format. Other studies follow a routine of synthesizing configurations directly from natural-language intents, taking advantage of large language models (LLMs) and retrieval augmented generation techniques. The results are yet far from satisfactory in practice due to the complexity of the network configuration requirements. A recently proposed example-driven configuration synthesis method (CEGS), attempts to learn from configuration examples provided by vendors. However, its effectiveness is bounded by example coverage, and the method struggles to generalize to new scenarios. To address this, we present LCLE, an end-to-end sketch generation framework that learns how to configure networks from device configuration guides and command references, much as human engineers do. Specifically, LCLE automatically generates configuration sketches from natural-language intents by LLMs with a structured device configuration model (DCM) extracted from vendor manuals. The DCM organizes configuration workflows, command syntax, and view hierarchies into a unified knowledge base that supports LCLE’s retrieval-augmented generation through a three-stage pipeline of intent parsing, sketch generation, and sketch optimization. Extensive experiments on Huawei and Cisco devices show that LCLE significantly improves the semantic completeness and syntactic correctness of the generated configuration sketches. In addition, the framework can be easily extended to new devices and protocols through DCM updates, promising a scalable solution for automated network configuration.	10.1109/TNSM.2026.3710600
Behrooz Farkiani, Fan Liu, Ke Yang, John DeHart, Jyoti Parwatikar, Patrick Crowley	Hermes: A General-Purpose Proxy-Enabled Networking Architecture	2026	Early Access	Tunneling HTTP Joining processes	We introduce Hermes, a general-purpose networking architecture that aims to improve service delivery over the Internet. Hermes delegates networking responsibilities from applications and services to proxies and is designed as a portable, adaptable solution to four fundamental challenges of efficient service delivery over the Internet: end-to-end traffic management, backward compatibility, data-plane security and privacy models, and adaptable communication layers. The design centers on an overlay of reconfigurable proxies and HTTP tunneling and proxying techniques, utilizing assisting components to extend proxy functionality when needed. Through prototyping and emulation, we demonstrate that Hermes improves key performance metrics across multiple use cases: it provides backward compatibility through protocol translation and tunneling, improves reliability by delegating retry logic to proxies, enables unified policy-based Layer 3 routing across network segments, and serves as an efficient substrate for future architectures like NDN, facilitating their operation over the Internet. Beyond evaluating Hermes across various use cases, we measured the overhead of Hermes’ HTTP tunneling and proxying mechanisms and found it to be modest, typically under 2 ms per proxy pair traversal in an isolated collocated setup. Although the HTTP proxying and tunneling techniques used by Hermes increase single-connection processing overhead, we also show that, with up to 1,000 concurrent requests, proxies can amortize connection setup time and reduce end-to-end latency by utilizing connection pooling and multiplexing.	10.1109/TNSM.2026.3705327
Masoumeh Safkhani, Mohammad Reza Servati, Fatemeh Rezaei	HEIoT: A Novel Three-Factor Authentication Protocol for Enhanced Security in IoT and Next-Generation Networks	2026	Early Access	Authentication Internet of Things Protocols	The Internet has a significant impact on contemporary society, enabling a wide range of applications, including advanced cellular networks such as 4G, 5G, and 6G. Since these communications occur over shared or open channels, ensuring secure data exchange is of critical importance, as any weakness in the communication infrastructure may compromise system reliability. Device authentication in the Internet of Things (IoT) and user authentication in smart environments, such as smart homes, remain fundamental security challenges. As the first line of defense, authentication mechanisms must be robust, since vulnerabilities at this stage can expose the entire system to serious threats. To address these challenges, numerous authentication schemes based on cryptographic primitives, including Elliptic Curve Cryptography (ECC), have been proposed. In this paper, we present a comprehensive security analysis of an ECC-based three-factor authentication protocol proposed by Yuan et al. Our analysis shows that the protocol is vulnerable to desynchronization, user impersonation, traceability, and insider attacks, all of which succeed with probability 1 by exploiting at most two protocol phases. To mitigate these weaknesses, we propose an improved authentication scheme, called HEIoT. The proposed scheme is formally analyzed under the Real-or-Random (RoR) model to establish session-key security and is further verified using the Scyther tool. Moreover, a Python-based implementation is provided to demonstrate the practicality of the proposed protocol. Comparative results indicate that HEIoT achieves stronger security while maintaining acceptable communication, computational, and storage overhead.	10.1109/TNSM.2026.3702041
Dev Gurung, Shiva Raj Pokhrel	LLM-QFL: Distilling Large Language Model for Quantum Federated Learning	2026	Early Access	Modeling Federated learning Large language models	As Quantum Federated Learning (QFL) scales toward distributed quantum networks, managing heterogeneous resources and communication bottlenecks becomes a critical challenge. This research proposes LLM-QFL, an adaptive network service management framework that leverages Large Language Models (LLMs) to optimize the operational efficiency of QFL systems. We introduce a federated distillation method in which locally fine-tuned LLMs serve as autonomous network agents. These agents adaptively manage service parameters by: i) dynamically adjusting local computation intensity (optimizer steps) based on loss gradients, ii) performing variance-aware client selection to minimize network-wide heterogeneity, and iii) implementing intelligent early stopping criteria to conserve bandwidth. By serving as an orchestration layer, LLM-QFL provides a synergy between LLMs and quantum networking. Our contributions include: i) Adaptive Performance and Efficiency: Reducing idle computation and significantly cutting communication overhead; ii) Theoretical Rigor: Convergence guarantees of O(1/T) for the adaptive management protocol; and iii) Scalable Deployment: Implementing PEFT (LoRA/QLoRA) for resource-constrained quantum service nodes.	10.1109/TNSM.2026.3712394
Ci-Yi Hung, Li-Yu Yang, Li-Der Chou	LMM: A Reinforcement-Learning-Based Mitigation Mechanism of Lateral Movement in Kubernetes	2026	Early Access	Modeling Advanced driver assistance systems Containers	With the growing adoption of microservices architecture, Kubernetes—while offering a variety of built-in security modules—remains vulnerable to lateral movement due to its highly interconnected network architecture and frequent misconfigurations in permission settings. This study proposes the Lateral Movement Mitigation (LMM) mechanism, which integrates event tracking, risk assessment, and reinforcement learning (RL) to enhance Kubernetes' defense against lateral movement. LMM leverages Falco with custom rules to capture container-level events and utilizes a high-order Markov chain to construct a transition probability matrix for estimating the likelihood of command sequences. These transition probabilities are then used for risk assessment and provided as input states to the RL agent. The RL agent selects mitigation actions based on recommendations from the MITRE ATT&CK framework, thereby dynamically strengthening Kubernetes' native security modules. Experiments show that LMM improves accuracy by 17.00% over Warp and F1-score by 23.30% over ADA in Kubernetes namespace bypass. In the Role-Based Access Control (RBAC) misconfiguration, LMM outperforms Warp by 18.53% in accuracy and 28.27% in F1-score. In terms of mitigation latency, LMM achieves up to 98.54% and 98.38% faster response times compared to Warp and ADA, respectively, demonstrating its effectiveness and real-time responsiveness. In summary, LMM combines monitoring, risk modeling, and automated decision-making to deliver an efficient and accurate proactive solution against lateral movement in Kubernetes.	10.1109/TNSM.2026.3713179
Deemah H. Tashman, Soumaya Cherkaoui	Trustworthy AI-Driven Dynamic Hybrid RIS: Joint Optimization and Reward Poisoning-Resilient Control in Cognitive MISO Networks	2026	Early Access	Reconfigurable intelligent surfaces Reliability Optimization	Cognitive radio networks (CRNs) are a key mechanism for alleviating spectrum scarcity by enabling secondary users (SUs) to opportunistically access licensed frequency bands without harmful interference to primary users (PUs). To address unreliable direct SU links and energy constraints common in next-generation wireless networks, this work introduces an adaptive, energy-aware hybrid reconfigurable intelligent surface (RIS) for underlay multiple-input single-output (MISO) CRNs. Distinct from prior approaches relying on static RIS architectures, our proposed RIS dynamically alternates between passive and active operation modes in real time according to harvested energy availability. We also model our scenario under practical hardware impairments and cascaded fading channels. We formulate and solve a joint transmit beamforming and RIS phase optimization problem via the soft actor-critic (SAC) deep reinforcement learning (DRL) method, leveraging its robustness in continuous and highly dynamic environments. Notably, we conduct the first systematic study of reward poisoning attacks on DRL agents in RIS-enhanced CRNs, and propose a lightweight, real-time defense based on reward clipping and statistical anomaly filtering. Numerical results demonstrate that the SAC-based approach consistently outperforms established DRL base-lines, and that the dynamic hybrid RIS strikes a superior trade-off between throughput and energy consumption compared to fully passive and fully active alternatives. We further show the effectiveness of our defense in maintaining SU performance even under adversarial conditions. Our results advance the practical and secure deployment of RIS-assisted CRNs, and highlight crucial design insights for energy-constrained wireless systems.	10.1109/TNSM.2026.3660728
Jing Zhang, Chao Luo, Rui Shao	MTG-GAN: A Masked Temporal Graph Generative Adversarial Network for Cross-Domain System Log Anomaly Detection	2026	Early Access	Anomaly detection Adaptation models Generative adversarial networks	Anomaly detection of system logs is crucial for the service management of large-scale information systems. Nowadays, log anomaly detection faces two main challenges: 1) capturing evolving temporal dependencies between log events to adaptively tackle with emerging anomaly patterns, 2) and maintaining high detection capabilities across varies data distributions. Existing methods rely heavily on domain-specific data features, making it challenging to handle the heterogeneity and temporal dynamics of log data. This limitation restricts the deployment of anomaly detection systems in practical environments. In this article, a novel framework, Masked Temporal Graph Generative Adversarial Network (MTG-GAN), is proposed for both conventional and cross-domain log anomaly detection. The model enhances the detection capability for emerging abnormal patterns in system log data by introducing an adaptive masking mechanism that combines generative adversarial networks with graph contrastive learning. Additionally, MTG-GAN reduces dependency on specific data distribution and improves model generalization by using diffused graph adjacency information deriving from temporal relevance of event sequence, which can be conducive to improve cross-domain detection performance. Experimental results demonstrate that MTG-GAN outperforms existing methods on multiple real-world datasets in both conventional and cross-domain log anomaly detection.	10.1109/TNSM.2026.3654642
Madhura Adeppady, Yenchia Yu, Ali Rahmanian, Ahmed Ali-Eldin Hassan, Carla Fabiana Chiasserini	Efficient Management of Composite Heterogeneous Applications at the Network Edge	2026	Early Access	Central Processing Unit Servers Resource management	Edge computing is a promising paradigm for deploying latency-sensitive applications (Apps) as it brings resources closer to end users. Edge Apps often adopt a microservice (MS) architecture, breaking monolithic Apps into lightweight, containerized MSs that can be dynamically and independently deployed. However, managing such Apps involves three key challenges: (i) optimizing the placement of MSs to reduce both response time and resource overhead, (ii) handling MS migration or relocation as users move while minimizing App service disruption (App downtime), and (iii) enabling MS sharing across Apps while ensuring performance guarantees. We formulate this as an optimization problem, named Multi-microservice Application Placement (MAP), prove its NP-hardness, and introduce STEP (State and Topology-aware Edge-MS Placement), a polynomial-time heuristic. STEP distinguishes itself from prior work by: (i) jointly considering stateful and stateless MS characteristics in deployment decisions, (ii) exploiting MS shareability to reduce resource usage, (iii) balancing response latency, App downtime, and resource utilization, and (iv) leveraging multiple versions of the same MS to adapt quality of service to available edge resources. Our results in a small-scale scenario show that STEP achieves near-optimal performance with only 7% higher CPU cost than the optimal solution. Large-scale real-time experiments on a Kubernetes cluster demonstrate that STEP consistently outperforms competing methods, achieving up to 50% lower deployment costs while delivering 50% gain in app quality and saving 15% in radio resources with over 90% request success rates.	10.1109/TNSM.2026.3709656
Lu Wei, Yong Yu, Jie Cui, Xianfeng Xie, Jing Zhang, Irina Bolodurina, Hong Zhong	Toward Stable and Low-Latency Task Offloading: A Multi-Agent Framework for Vehicular Edge Computing	2026	Early Access	Vehicles Delays Stability	With the rapid growth of Vehicular Edge Computing (VEC) and Mobile Edge Computing, efficient task offloading is essential for enhancing the computing and communication capabilities in vehicular networks. However, many existing methods suffer from slow convergence, load imbalance, and instability in dynamic, latency-sensitive environments. To address these challenges, we propose MAPPO-Lyapunov (MAPPO-L), a multi-agent offloading framework that integrates Multi-Agent Proximal Policy Optimization (MAPPO) with Lyapunov optimization. MAPPO-L enables distributed coordination among vehicles, roadside units (RSUs), and cloud servers, minimizing delay, improving resource utilization, and ensuring long-term stability. Lyapunov theory transforms long-term stability into per-slot optimizations, while MAPPO ensures efficient policy learning. An adaptive exploration mechanism dynamically adjusts exploration rates based on network dynamics, accelerating convergence and stabilizing training. Extensive simulations with real-world data show that MAPPO-L maintains task completion rates above 80%, converges 25%–37.5% faster than baselines, and reduces training fluctuations to 2.3%. Ablation studies confirm the critical roles of location, channel, and queue information, validating the robustness of MAPPO-L in practical VEC environments.	10.1109/TNSM.2026.3713305