TNSM Searchable Index

Author(s)	Title	Year	Publication	Keywords
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
Lanlan Rui, Yuanrui Yang, Peng Yu, Zhipeng Gao, Yang Yang, Shaoyong Guo	LogPISA: An Improved Pre-training and Tuning Pipeline for Log Understanding with Invariant and Semantic-Aware Objectives	2026	Early Access	Modeling Training Anomaly detection	With the rapid development of computer and network technology, network and software logs generated by a multitude of devices contain a wealth of knowledge and serve as a critical resource for intelligent fault diagnosis and efficient system operations. In recent years, various deep learning methods and the pre-training and fine-tuning paradigm of large language models have achieved significant success in log understanding. However, most existing approaches directly adapt models designed for natural language, overlooking the unique characteristics inherent to log data, such as its distinct vocabulary distribution, structural patterns, and semantic expressions. Concurrently, some template-based methods lack flexibility and are limited in their ability to mine deep semantic information. Therefore, we propose LogPISA, an improved pre-training and tuning pipeline for log understanding with invariant and semantic-aware objectives. Our framework employs a hybrid attention mechanism, combining standard self-attention with our novel Keyword-Aware Sparse Attention to enable the model to capture critical signals more efficiently. During the pre-training phase, we introduce two innovative self-supervised tasks: a non-contrastive learning task based on permutation invariance to capture the flexible ordering within log blocks, and a contrastive learning task based on log summarization to guide the model to focus on core semantics over superficial textual forms. Experimental results on several public benchmarks demonstrate that our model achieves excellent performance on downstream anomaly detection tasks. This validates that our proposed framework learns more robust and generalizable log representations, providing a high-quality representational foundation for various downstream log analysis tasks.	10.1109/TNSM.2026.3719139
Alessandro Buratto, Marco Levorato, Leonardo Badia	DCP: a TCP-Inspired Domain Adaptation in Dynamic Data Drift	2026	Early Access	Modeling Timing Costing	Mobile devices are affected by computing limitations, battery life, and connectivity issues, making it difficult to execute complex machine learning models or frequently transmit data. These challenges hinder real-time adaptability in dynamic environments. To tackle these issues, we introduce a framework between mobile devices and edge servers, where the edge server assists mobile devices by continuously fine-tuning a lightweight classifier to keep up with changes in data patterns. Our approach, called the Drift Control Protocol (DCP), is inspired by how TCP manages network congestion. Just as TCP interprets packet loss as network congestion and throttles transmission, DCP treats spikes in classification error as congestion and implements an additive increase multiplicative decrease mechanism to dynamically control the frequency of model retraining, optimizing the trade-off between the classifier update rate and the communication overhead with the edge server. It dynamically adjusts the update frequency based on how the data distribution shifts and controls how many samples the mobile device sends to the edge server, while improving overall accuracy. We test different versions of DCP on both synthetic and real-world datasets. Our results show that DCP procedures obtain better tradeoffs in mean error and communication costs when compared with constant interval updates policies.We demonstrate this capability across real-world data and parametrized synthetic datasets explicitly designed to simulate both bursty and gradual drift scenarios.	10.1109/TNSM.2026.3719234
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
Anselme Ndikumana, Kim Khoa Nguyen, Oscar Delgado, Adel Larabi, Mohamed Cheriet	Empowering Rural Areas with Energy-Efficient 5G IAB-Based Fixed Wireless Access Network	2026	Early Access		Fixed Wireless Access (FWA) has recently emerged as a cost-effective alternative to optical fiber in rural areas, particularly where fiber deployment is economically infeasible. To extend coverage and increase capacity, FWA networks have begun to integrate Integrated Access and Backhaul (IAB) with mid- and high-band spectrum. However, the energy consumption of multi-hop IAB networks scales significantly with the number of hops, a challenge that prior research has not adequately addressed. This paper proposes an energy-efficient framework that minimizes network energy consumption by maximizing Resource Block (RB) utilization while avoiding both over- and under-allocation in multi-hop IAB-based FWA deployments. The proposed method jointly allocates RBs and selects modulation and coding schemes across a mixed set of 5G numerologies to satisfy data rate requirements while minimizing energy consumption. The inherent dynamic interactions among IAB stations render the problem highly complex and non-convex; therefore, we design a disciplined multi-convex programming supported by dynamic programming algorithms to obtain tractable solutions. Furthermore, we introduce a transformer-based prediction to forecast RB distribution, thereby mitigating the need for frequent short-timescale coordination among IAB stations. Our simulation results demonstrate that the proposed approach achieves the required data rates while reducing energy consumption by 14%.	10.1109/TNSM.2026.3719631
Yang Wu, Xu Cheng, Wenguang Zheng, Yingyuan Xiao	A Geo-Aware Personalized Network for User and Service Representation and Bilinear Interaction Modeling in QoS Prediction	2026	Early Access		With the rapid growth of the Internet, the proliferation of functionally similar web services has made Quality of Service (QoS) prediction, which measures service performance, increasingly critical. In QoS prediction, the QoS values observed from user-service invocations are often significantly affected by their geographical location. However, existing QoS prediction methods typically assume static user and service representations, overlooking geographic differences. We argue that even the same user or service should have personalized representations based on different geographic locations. To address this, we propose GeoPerNet, a Geo-Aware Personalized Network for QoS Prediction. Specifically, we design the Geographical Aware Personalization Module, which models the geographical similarity between users and services to select the most relevant top-k neighbors for the target user or service. We then apply geographic similarity-based weighting to highlight key neighbor information. Next, we leverage the designed GeoTransformer to model the complex dependency relationships among neighbors. Finally, the refined neighbor representations are fused with the original embeddings to generate personalized user and service representations. Additionally, we design the Bilinear Interaction Module to capture fine-grained interaction relationships between users and services using a bilinear function. Experiments on the large-scale WS-DREAM dataset demonstrate that GeoPerNet outperforms state-of-the-art approaches.	10.1109/TNSM.2026.3719697
Bin Dai, Yuntao Wang, Jianhai Zheng	HAT-Route: A Physics-Aware Hierarchical Transformer Framework for Scalable Cloud-Edge Collaborative Routing	2026	Early Access	Routing Modeling Optimization	Routing optimization in cloud-edge collaborative networks faces a fundamental conflict between global strategic planning and local real-time responsiveness, further complicated by structural heterogeneity and stochastic traffic patterns. Traditional protocols lack adaptivity, while existing Deep Rein-forcement Learning (DRL) approaches based on Graph Neural Networks (GNN) struggle with limited receptive fields and over-smoothing issues in large-scale topologies. In this paper, we propose HAT-Route, a Transformer-driven hierarchical routing framework supported by the Network Digital Twin (NDT). Our contributions are threefold: (1) We establish a cloud-edge collaborative architecture operating under the Centralized Training and Decentralized Execution paradigm. This architecture balances the trade-off between global optimization and real-time inference. (2) We introduce FlowFormer, a Spatiotemporal Transformer for the NDT. FlowFormer integrates a novel Edge-Conditioned Spatial Attention (EC-SAT) mechanism to capture physical link constraints and distinguish between congestion and Head-of-Line (HOL) blocking. (3) We design HAT-Route, a hierarchical DRL agent that utilizes Graph Transformers for global policy learning in the cloud, coupled with knowledge distillation to deploy lightweight policies at the network edge. Extensive experiments demonstrate that our framework outperforms traditional protocols and GNN-based baselines in terms of QoS optimization, training stability, scalability, and generalization capability on large-scale network topologies.	10.1109/TNSM.2026.3718004
Ping He, Yu Yao, Xu Li, Yao Hu, Wei Yang	FIGAN: Diversity-Oriented Traffic Generation for Industrial Protocol Format Inference	2026	Early Access	Protocols Modeling Fuzzing	Protocol Format Inference is a pivotal step in the reverse engineering of proprietary protocols, yet its effectiveness is constrained by the scarcity of high-quality training data. In industrial control systems, the rigid and cyclical nature of traffic results in a "long-tail" distribution, where diverse functional scenarios are severely underrepresented. Existing generative approaches, primarily designed for fuzzing or intrusion detection, fail to resolve the intrinsic conflict between syntactic validity and semantic diversity required for protocol format inference. To bridge this gap, we propose FIGAN, a stage-wise decoupled generative framework tailored to synthesize high-fidelity traffic for protocol format inference. By isolating flexible distribution learning from rigid syntax enforcement, FIGAN liberates the generative process to extrapolate novel payload variations from a continuous latent space, effectively surmounting the limitations of sparse seed data. Specifically, the framework integrates three synergistic modules: first, heuristic pre-processing that constructs semantic templates as a prior knowledge base; second, a generative adversarial architecture optimized via discrete relaxation to explore high-dimensional payload patterns independently of syntax rules; and finally, a closed-loop verification mechanism that performs syntactic calibration and functional validation against simulated device responses. Evaluations on four real-world protocols (Modbus TCP, S7Comm, Omron FINS, and DNP3) demonstrate that FIGAN significantly outperforms state-of-the-art baselines. The source code has been open-sourced https://github.com/MissHP111/FIGAN.	10.1109/TNSM.2026.3717268
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
Wei Sai, Yihui Lu, Xin Guo	A Privacy-Preserving Security Framework for Multi-Party Data Fusion Computing Based on Homomorphic Encryption	2026	Early Access	Security Protocols Information rates	To prevent plaintext exposure in multi-party collaborative computing, this paper proposes a distributed secure multi-party computation protocol based on the Cheon-Kim-Kim-Song (CKKS) homomorphic encryption scheme. Data is encoded and encrypted at the source into CKKS complex polynomial ciphertext, enabling vectorized fusion under shared evaluation keys and threshold decryption in a decentralized architecture without a trusted central authority. Experiments on heterogeneous multi-institution datasets demonstrate low numerical error (9.0×10⁻⁷ at polynomial order 2¹⁶ and depth 12), effective scalability (throughput increasing from 1.12×10⁵ to 1.32×10⁵ ops/s and latency decreasing from 56 ms to 38 ms as nodes scale from 4 to 16), and strong robustness (70% decryption success at a 60% threshold and 95% recovery under malicious interference), showing that the framework achieves efficient computation with strict privacy protection for cross-party data fusion.	10.1109/TNSM.2026.3717343
Mengdi Liu, Mingwei Lin, Shenbao Yu, Riqing Chen, Xin Luo	A Lightweight and Robust Low-rank Tensor Embedding-integrated VAE for Network Anomaly Detection	2026	Early Access	Modeling Tensors Anomaly detection	As networks scale exponentially, robust anomaly detection has become critical. Although numerous detection algorithms have been proposed, most rely on linear matrix-based models that fail to effectively capture multidimensional structural characteristics (including temporal patterns, node locations, and more) or the underlying nonlinear relationships. Moreover, periodic temporal patterns are often overlooked, degrading detection performance. To address these limitations, we propose a novel Tensor-based Embedding Variational Autoencoder (TEVAE) model. TEVAE introduces a tensor representation framework that integrates deep temporal feature extraction with probabilistic reconstruction. It employs bidirectional long short-term memory networks to distill temporally enriched latent representations from contaminated network tensors. These representations are processed through a probabilistic variational framework, enabling robust learning of latent embeddings and high-fidelity reconstruction of denoised outputs. To further reduce computational demand, we propose LightTEVAE. It decomposes the network tensor via dimension preserved tensor decomposition into low-rank factors, which are embedded as compact representations for encoding, thereby reducing complexity while preserving structural integrity. Experimental results on public network datasets demonstrate that: 1) TEVAE achieves superior anomaly detection performance, consistently outperforming nine main-stream methods by a clear margin; 2) LightTEVAE effectively maintains this high detection performance while substantially reducing computational overhead. These results demonstrate the superiority of both proposed models in temporal feature extraction, complex pattern learning, and large-scale processing efficiency for network anomaly detection.	10.1109/TNSM.2026.3716561
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
Qing Wu, Xijia Dong, Leyou Zhang, Yue Lei, Zilong Yan	Cloud-Assisted Verifiable and Updatable Private Set Union Protocol for Enhancing Network Intrusion Detection	2026	Early Access	Protocols Clouds Security	As network intrusion detection systems (NIDS) play an increasingly critical role in large-scale network environments, multiple organizations, Internet Service Providers (ISPs), and security service providers often maintain independent IP blacklists. Due to the dynamic nature of malicious IP addresses and their cross-organizational propagation, inter-organizational blacklist sharing is essential for improving network intrusion detection. However, traditional blacklist exchange mechanisms risk exposing participants’ complete blacklist information, and curious organizations may infer another organization’s detection strategies from the shared IP intersection, leading to privacy breaches.To address this issue, this paper proposes a Cloud-Assisted Verifiable and Updatable Private Set Union (CVU-PSU) protocol, which leverages the multi-query Reverse Private Membership Test (mq-RPMT) protocol and Oblivious Transfer (OT) technology to ensure privacy-preserving inter-organizational blacklist sharing. The protocol utilizes cloud computing to reduce the computational and communication overhead of participants in the mq-RPMT protocol while incorporating a verification mechanism to ensure the correctness of the cloud’s returned results. Furthermore, the protocol supports real-time blacklist updates, enabling adaptation to rapidly changing malicious IP addresses.Experimental results demonstrate that the proposed protocol achieves efficient inter-organizational blacklist sharing with low communication and computational costs while preserving privacy, thereby enhancing the real-time performance and accuracy of network intrusion detection systems.	10.1109/TNSM.2026.3716071
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
Zening Li, Pin-Han Ho	Topology-Constrained Generative Modeling for Performance Monitoring and Structural Anomaly Definition in Optical Transport Networks	2026	Early Access	Windows Modeling Semantics	Electrical-layer performance monitoring (PM) in optical transport networks is reported as window-aggregated counters, under which propagation delay and signal regeneration boundaries are not directly observable. Consequently, different impairment mechanisms can lead to statistically similar PM trajectories, leaving PM-only methods without a principled notion of nominal behavior or anomaly detectability. To address this limitation, this paper proposes Topology-Constrained Partially Observed Dynamical System (TC-PODS). TC-PODS defines a topology-consistent nominal predictive reference directly in PM space by introducing a latent impairment state (LAT) that enforces service-path propagation and regeneration semantics as supervisory constraints, while explicitly modeling the irreversible projection induced by windowed PM observation. This formulation disambiguates nominal variability from anomalous behavior under partial observability and defines anomalies as PM trajectories that are incompatible with any topology-consistent nominal generation. By shifting PM analysis from algorithm-centric detection toward detectability-aware inference under the deployed observation protocol, TC-PODS provides a principled framework for forecasting and anomaly interpretation under partial observability, with future applications to structure-aware failure management.	10.1109/TNSM.2026.3715222
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
Kaifei Peng, Yanbiao Li, Wenbin Li, Yuxuan Chen, Xian Yu, Xin Wang, Bo Pang, Gaogang Xie	Rethinking Virtual Network Construction for Network Emulation at Scale: Analysis, Modeling, and Optimization	2026	Early Access	Virtual machines Memory Timing	Network emulation has become an indispensable methodology for evaluating next-generation network architectures, offering a critical balance between experimental fidelity and operational scalability. However, its effectiveness is fundamentally constrained by inefficiencies in emulating large-scale networks, particularly during virtual network construction. This bottleneck arises from mandatory serialization of virtual link instantiation and operating system (OS) kernel-level notification overheads, which collectively degrade performance by orders of magnitude on 10K-node topologies. Departure from the current practice that employs a multi-machine framework for improvements, we propose SplitNN (Split Network and Namespace), a novel single-machine network emulation paradigm that breaks the serialization constraint through multi-VM (virtual machines) partitioning, and reduces notification overheads via namespace segmentation. Extensive evaluations show that SplitNN constructs 10K-node virtual networks within 1–5 minutes on a single machine, achieving a 98.5%–99.2% reduction in construction time compared to state-of-the-art emulators. While primarily a single-machine solution, SplitNN seamlessly integrates with multi-machine deployments, complementing them by enabling cumulative gains in both scalability and efficiency.	10.1109/TNSM.2026.3715559
S A Harish, S Vignesh, Divya Pathak, Anil Kumar Sharma, Praveen Tammana	Anomaly Detection in In-Network Fast ReRoute Systems	2026	Early Access	Fluid flow Planing Delays	High-speed programmable data planes provide opportunities to implement data-driven fast reroute systems that quickly adapt to varying network conditions (e.g., congestion, failures) and improve network performance. The core of these systems has packet-processing algorithms running in the data plane that continuously look for traffic patterns (e.g., too many retransmissions) specific to a network condition (e.g., link failure) and take appropriate action (e.g., reroute). Despite their benefits, they also increase the potential attack surface. Adversaries can generate malicious traffic patterns resembling those anticipated by a fast reroute system and trick the system. Doing so would lead to poor network performance due to incorrect reroute decisions. In this paper, we propose a mechanism to detect whether the fast reroute systems are under the influence of malicious traffic patterns. Our key idea is to model the expected behavior using benign traffic features and use the model as a reference to determine whether the system is under the influence of adversaries. Using realistic attack traces, we demonstrate attacks on two fast reroute systems and successfully detect those attacks using the proposed detection mechanism.	10.1109/TNSM.2026.3715353
Abdullatif Albaseer, Elmahdi Bentafat, Mohamed Abdallah, Saif Al-Kuwari, Marwa Qaraqe	Incentive-Driven Honeypot Defense: A Multi-Agent DRL Framework for Securing Smart Grid Networks	2026	Vol. 23, Issue	Pricing Timing Learning (artificial intelligence)	Honeypot defenses present a robust solution for securing the Advanced Metering Infrastructure (AMI) against sophisticated cyberattacks. The efficacy of AMI defenses relies on the strategic deployment of honeypots by Small-scale Power Suppliers (SPSs) and the subsequent exchange of defense data with Traditional Power Retailers (TPRs). However, existing methods are limited by their requirement for prior information exchange and their inability to specify targeted services (i.e., protocols), making them impractical for dynamic environments. In addition, previous approaches have predominantly overlooked critical aspects such as service allocation and competition among SPSs. To address these challenges, we propose a novel Deep Reinforcement Learning (DRL)-based Stackelberg leader-follower game framework that facilitates tailored service allocation and fosters competition among SPSs. We leverage the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm, which combines centralized training and distributed execution, to enable each SPS to autonomously learn optimal strategies that consider defense data quality and attack rates, without requiring prior knowledge of deployment costs or the actions of other SPSs. This adaptability minimizes the overhead associated with information exchange and improves the defense of vulnerable services. In particular, our method operates by requiring data collection, as each SPS functions as an independent learning agent that generates its own training experiences. Extensive simulations demonstrate that our proposed DRL-driven approach significantly outperforms baseline methods in different performance metrics. It achieves higher utility for both SPSs and TPR while maintaining lower operational costs.	10.1109/TNSM.2026.3715400
Tong Li, Shicheng Wei, Wencheng Yang, Yan Li	HotPatchCaps: A Capsule Network With Runtime Hot Patching for Zero-Day API Attack Detections	2026	Vol. 23, Issue	Modeling Application programming interfaces Security	Modern services are awash in Application Programming Interfaces (APIs), yet most security pipelines end at pre-deployment testing using fuzzers and scanners. This leaves a run-time 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