In heterogeneous wireless emergency networks, network fault diagnosis plays a critical role in ensuring reliable and secure communication. To improve network transmission quality, the complexity of network equipment-both in hardware and software design-has increased, which inevitably gives rise to equipment failures with complex root causes, significantly elevating the difficulty of fault diagnosis. Current fault diagnosis algorithms are inadequate for addressing the challenges in fault diagnosis of complex emergency access networks, primarily due to their high diagnostic costs and low accuracy. In this study, we first propose a diagnosis framework and a deterministic fault propagation (DFP) model, and a hierarchical fault diagnosis framework (HFDF). Second, we develop three algorithms to construct a fault cause relationship graph (FCRG), which supports identifying the logical relationships among various fault causes associated with a specific fault. Third, we propose a fault diagnosis algorithm based on relational graph inference (FDRGI). Finally, we conduct extensive experiments in real-world wireless access networks. The experimental results demonstrate that our algorithm satisfies the requirements for root cause diagnosis of access failures in emergency networks, and outperforms other comparative algorithms in terms of diagnostic cost and accuracy: it reduces the average diagnostic cost by 13.71%-69.88% and improves the average diagnostic accuracy by 39.06%%-1.98-fold.

In this paper, we propose a novel dependency-aware task scheduling strategy for dynamic unmanned aerial vehicle-assisted connected autonomous vehicles (CAVs). Specifically, different computation tasks of CAVs consisting of multiple dependency subtasks are judiciously assigned to nearby CAVs or the base station for promptly completing tasks. Therefore, we formulate a joint scheduling priority and subtask assignment optimization problem with the objective of minimizing the average task completion time. The problem aims at improving the long-term system performance, which is reformulated as a Markov decision process. To solve the problem, we further propose a diffusion-based reinforcement learning algorithm, named Synthetic DDQN based Subtasks Scheduling, which can make adaptive task scheduling decision in real time. A diffusion model-based synthetic experience replay is integrated into the reinforcement learning framework, which can generate sufficient synthetic data in experience replay buffer, thereby significantly accelerating convergence and improving sample efficiency. Simulation results demonstrate the effectiveness of the proposed algorithm on reducing task completion time, comparing to benchmark schemes.

We propose a robust website fingerprinting recognition method, termed FRF (Fluctuation-Resilient Fingerprinting), which aims to enhance the stability and accuracy of regulatory recognition under multiple traffic disruption conditions. FRF constructs a robust traffic representation, namely the Traffic Sliding Window Matrix, to reliably extract discriminative information from network traffic. It further integrates a hybrid CNN-BiGRU architecture to capture key features from the traffic matrix for classification, while incorporating data augmentation to improve the model’s generalization capability. Extensive experiments are conducted in both closed-world and open-world scenarios, with systematic comparisons against several state-of-the-art website fingerprinting recognition methods. The results show that, in the closed-world setting, FRF achieves an average accuracy improvement of approximately 1.57% over the best existing method, RF, under real traffic conditions, and a further improvement of about 2.56% under disruption, confirming its robustness across diverse disruption environments. Moreover, in the open-world setting, FRF consistently outperforms competing methods in terms of precision–recall performance, a widely recognized and effective evaluation metric, across all decision thresholds, indicating stable and reliable recognition under realistic network conditions with multiple traffic disruptions.

While improving the convenience of IoV applications, its communication process faces serious privacy and security threats. Especially in the process of network layer transmission, the analysis of side-channel characteristics (such as packet size and traffic transmission mode) may lead to user privacy leakage. Existing research mostly focuses on website fingerprint camouflage or overall traffic statistical feature camouflage, and usually only designs a camouflage mechanism for a single application layer protocol, which is challenging to achieve unified protection in the multi-protocol scenario of the Internet of Vehicles. This paper proposes LeakGAN Packets Traffic Camouflage (LPTC), which simultaneously supports traffic camouflage for multiple application-layer protocols. The LPTC learns the sending pattern of target traffic packets and generates camouflaged traffic through an adversarial game between a generator and a discriminator, to effectively resist side-channel attacks without affecting communication performance. Experimental results indicate that the proposed LPTC scheme significantly degrades the classification performance of traffic transmission models based on K-Nearest Neighbors, Support Vector Machine, Bagged Trees, and Random Forest algorithms, reducing classification accuracy by 41.23%. In addition, the precision, recall, F1 score, and AUC value decreased by 41.82%, 61.85%, 49.57%, and 42.00%, respectively. It also defends against adaptive attackers who retrain deep learning classifiers on camouflaged traffic. The LPTC substantially enhances the indistinguishability between target and camouflaged traffic, thereby effectively mitigating user privacy leakage caused by side-channel analysis. Meanwhile, the system’s bandwidth and delay performance is sufficient to support the transmission requirements of real-time vehicular networks.

Traditional key agreement-based encryption mechanisms in public security communications struggle to address increasingly sophisticated network attacks and real-time threats due to infrequent key updates and exposure risks. To overcome these limitations, this study proposes a chaotic encryption-based tunneling method (CETM) at the network layer. CETM integrates a 6-D hyperchaotic system with the advanced encryption standard with 256-bit key length (AES-256) algorithm, leveraging the system's extreme sensitivity to initial conditions to generate high-strength key sequences, eliminating the need for traditional negotiation. A self-synchronizing key update mechanism is also introduced, allowing both parties to update keys automatically without network transmission, thereby eliminating the risk of key leakage. The proposed system demonstrates strong complexity and randomness, as validated by multiple chaotic metrics including the Lyapunov exponent, Shannon entropy, correlation coefficient, permutation entropy, and key space. The generated keys pass both the SP800-22 and FIPS 140-2 tests, meeting established security standards. Performance evaluations across local, cloud, and satellite environments show that CETM significantly outperforms conventional methods in terms of bandwidth efficiency, latency, jitter, and packet loss.

The Vehicle-to-Ground (V2G) emergency communication network is a dedicated network established to respond to emergencies, such as natural disasters and traffic accidents, and it plays a crucial role in ensuring the safe and smooth operation of vehicles. Composed of numerous devices, this network is inevitably exposed to failure risks due to prolonged operation, complex designs, and insufficient management and maintenance. Faults in network nodes may undermine the reliability of vehicle-to-ground communication. Rapid fault localization is critical to the maintenance and management of network device. However, current localization methods face issues like excessively long probing paths, high localization costs, and low accuracy—all of which lead to subpar performance in real-world fault localization scenarios. To address these problems, we introduce a novel Multi-stage Group Probe (MGP) localization method, designed to balance localization cost and accuracy effectively. Specifically, we first present a network localization model and the concept of "uncertain information volume of network node states," which quantifies the cost and efficiency of localization. Second, leveraging graph theory, we propose the idea of network probing subgraphs and constrain the number of probing stations and probe lengths, while developing algorithms for selecting probing stations and planning probing paths. Additionally, we introduce a group probe localization method that incorporates information feedback to reduce costs. Finally, we evaluate the MGP against other probe localization approaches across different networks. Experimental results demonstrate that MGP outperforms comparative methods in terms of localization cost, accuracy, and efficiency.

In the context of multi-cloud networks, enhancing transmission efficiency while maintaining security in communication relations is of paramount importance. Existing methods have been criticized for imposing high encryption overhead on the sender side and lacking effective short-term dynamic modeling of network performance. This paper proposes an optimization method that integrates a trusted obfuscation forwarding framework with the joint spatio-temporal network performance prediction. Firstly, to conceal communication relationships, we design a hop-by-hop encryption and proxy forwarding framework leveraging obfuscated forwarding nodes. Second, node selection is optimized by predicting network performance metrics. We extract the spatial correlations of the topology using a Chebyshev Network (ChebNet), and integrate this with Long Short-Term Memory (LSTM) models to capture temporal variations in performance. Experimental results show that the proposed method reduces the probability of communication relationship exposure to 3.57% from 100%, approaching the theoretical optimum. The prediction accuracies of bandwidth, delay, jitter and packet loss rate are significantly improved in terms of mean absolute error, root mean square error, and mean absolute percentage error metrics when compared to the graph convolutional network (GCN) and temporal convolutional network (TCN) benchmark models. Moreover, the node selection probability during obfuscated forwarding approaches a uniform distribution. The proposed method distributes the encryption load across intermediate nodes, thereby reducing the burden on the sender.

Remote driving is crucial for the commercialization of autonomous driving. However, technological and network limitations typically compromise communication reliability in remote driving, thereby undermining control safety and precision. We propose an asymmetric reliability-enhanced (ARE) scheduler based on heterogeneous cellular networks. ARE employs targeted algorithms for uplink and downlink transmissions, which are tailored to their respective traffic characteristics and service requirements. For uplink data, we design an adaptive minimum round-trip time (AE minRTT) scheduling algorithm that leverages buffer overflow awareness to avoid congestion during transmission. For downlink data, given the low traffic volume but high importance, the proposed scheduler uses limited bandwidth to reduce packet arrival latency and packet loss. The simulation results demonstrate that compared with baseline algorithms, AE_minRTT increases uplink aggregated bandwidth by 29%-100%. The field experiments further demonstrate a 14-20 point improvement in video multimethod assessment fusion scores for uplink video transmission and a 10%-20% reduction in downlink one-way latency. These results demonstrate the effectiveness of the proposed scheduler in enhancing reliability in remote driving.

Multi-interface servers can achieve high bandwidth and reliable transmission through multipath transmission and bandwidth aggregation technologies. These capabilities meet the transmission quality requirements of emergency communication systems. However, it is important that multi-interface servers come with the risk of single-point failures, which significantly limit the reliability of emergency communication systems. Currently, existing solutions encounter challenges in managing the single-point failures of multi-interface servers. In this paper, we analyze the complexity of emergency communication systems, and propose a novel intelligent disaster recovery mechanism called MultiS-IDRM to achieve intelligent disaster recovery for multi-interface servers for the first time. Firstly, we propose a multi-dimensional attribute joint decision model (MDAJD) as the basis for determining the working status of multi-interface servers. Secondly, we design a route encapsulation method applicable to multipath backup scenarios. This method achieves the failover of independent paths with mixed multiple network tunnels. Additionally, we design a server status detection algorithm and an intelligent switching mechanism to realize intelligent switching of multi-interface servers in different application scenarios. Finally, we compare MultiS-IDRM with other protocols, including the Virtual Router Redundancy Protocol, Gateway Loading Balance Protocol, and Hot Standby Redundancy Protocol, in real networks. The experimental results demonstrate that MultiS-IDRM effectively resolves the single-point failure issue of multi-interface servers while enhancing the reliability, invulnerability, and intelligence of emergency communication systems.

Multipath transmission enables path diversity and improves network resource utilization. However, suboptimal path selection can significantly increase the risk of degraded multipath transmission ability due to the inherent notable differences across multiple heterogeneous paths and the frequent jitter during vehicular mobility. To address this issue and achieve collaborative transmission enhancement in such heterogeneous jitter environments, this paper proposes a delay-difference-constrained packet scheduler (DDCPS) that incorporates both network risk and benefit. Specifically, network risk is quantified by the delay differences and variance, whereas network benefits encompass both the transmission ability and application adaptability. The DDCPS prioritizes paths with the lowest residual risk by minimizing the difference between network risk and benefit. We evaluate the DDCPS across six key dimensions, including a real-world IoV test, to ensure a comprehensive assessment. The results demonstrate that the DDCPS achieves an improvement of 44% in latency stability while simultaneously achieving a 4.81% increase in transmission ability compared to best-performing scheduler.

With the improvement of intelligence level and the innovation of surveillance technology, multipath technology has been widely used in real-time HD video surveillance because of its reliability advantage to ensure the critical demand of high reliability and low latency data transmission for train operation safety. However, traditional multipath scheduling often results in video quality degradation due to the presence of out-of-order packets (OFOs) when dealing with mobile and asymmetric network environments. Therefore, this paper proposes an algorithm called “Adaptive Redundancy and Buffering Scheduler (ARBS)”, which aims to improve the reliability of video transmission by dynamically adjusting the transmission policy of redundant packets and the buffer size at the receiving end. The method is implemented in the Zhirong logo network and tested with existing scheduling algorithms in a real scenario simulation. The experimental results show that ARBS reduces the OFO rate to 10.37% of the original level, and the video low latency percentage is higher than the rest of the schedulers by 14.68% to 49.14%.

In recent years, the Internet industry has been developing rapidly, and various mobile terminal network applications such as short videos and live broadcasts have emerged rapidly, with a sharp increase in the number of users, leading to a rapid growth in network bandwidth demand. Considering that smartphones and other smart terminal devices currently provide multiple wireless network access methods such as WIFI, LTE, 5G, satellite, etc., multipath parallel transmission technology has become an ideal solution. However, most of the current researches tend to refine the scenarios and design the scheduling and transmission mechanisms for specific scenarios, and as the scenarios continue to be refined, the adaptability of the scheduling algorithms decreases, and the avalanche effect of the system caused by the degradation of some links that exists under the heterogeneous wireless networks is not taken into account. Based on the analysis of the above problems, this thesis designs a scheduling and transmission mechanism that can adapt to the changes in transmission traffic while avoiding the avalanche effect under heterogeneous network environments. Compared to the classic minRTT algorithm, the overall transmission delay is reduced by 30% in the test environment, and the transmission bandwidth and video transmission effect are both greatly improved.

Multipath cooperation technology alleviates transmission pressure by leveraging path diversity. However, in next-generation service-oriented environments with computation-intensive services, the limited computing capability of transmission paths can degrade end-to-end service quality, even when bandwidth is sufficient. This issue becomes more pronounced in dynamic mobile scenarios, where fluctuating link status and computational resources introduce new challenges in path selection. To address these challenges, we propose a novel path selection approach that jointly considers both network and computation constraints for computation-intensive services. First, we construct a computation-integrated multipath transmission framework to support real-time monitoring of both link-level computing capabilities and network conditions. Second, we introduce a packet structure embedding device identifiers and computing capability, enabling adaptive scheduling. Finally, we develop a computing capability-constrained delay-minimizing packet scheduler (C2-DMPS) to balance bandwidth and computational load, ensuring low-latency transmission for emerging service demands. The results demonstrate the critical role of computational capacity in maintaining service performance, especially under volatile network conditions, highlighting potential risks to service continuity in next-generation environments.

Integrated Vehicular Networks (VANETs) constructed through the collaboration of various heterogeneous networks, such as 4G, 5G, satellite networks, and Unmanned Aerial Vehicle (UAV) networks, provide an effective solution to the resource constraints between vehicles and edge fog computing nodes. Reordering Buffer (RB) is crucial in concurrent data transmission between vehicles and edge fog computing nodes via heterogeneous VANETs. RB is in charge of storing out-of-order packets, waiting for packets with smaller sequence numbers, and delivering in-order packets to upper-layer applications. However, current packet reordering mechanisms are challenging in providing stable and high goodput due to the inappropriate timeout timers and uneven delivery rules. In this paper, we propose a PPO-assisted packet reordering mechanism (PPORM) to achieve optimal control of packet delivery. We first transform the goodput maximization problem into the optimal timeout threshold of RB and the optimal delivery moment of each packet. Secondly, we introduce a Proximal Policy Optimization-assisted Timeout Threshold Updating (TTU) algorithm to dynamically adjust the threshold in response to real-time changes in network conditions. Further, we present a Multifactor Smooth Delivery (MSD) algorithm to regulate the optimal queuing delay for each packet and enhance the stability of the real-time throughput as much as possible. Experimental results show that PPORM improves goodput by 6.94%∼45.57% and improves stability by 32.5%∼49.58% compared with other baseline algorithms.

The digital age has promoted the growth of the demand for a large number of computing resources, and the computing power network was born. Through distributed architecture, the computing power network can cover clouds, edges, and terminals to meet different levels of computing needs. Then, in reality, the network resources are complex, and there are many high-concurrent transmission tasks in the computing network. The OSPF routing algorithm adopted in the traditional network only supports multiple transmission tasks to seize the bandwidth resources of a single link, so it is difficult for users to flexibly choose computing resources according to the task requirements. Therefore, in order to achieve the heterogeneity of network resources, we propose a cross-network fusion scheduling mechanism for high-concurrent transmission tasks in the computing network. By sensing the performance parameters of multi-link heterogeneous networks, we assign a link with relatively abundant bandwidth and relatively short time to complete all transmission tasks to the new task flow for transmission, so as to ensure the bandwidth demand of high concurrent transmission tasks. Finally, the test results show that this method can make full use of network resources, and the throughput of each task flow is improved by more than 100 Mbps, and the transmission time of each task is reduced by more than 5 s.

Medium and large enterprises need to exchange confidential information frequently. Traditional encryption techniques can only protect the security of communication data, but the privacy of the communicating entities' identity information cannot be effectively protected. Currently, common solutions include virtual private networks (VPNs) and single/multiple agent-based forwarding mechanisms. However, the egress nodes of VPNs are susceptible to attacks and monitoring; agents in single-agent forwarding mechanisms are easy targets for single points of failure and attacks. Medium and large enterprises urgently need a solution that ensures communication privacy while being cost-effective and easy to manage. Therefore, this paper proposes a lightweight privacy communication network system based on a multi-agent mechanism to realize the non-association of communication parties and provide transparent transmission services. In order to schedule and manage the whole system flexibly, this paper implements a remote service and a visualization service in a Directory Server. In addition, an intelligent multi-objective optimization analysis of the route selection algorithm is introduced to ensure the anti-tracking and high efficiency of data transmission; meanwhile, traffic obfuscation technology is used to prevent attackers from recognizing the communication content and the identities of both parties through the traffic pattern, which improves the transmission and management security of the system. Test results show that the system effectively protects information communication relationships in medium and large enterprises.

In this paper, we investigate a deterministic transmission scheduling problem for satellite-terrestrial integrated networks (STIN), in which satellite networks supplement terrestrial networks endowed with deterministic mechanisms (e.g., cycle specified queuing and forwarding) to improve scheduling success ratio and reduce overall transmission delay (i.e., network revenue) while maintaining users’ quality of service (QoS). We propose a fixed-mobile-satellite integrated architecture and formulate the transmission scheduling problem as a two-hierarchical routing and queuing problem. Then, a deep reinforcement learning-based Transient Routing And Varied quEue aLgorithm (TRAVEL) is developed to address the two-hierarchical decision problem. Specifically, according to the intrinsic properties of routing and queuing, we decouple the decision problem into two sub-problems, namely route planning at the macro level and queue selection at the micro level. The proposed TRAVEL can realize efficient decision making to perform the differential transmission scheduling of intricate tasks. Simulation results demonstrate that the TRAVEL delivers robust and effective performance, thereby enhancing network operation revenue in terms of different traffic proportion.

With the continuous development of network technology, people are trying to build various dedicated networks for encrypted communication and emergency communication scenarios. Routing devices are essential core devices in the use of dedicated networks, and monitoring their working status, diagnosing and repairing anomalies, is crucial to ensure rapid communication recovery and enhance the stability of communication systems. However, as people's demand for communication continues to increase, network devices are gradually developing towards complexity, customization, and specialization. The structure and function of network devices may vary greatly in different scenarios, posing significant difficulties for users and maintenance personnel in diagnosing network faults. Therefore, in order to assist decision-making and achieve timely and accurate fault diagnosis and maintenance of network equipment, a knowledge graph based fault diagnosis mechanism is proposed. Based on various types of fault knowledge of network devices, construct a knowledge graph, locate possible causes and possibilities from the knowledge graph based on abnormal phenomena, and provide corresponding solutions for each cause.

Low Earth Orbit (LEO) satellite networks have recently been regarded as a promising solution to provide ubiquitous user access and flexible content delivery. To meet the increasing demands of satellite services, multipath routing is leveraged to transmit each flow via multiple transmission paths concurrently. However, dynamically changed satellite topologies have brought challenges to multipath route planning and traffic splitting among different paths. Thus, in the paper, we propose a GNN-enabled Multipath Routing (GMR) scheme to maximize the network efficiency. Particularly, we firstly formulate the mutipath routing issue as a stochastic optimization problem under the bandwidth limitation and flow conservation constraint, and further present a Link Disjoint Multipath Routing (LDMR) scheme for the central route calculation, where available transmission paths and uneven traffic density are jointly considered to maximize the link utilization. Then, to balance the load distribution of different paths, a GNN-based MultiPath Traffic Engineering (GNN-MPTE) algorithm is designed for dynamic flow splitting based on estimated path quality. Finally, we implement the proposed GMR scheme and corresponding algorithms in Network Simulator NS-3 and make comparisons with other benchmarks. Simulation results demonstrate that the proposed GNN-MPTE can be extended to arbitrary inclined/polar orbit constellations without repetitive training, and significantly prompt the transmission quality including average delay, throughput, and flow completion ratio.

Multipath transmission is a critical enabling technology to enhance QoE for edge users. The packet scheduler plays an irreplaceable role in overcoming heterogeneity and dynamicity in multipath transmission. However, current schedulers depend on an inaccurate delay estimation and lack systematic traffic intensity awareness, performing poorly in wireless heterogeneous networks (HetNets). In this paper, we propose a novel traffic-aware two-level packet scheduler (TA2LS) to address the problem and improve aggregated bandwidth while trading off delay. In particular, we design a multipath transmission state machine (MTSM) to perceive link traffic intensity. MTSM replaces network prediction algorithms by identifying the contribution of each link in multipath transmission in a cost-effective way. Further, we propose a scheduling mechanism based on a two-level optimal-path evaluation method (2LOSM) to adjust the packet scheduling policy adaptively. 2LOSM increases the priority of links with low traffic intensity during scheduling, improving aggregated bandwidth performance and reducing end-to-end delay. We have built a real-world 4G/5G/WiFi testbed and deployed 47 dynamic scenarios to evaluate TA2LS and other five schedulers. In 4G/5G/WiFi scenarios, TA2LS improves aggregated bandwidth by 10.32%-48.27% compared to the second-best scheduler and reduces end-to-end delay by 5.04%-39.98% under the premise of fewer or equivalent overheads.