Congestion control is a cornerstone of reliable data transport, but current fixed-rule algorithms struggle with the diverse link characteristics of Internet of Things deployments. A more critical challenge is that many resource-constrained Internet of Things devices lack the computational power to run advanced, data-driven methods locally, hindering performance and adaptability. To address this, we introduce KONTROL, a service framework that decouples congestion control intelligence from end devices by offloading the decision-making logic to a centralized congestion control server. This enables lightweight clients to leverage sophisticated control strategies without bearing the computational burden. As a critical instance within our framework, we implement Deep Reinforcement Learning agents on the server, which learn to optimize window adjustments for each sender based on real-time relayed network statistics via kernel modifications. Our experimental evaluation across challenging simulated environments shows that the proposed scheme achieves consistently high throughput and low latency while maintaining fairness. These results validate the viability of the offloading paradigm, paving the way for more flexible transport protocols for the broader Internet of Things ecosystem.

Future Internet of things (IoT) networks will host applications that involve data collection and computation tasks on one or more servers. To this end, this paper proposes the first mixed integer linear program (MILP) to schedule and embed applications on energy harvesting nodes, where it optimizes (i) the sampling time of devices, (ii) whether to run an application, and (iii) the energy usage of devices, gateways and servers. To ensure applications are run often, we adopt the maximum age of service (AoS) metric, and set the MILP’s objective to minimize the maximum AoS or min-max AoS of applications. This paper also proposes two novel solutions: (i) a receding horizon control (RHC) based method, and (ii) a solution that greedily embeds applications according to their AoS. The results show that the min-max AoS of RHC and greedy approach is respectively 1.07x and 1.13x higher than MILP.

The growing device intelligence and distributed apps of the Internet of Things (IoT) across multiple fields have caused a sharp rise in wireless network traffic, posing more intense uplink resource competition and traffic scheduling optimization challenges for traditional wireless networks. To address these challenges and enhance the network’s load-bearing capacity and efficiency, this paper proposes a dynamic load-aware Steiner traffic aggregation algorithm (DLSAG), which integrates load awareness with a multi-candidate subtree strategy and employs a neural network to accelerate the computation process, thereby efficiently aggregating data flows. Theoretical analysis demonstrates that the time complexity of DLSAG is significantly reduced compared to traditional algorithm. Through extensive experiments across varying network scales and four baseline algorithms, it has been shown that DLSAG can reduce maximum link utilization by approximately 4.4% to 28.8%, while achieving a 1.5% to 10.2% improvement in Packet Delivery Ratio (PDR) and maintaining a competitive end-to-end delay, which is only about 7.4% to 10.3% higher than the optimal algorithm.

With the rapid integration of the Internet of Things (IoT) and fifth-generation mobile communications (5G), the massive real-time computing demands generated on the terminal side have exceeded the processing capability of a single edge server (ES). How to efficiently offload computing tasks in parallel to multiple ESs for collaborative execution has emerged as a significant challenge in mobile edge computing (MEC). To address this, we propose a dynamic parallel offloading framework, CoESAC (Collaborative Edge with Soft Actor-Critic). On the one hand, we model the multi-objective offloading problem and the load allocation problem as a high-dimensional discrete decision-making task, demonstrating its intrinsic NP-hard complexity. On the other hand, based on a discrete Soft Actor-Critic (SAC) algorithm in deep reinforcement learning (DRL), the proposed method adopts a task sub-fragmentation and dynamic load adaptation mechanism to flexibly schedule the parallel computing capabilities of multiple ESs. Experimental results show that across diverse system conditions and execution scenarios, CoE-SAC reduces the average make-span by up to 50.16% compared with advanced baselines, and significantly lowers the failure rate by more than 30.73%. These improvements highlight the framework’s strong robustness and superior performance, offering new insights into multi-node collaborative computing under heterogeneous resources and high-concurrency workloads.

Federated Learning has been widely adopted in privacy-sensitive distributed machine learning. However, in heterogeneous scenarios where significant differences exist in user data distribution, computational capabilities, or network conditions, existing models often underperform. This paper models key metrics such as communication delays and employs an integer programming approach to select participating devices for training intelligently. Additionally, an efficient federated learning algorithm, AdapFed, is proposed. In the early stages of training, AdapFed prioritizes updates from devices with rapidly changing gradients, while later rounds involve a broader range of diverse devices. The algorithm also incorporates a constraint on the average waiting time among device sets throughout the training process. Experimental results on three public datasets demonstrate that, compared to five baseline algorithms, the proposed framework improves model accuracy by 6.4% within the same number of training rounds and reduces the average waiting time among users by 71.4%.

As advancements in intelligent infrastructure and machine learning technologies continue, implementing congestion control using intelligent algorithms can significantly improve network flow management, ensuring a seamless user experience. However, both extending the intelligent congestion control framework and its deployment on existing devices present practical challenges. In this paper, we introduce a reconfigurable approach to enhance congestion control in the kernel using deep reinforcement learning. We first modify the congestion control logic and establish a connection to an extensible online learning framework in user space. Using this framework, we augment the congestion control state model with a Markov Decision Process and introduce an intelligent congestion control scheme called KerDqn, which combines state transitions and utility. Experiments indicate that under conditions of high packet loss, KerDqn's goodput remains unaffected, while both Cubic and Vivace experience severe goodput losses. Under low buffer conditions, KerDqn achieves a throughput 1.6 times that of Cubic and 1.2 times that of Vivace at a 15KB buffer size. In high buffer scenarios, KerDqn exhibits low latency without bufferbloat. In terms of TCP friendliness and fairness, KerDqn ranks highest.

With the application of new network technologies and hardware, exploring patterns of large-scale latency variations is a long-term endeavor crucial for characterizing wide-area network states. Wide-area network measurement schemes focus on link Quality of Service (QoS) metrics, and the measured data are essential for network performance management, SLA verification, planning, and optimization. However, in the complex environment of wide-area networks, mature one-way delay measurement schemes face challenges such as inaccurate clock synchronization and network security policy restrictions. Existing research lacks precise quantification and extensive data support for the dynamics of one-way delay variations. In this paper, we propose a correlation analysis framework based on relative latency measurement. By customizing Network Time Protocol (NTP) packet transmission and establishing a large-scale packet parsing process, we selected 5003 global target servers, gathered over 1TB of network dataset, and conducted correlation studies between one-way delay, hop count, and geographical location. The experimental results indicate that, under identical hardware conditions, detection speed has been improved by at least 10%. In the correlation study, counter-intuitively, it was revealed that 91.2% of measurement points showed a non-positive correlation between relative latency and routing hop count. Additionally, we found that the correlation with average reverse latency is even more significant, reaching up to 33% in some cases. The experimental data is available at https://github.com/Network-Optimization/Relative-delay-network-measurement-data.

Ensuring quality of service (QoS) for data transmission in dynamic network environments presents a significant challenge. To solve the conflict between complex network conditions and strict transmission quality, effective solutions employing artificial intelligence have recently been introduced. Nonetheless, the prevalent methods primarily emphasize throughput oriented designs, concentrating on data’s downlink transmission while overlooking the upload of delay sensitive data during bidirectional interactions. Resource-constrained mobile devices struggle not only with executing long-term and continuous online computing tasks but also with rapidly adjusting to the constantly changing wireless networks. Models running on mobile devices must adapt to the changing wireless links. Diverse application demands and limited hardware capabilities pose significant obstacles to deploying relevant models.Consequently, we introduce a novel congestion control scheme named Colla, which enables mobile devices to achieve on-demand configurations through edge computing services. Colla reduces the computational load on the devices by incurring manageable communication overhead. It offers flexibility to balance latency and throughput according to the application’s requirements, thereby enabling low-latency data uploads. Experimental evaluations using real-world traces across 28 wireless scenarios demonstrate that Colla achieves effective bandwidth utilization, ranging from 66% to 93% in various conditions. Moreover, while maintaining comparable goodput, Colla significantly diminishes delay, reducing the average and the 99.9th percentile delays by 40% and 64%, respectively. Notably, even under conditions of unreliable edge links, Colla outperforms its competitors.

The burgeoning domain of the metaverse has sparked significant interest from a diverse array of industries, including healthcare services. However, the metaverse and its associated applications present various challenges to existing networks. First, to meet the increasing demands of the metaverse, there is a need for enhanced bandwidth, reduced latency, and improved packet loss control. Furthermore, the transmission mechanism should exhibit flexibility to automatically adapt to the diverse hybrid needs of different healthcare services. In this article, a multipath transmission-based paradigm tailored for the metaverse-based healthcare services is developed. Significantly, we devise an orchestration framework to reconcile edge-side subflow management with diverse healthcare applications. Using machine learning techniques, the framework can produce near-optimal subflow adjustment strategies for client nodes and miscellaneous services. Comprehensive experiments are performed on applications with diverse requirements to validate the adaptability of the framework to the application needs. The experimental results demonstrate that the proposed method enables the network to autonomously adapt to changing network conditions and service requirements. This includes applications' preferences for high throughput, low delay, and high stability. Moreover, the test results show that the proposed approach can notably decrease the occurrences of network quality falling below the minimum requirement. Given its adaptability and impact on network quality, this work paves the way for future metaverse-based healthcare services.

Emerging mission-critical industrial applications pose serious challenges to the reliability of the industrial networks currently. Heterogeneous wireless terminals access the network through various industrial gateways, but the vulnerability of edge links hinders the safe operation on the site. Existing service protection methods of wireless links satisfy reliability requirements through naive redundancy mechanisms. However, these methods are difficult to ensure the long-term safety of industrial applications and consume excessive energy. Therefore, this article proposed an on-demand service protection (OSP) mechanism, which can retransmit data on time to meet reliability demands intelligently. First, a long-term reliability model and an energy consumption optimization problem are formulated. Second, we introduced a flexible module inside the industrial gateway to identify various requirements and detect crucial data loss events. Then, an intelligent agent is designed to avoid unexpected data loss, which can generate a retransmission policy according to application requirements. Finally, multiple validations on the OSP were conducted. Both numerical results and prototype experiments illustrate that the proposed solution outperforms existing candidates in terms of reliability and energy consumption.

Recognition of road surface conditions should always be at the forefront of intelligent transportation systems for the enhancement of transportation safety and efficiency. When road surfaces are covered by ice or snow, accident rate would increase due to the reduction of road surface roughness and also friction between tire and road. High-resolution recognition of natural and man-made surfaces has been proved to be achievable by employing radars operating at low-terahertz (low-THz) frequencies. In this work, we present theoretical investigations on surface condition recognition of an icy road by employing low-THz frequencies. A theoretical model combining integral equation method (IEM), radiative transfer equation (RTE), and Rayleigh scattering theory is developed. Good agreement between the calculation results and measured data confirms the applicability of low-THz frequencies for the evaluation of icy road surface in winter. The influence of carrier frequency, ambient temperature, impurities inside the ice layer, and frozen soil surface conditions on the efficiency of this method is presented and discussed.

The evolution of cyber-physical system (CPS) benefits from substantial supports of many cutting-edge technologies. However, as a significant medium to bridge virtual and reality parts, the dependability of various network components is facing unprecedented challenges and threats. In this article, we propose a smart collaborative balancing (SCB) scheme to dynamically adjust the orchestration of network functions and efficiently optimize the workflow patterns. First, mathematical models of bandwidth allocation for multiuser with appropriate probability distribution are established. Matrix operations are utilized to solve the relevant issues based on individual congestion windows. Invasion defense mechanisms are also provided and discussed. Second, specific procedures of collaboration among different network components are presented. The capabilities of CPS, in terms of bandwidth allocation and invasion defense, are guaranteed via novel queueing policies and access control mechanisms. Third, we build a comprehensive prototype including multiple domains and users for validations. Experimental results in two scenarios illustrate that SCB not only supports service reliability of end hosts with different priorities, but also resists malicious attacks which are targeting the corresponding terminals inside domains. Compared to the benchmarks in software defined networks and traditional Internet, our scheme performs better in both available resource management and abnormal flow recognition aspects.

Ubiquitous power Internet of Things (IoT) is suffering unprecedented constraints and reliable tracing is a typical example. Motivated by software-defined and function virtualization capabilities of edge-cloud interplay, we propose a smart collaborative tracking scheme by investigating advanced parameter prediction skills and improved particle filter approaches. First, the range-based positioning issues are transformed into the vector nonlinear suboptimal estimation problem based on information fusion. Second, the importance of density function is provisioned to calculate locations and trajectories of the mobile node by obtaining cubature points, updating state estimation, and revising vector estimation. The Gauss-Newton iterative method has been utilized to achieve higher accuracy. Third, we implement our scheme into the simulation platform and prototype system. The practical deployment has been validated from multiple perspectives. Comparing with existing candidates, experimental results illustrate that the proposed algorithm is able to enhance the performance and demonstrate acceptable reliability. Potential usages are being expected in dynamic surveillance, equipment maintenance, and other emerging IoT scenarios.

Artificial intelligence is being utilized in multipath industrial networks to enhance service supporting ability. However, existing obstacles in controlling receive buffer restrict throughput even when higher bandwidth is available. Therefore, in this article, we propose a smart collaborative automation (SCA) scheme to improve resource usage and overcome buffer limitations. First, a mathematical model is established to describe primary system operations with considerations of chunk loss. The inf-supremum methodology and probability theory are adopted to track congestion window variations. Second, differences in disordered chunk expectations are analyzed to locate the critical condition of round numbers. Specific algorithm details are provided via simplifying comparison to achieve comprehensive policy selections. Third, evaluation topologies and environments are created with reasonable parameter settings. Validation results demonstrate that model-driven SCA can reduce unexpected occupations at the receiver-side. Comparing to intuition-driven schemes, overall performances, in terms of the sender's transmission capacity and receiver's buffer utilization, are improved under different experimental configurations.

Summary Multipath TCP (MPTCP) is a promising transport protocol that allows a multihomed device to simultaneously use multiple network interfaces to send application data over multiple paths. However, although applying MPTCP to data delivery introduces many and attractive benefits, the MPTCP is vulnerable to network attacks. When a path within the MPTCP connection suffers from some types of attacks (eg, a denial‐of‐service attack) and becomes underperforming, it will undoubtedly cause transmission interruption in the stable paths and thus degrade the application‐level performance. Unfortunately, the MPTCP path management mechanism is very simple and cannot timely prevent the usage of underperforming paths in multipath transmission. In this paper, we introduce a new “potentially underperforming” (PU) concept to MPTCP and propose a novel PU‐aware path usage management mechanism ((PU)2M2) for MPTCP aiming to (1) detect and declare an underperforming path and prevent the usage of underperforming paths in multipath transmission, (2) provide a finite‐state‐machine model to change per‐path's state accordingly and effectively manage multiple paths for data transmission, and (3) alleviate the packet reordering problem and make MPTCP avoid throughput performance degradation during network underperforming. We demonstrate the benefits of applying (PU)2M2 to MPTCP.

Mobile Fog Computing (MFC), as a crucial supplement to cloud computing, has its own special traits in many aspects. As smart mobile devices grow and vary in shapes and formats over the years, the need for real‐time interactions and an easy‐to‐use network is imminent. In this paper, we propose a smart collaborative policy for MFC scenarios by considering the target of rural vitalization. The challenges and drawbacks of extending cloud to fog are reviewed at the beginning. Then, the analysis of policy design is presented from the perspectives of feature comparisons, urgent requirements, and possible solutions. The details of policy establishment are introduced with necessary examples. Finally, performance evaluations are provided based on simulation platforms. Validation results related to round trip time and transmission time illustrate the significant improvements of our proposal in certain ways compared to the original candidate, which enables larger deployment in impoverished areas.

Virtual machine (VM) live migration, the key problem of modern virtualized data centers, is a challenging task since 1) Frequent traffic across data center between coupling VMs limits the efficiency of current methods. 2) Most existing approaches suffered from poor scalability issues as multi-objective optimization is still an open question in these designs. To address these problems, in this paper, a novel multi-objective VM migration algorithm is proposed. Given the definition of dominant resource fairness, a max-min fair model subject to server-side constraints is introduced. Then, we further formulate the VM migration as an optimization problem which considers application dependencies to reduce network traffic caused by migration. By incorporating the two basic VM migration algorithms, we conduct a joint formulization for maximizing the utilization of physical machines while minimizing the traffic burden across dependent VMs. The simulation result demonstrates the accuracy of the theoretic model and it is shown that our proposed method decreases network traffic by up to 82.6%, significantly improving the efficiency of data centers.

By identifying the shortcomings of using RTT to evaluate the quality of different paths in a multipath scenario, we propose a Relative Delay Estimator (RDE) to compare the relative one way delay of different paths without clock synchronisation. This estimator enables the comparison and selection of the best forward and backward paths, in terms of delay. As an initial application of RDE, we design a novel retransmission policy (NcRDE). The main novelty of this policy is that, from the multiple paths available, the path chosen for retransmission is according to the value of one way delay. We also present an extension to this scheme that takes path failures into account (PF-NcRDE). Simulation results show that, when compared with recently proposed retransmission policies, NcRDE can improve throughput when the different paths have different forward and backward delays. Also, in case of path failure PF-NcRDE enhances the performance significantly over NcRDE.