Yu Guo, Shenghao Ye, Shuangwu Chen, Zijian Wen, Tao Zhang, Bai Qirui, Dong Jin, Yunpeng Hou, Huasen He, Jianyang, Xiaobin Tan. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Mixture-of-Experts (MoE) models outperform traditional dense models through sparse expert activation, where each token is dynamically routed to a small subset of experts. Across many tokens, these sparse Dispatch operations induce all-to-all traffic, making communication a major bottleneck for both training and inference: unicast replication wastes bandwidth, while table-driven multicast struggles with receiver-set churn and incast. In this paper, we propose SprayCast, a congestion-adaptive native RDMA multicast scheme for dynamic sparse token Dispatch. To avoid maintaining multicast forwarding tables in switches, SprayCast encodes each packet’s destination node set in its packet header using hierarchical bitmaps, enabling table-free in-network replication. It uses in-band network telemetry (INT) feedback to steer replication away from congested multicast branches and range-based negative acknowledgments (NACKs) for localized loss recovery, saving bandwidth and reducing tail latency in dynamic all-to-all communication. In htsim simulations on a 128-server fat-tree, SprayCast achieves better scalability as top-K dispatch fanout increases and reduces P99 dispatch tail latency by up to 6 × at K = 8 compared with representative baselines.

Multicast is an effective technology for improving the quality of experience (QoE) for multiple users in 360° video services. However, unlike traditional video, 360° video multicast presents several unique challenges, including efficient user grouping, joint optimization of resource allocation and bitrate decision, and the low transmission delay and high computing demands in the transmission. To address these issues, this paper proposes a novel two-tier grouping 360° video streaming multicast scheme. We first formulate an optimization model that ensures QoE while jointly optimizing multi-user two-tier grouping, resource allocation and bitrate decisions. Subsequently, we introduce a two-tier grouping strategy and a corresponding resource allocation approach for each tier, leveraging the Shapley value for fair resource distribution. We then propose a heuristic algorithm, named Two-Tier Grouping Multicast (TTGM), which can achieve an optimal solution in certain iterations with low time complexity. Simulation results demonstrate the effectiveness of TTGM, showing that it significantly outperforms existing algorithms such as BF, oneG, RTOP, VG and Dragonfly in achieving optimal QoE performance regardless of the setting of scenarios.

Logical anomalies occur when a product's assembly violates prescribed logical rules, which widely exist in industrial assembly and packaging processes. Due to the difficulty in comprehending such complex logical relationships, a paucity of research has focused on the industrial logical anomaly diagnosis (LAD). Recently, large language models (LLMs) have demonstrated strong semantic understanding and zero-shot reasoning capabilities, making them a promising tool for LAD. However, directly applying LLMs to LAD still face two critical challenges: 1) the scarcity of abnormal samples in real-world settings, and 2) the propensity of LLMs to generate hallucinated or unreliable diagnostic conclusions. To address these challenges, we propose LogiDiag, a novel diagnostic reasoning method for LAD, to pinpoint where and why a product fails to comply with the logical rules, thereby elevating product quality and reducing remedial intervention cost. We design a visual descriptor that identifies product component attributes even in out-of-distribution abnormal images and organizes them into component descriptions for LLMs' comprehension. To mitigate hallucinations, we devise a planner to guide LLMs in diagnostic reasoning through rule orchestration, tool allocation, and chain-of-diagnosis generation. Experimental results on multiple benchmark datasets validate the competitive performance of LogiDiag.

The exponential surge in network traffic has imposed significant challenges on traditional Internet architectures, resulting in high latency and redundant transmissions. Cache-enabled networks alleviate these issues by deploying content closer to end-users, making the planning of such networks a research focus. However, regional heterogeneity in user demand and caching interdependencies among hierarchical nodes complicate the planning process. Most existing approaches rely on simplistic even allocation or empirical methods, which fail to simultaneously meet user performance expectations and minimize deployment costs. This paper proposes a network planning framework based on the Equivalent Characteristic Time Approximation (ECTA). The approach begins by establishing a performance–resource mapping. Using ECTA, we decouple the tightly coupled characteristic time relationships across hierarchical nodes, thereby accurately estimating the required cache capacity and bandwidth needed to achieve user performance targets. Building on this foundation, we formulated the network planning as a constrained convex optimization problem that minimizes deployment cost while satisfying user performance constraints. We conducted extensive experiments on a large-scale simulation platform (ndnSIM) and a real-world cache-enabled network testbed (CENI-HeFei). The results demonstrate that, under identical network topologies and total resource constraints, our method significantly improves cache hit probability while reducing deployment costs compared to homogeneous resource allocation schemes. This work provides a practical theoretical foundation and valuable insights for the design, deployment, and optimization of future cache-enabled networks.

The massive parameter scale of sparsely-activated Mixture-of-Experts (MoE) models necessitates distributed training with hybrid parallelism. Placing such training tasks,i.e.mapping the logical partitions of an MoE model to available physical NPUs, is challenging. Due to the bandwidth and latency discrepancies between intra- and inter- Pods, the cross-Pod communication usually becomes a bottleneck. The high dispersion of NPUs in multi-tenant clusters exacerbates this issue further. However, a paucity of studies has considered the cross-Pod model placement problem. To address this challenge, we propose a novel model placement scheme tailored for MoE model training with hybrid parallelism in multi-tenant clusters. By quantifying the cross-Pod communication overhead incurred during MoE model training, the model placement is formulated as a 0-1 integer quadratic problem, which is NP hard. Motivated by the traffic difference between different parallelism, we decompose this problem into two subproblems. To solve the subproblems, we propose a lightweight two-stage algorithm based on Best-Fit strategy and neighborhood search. Experiments under different models and network topologies show that our model placement scheme can reduce cross-Pod traffic by 35.9% and cut communication time by 18.7% compared to state-of-the-art methods.

The exponential growth of data creates significant challenges for distributed storage systems. Conventional cache management architectures face limitations in performance and scalability, primarily due to uncoordinated resource utilization across front-end and back-end networks, skewed data access patterns, and the demands of ultra-high concurrency. In this paper, we propose CacheMon, an in-network cache coordination system based on hybrid topology for massively scalable distributed storage. CacheMon establishes a control plane in the programmable switch to integrate front-end and back-end resources, ending the inefficiency of traditionally isolated networks. We design two key mechanisms in CacheMon: a cache tracking mechanism for maintaining data consistency through location recording, and a unified load balancing mechanism that leverages in-network measurements to optimize resource allocation and adaptively counter workload skew. Experimental results confirm that CacheMon significantly improves throughput performance and system scalability under skewed and concurrent workloads, leveraging its hybrid topology to efficiently coordinate network resources.

Straggler nodes are a major bottleneck in large-scale distributed training, degrading efficiency and stability. However, current solutions, including In-Network Aggregation (INA), lack the adaptability to effectively manage these stragglers in dynamic environments. This paper proposes Straggler Dynamic Management (SDM), an adaptive method for large-scale distributed training that performs dynamic straggler management by coordinating the data and control planes to achieve accurate, time-based detection and efficient mitigation via a performanceaware redundancy strategy and semi-asynchronous aggregation. SDM manages stragglers through a coordinated architecture that decouples the data and control planes for efficient detection and response. It leverages the data plane to estimate each node's remaining completion time, ensuring accurate and low-overhead straggler identification. The control plane then mitigates their impact using two key strategies: a performance-aware redundancy scheme to reduce waiting delays, and a semi-asynchronous aggregation mechanism that dynamically adjusts synchronization to alleviate gradient staleness and improve model convergence. We implement and deploy SDM on a real-world hardware testbed and evaluate its performance under various straggler scenarios. Experimental results demonstrate that SDM significantly improves training efficiency and convergence stability in the presence of straggler nodes, particularly when multiple stragglers occur simultaneously, exhibiting greater robustness and adaptability than existing methods.

Fault detection (FD) is essential for wafer quality control in semiconductor manufacturing (SM), as it can identify abnormal wafers in the early stages. However, frequent chamber discrepancies leads to distribution shifts in the data from different chambers, which causes performance degradation of the existing model. In this paper, we conceive a robust cross-chamber fault detection method in SM, which to the best of our knowledge is the first work that employs domain generalization to address the issue of cross-chamber fault detection in SM. Our basic idea is to map the samples from the source chambers and target chambers to the same hypersphere space, making normal samples cluster around a specific feature center while faulty samples stay away from it. Due to the scarcity of faulty samples, we propose a one-class classification-based fault detection method relying on normal samples to establish classification boundaries. To generalize the model to unseen target chambers, we design a meta-learning-based one-class domain generalization approach. We also devise a strategy to enhance distribution alignment within hypersphere, making the classification boundaries of various chambers be close to each other. The evaluations on real-world data collected from a wet process equipment in SM verify the high robustness of the model to chamber discrepancies with limited faulty samples.

Website Fingerprinting (WF) attacks have posed a serious threat to the anonymity of the onion router (Tor) communication system, as attackers can passively pry into the encrypted traffic and infer the website visited by users. To defend against WF, recent studies focus on adversarial perturbations. However, most of them suffer from a high bandwidth overhead and a low defense performance. To address this problem, our basic idea is to generate perturbation only on the sensitive regions, which can effectively mask the website’s fingerprint, thus misleading the WF attack models and reducing the bandwidth overhead. In this paper, we formulate a joint optimization problem of perturbation position and magnitude by confining the perturbations within sensitive regions, which is rarely considered in the literature. We propose a robust low-overhead WF defense method based on reinforcement learning (RL), named RLpatch. RLpatch identifies the common sensitive regions of various surrogate models and adjusts perturbation according to the query result from a query WF model. It further employs the positional frequency of perturbations to generate a common perturbation paradigm for different traces of a same website. Experimental results show that RLpatch achieves higher defense performance, lower bandwidth overhead and better robustness against adversarial training compared to the state-of-the-art methods.

Distributed training of deep neural networks (DNNs) suffers from efficiency declines in dynamic heterogeneous environments, due to the resource wastage brought by the straggler problem in data parallelism (DP) and pipeline bubbles in model parallelism (MP). Additionally, the limited resource availability requires a trade-off between training performance and long-term costs, particularly in online settings. To address these challenges, this article presents a novel online approach to maximize long-term training efficiency in heterogeneous environments through uneven data assignment and communication-aware model partitioning. A group-based hierarchical architecture combining DP and MP is developed to balance discrepant computation and communication capabilities, and offer a flexible parallel mechanism. In order to jointly optimize the performance and long-term cost of the online DL training process, we formulate this problem as a stochastic optimization with time-averaged constraints. By utilizing Lyapunov’s stochastic network optimization theory, we decompose it into several instantaneous sub-optimizations, and devise an effective online solution to address them based on tentative searching and linear solving. We have implemented a prototype system and evaluated the effectiveness of our solution based on realistic experiments, reducing batch training time by up to 68.59% over state-of-the-art methods.

Recently, with the development of AI technology, load balancing methods are commonly integrated to enhance datacenter network throughput within a spine-leaf architecture. However, traditional load balancing inevitably leads to packet disorder, affecting system efficiency. Considering that network applications can actually tolerate a certain disorder degree, we think the transmission efficiency will be improved without requiring additional processing of packets within a certain disorder tolerance threshold. Thus, we propose a disorder tolerant load balancing scheme called Disorder Tolerant Packet Spray, dotPS. We design the architecture of dotPS, controlling the degree of disorder tolerance to improve transmission efficiency, and propose the concept of a group interval which can be dynamically adjusted as a measurement of disorder tolerance. Then we design a load balancing algorithm based on group interval, requiring strict ordering for inter-group packets while tolerating disorder in intra-group packets. Finally, simulation results dedicate that compared to the state-of-the-art load balancing methods under different network loads, the proposed method reduces the average flow completion time by 8% to 18%.

QoE metrics for video provides network operators with insight into the quality of service of their video delivery, giving them valid information to optimize bandwidth resource allocation. However, with the popularization of end-to-end encryption protocols (e.g., SSL/TLS), operators cannot directly obtain valuable information from encrypted traffic. In this paper, we present MultiQoE, which leverages multimodal features with multihead attention mechanism, enabling more accurate and wide-ranging real-time DASH video QoE measurements. We carefully select round-trip time (RTT) and throughput (THR) as multimodal input features so as to capture complementary information. Building on this, we develop a robust deep learning architecture that integrates convolutional neural network for effective feature extraction and multihead attention mechanism for enhanced contextual understanding. This combination allows the model to process complex relationships between the input modalities and deliver more precise measurement related to video QoE metrics. We evaluate MultiQoE on the real-world DASH traffic dataset collected from our platform, and it outperform existing methods in QoE measurement across four tasks. Resolution and rebuffering time classification improve by 2% and 0.54%, while MSE for rebuffering duration and end time decrease by 4.32% and 1.54%, respectively.

Congestion control is the main method to solve network congestion. The Bottleneck Band-Width and Round-Trip ropagation time(BBR) congestion control algorithm, proposed by Google in 2016, can achieve lower latency while maintaining higher throughput. However, in high-bandwidth, long-delay network conditions, BBR and other improved algorithms suffer from low bandwidth utilization and slow convergence. In order to ameliorate the above problems, the QM_BBR algorithm proposed in this paper, improves the transmission performance of each phase by 1) Improving the speed of the Startup phase based on comparison, 2) Adjusting the performance gain of the ProbeBw phase based on adaptation, 3) Adding a new Quick-Measurement phase based on the state judgment, which can adaptively adjust the pacing gain according to the current network latency and the network congestion to make the network congestion end more quickly. The experimental results show that QM_BBR improves the convergence speed by up to 18%, reduces the retransmission by 77%, and increases the throughput by 8.1% compared with BBR.

Cache-enabled networks present challenges in managing rapidly changing information demand and accommodating diverse user preferences. This paper proposes a cache placement strategy named SMAC. SMAC is specifically tailored for video scenes and comprehensively considers the spatiotemporal characteristics of contents. By deeply analyzing the characteristics of data across three dimensions: platform, style, and theme of videos, SMAC can accurately capture and predict demand patterns. Additionally, SMAC introduces a cache threshold adaptive adjustment mechanism based on a sliding module. The mechanism dynamically adjusts the content placement level of caching according to changes in user preferences over time. Experimental results indicate that, compared to some common strategies, under various experimental conditions, SMAC can increase the hit ratio by 3-8%, reduce server load by 5-30%, and decrease total delay by 3-22%. The demonstration of these network performance validates the effectiveness of the SMAC.

The importance of encrypted traffic classification for network management and security is self-evident. The emergence of programmable data plane (PDP) technology makes it possible to directly implement encrypted traffic classification in the data plane, which can classify network traffics in line-rate. In this paper, we propose a two-phase encrypted traffic classification (TP-ETC) scheme in programmable data plane. In TP-ETC, Convolutional Neural Network (CNN) is employed for classifying highly similar traffic with high accuracy in the first phase, and Long Short-Term Memory (LSTM) model is responsible for classifying all remaining traffic with low storage overhead in the second phase, achieving the best balance between accuracy and storage overhead. We also design a feature extraction method suitable for PDP, effectively reducing the overhead of feature storage. In addition, we design a table segmentation algorithm to reduce the growth rate of table entries to a linear level. The experimental results demonstrate the superiority of the proposed scheme TP-ETC.

With advancements in hardware and communication technologies, the demand for 360-degree video applications has surged. However, the overall spectral efficiency is compromised due to the inability of current multicast schemes to accurately identify 360-degree video users with similar characteristics, resulting in improper user grouping. To address these issues, we propose Hcast360, an adaptive user-correlation-based 360-degree video multicast strategy, in which we designed a user correlation metric based on the throughput difference between multicast and unicast to determine the feasibility of multicast among users. This strategy accurately assigns users to appropriate multicast groups, significantly improving the utilization efficiency of wireless resources. We decompose the 360- degree video multicast problem into three sub-problems: user grouping, resource allocation, and bitrate adaptation. To solve these problems, we propose an adaptive grouping algorithm that assigns users to suitable multicast groups without predefining the number of multicast groups. Finally, we employ a genetic algorithm to determine the bitrate for 360-degree video tiles. Experimental results show that our algorithm can improve users' Quality of Experience (QoE) significantly, achieving noticeable gains without any increase in bandwidth usage.

With the rise of edge devices, video analytics has become a key application in edge computing. Single-camera systems are limited by their Field of View (FoV), making them inadequate for complex environments such as traffic intersections. In this paper, we propose Adaptive Cross-Camera Video Analytics (ACCVA), a novel system for intelligent traffic monitoring. ACCVA introduces a novel camera selection algorithm that adaptively chooses the best camera based on vehicle location and historical data, as well as an adaptive retention algorithm to prevent occlusion and recover lost objects. ACCVA establishes dynamic segmentation of camera regions and cross-camera correlation, managing real-time video inference and result sharing. Implemented on the NVIDIA Jetson Orin NX and evaluated with a real-world traffic surveillance dataset, ACCVA significantly reduces end-to-end latency and enhances accuracy compared to existing state-of-the-art systems. ACCVA excels in complex scenarios by balancing low latency and high accuracy, providing an effective solution for intelligent traffic monitoring.

Website fingerprinting has emerged as a prominent topic in the area of network management. However, the proliferation of encrypted network traffic poses new challenges for website fingerprinting. In this paper, we analyze the behavior and correlations among the network flows generated by browsing a webpage and conclude that there exist specific spatio-temporal correlations among these network flows. Based on this finding, we propose the construction of an inter-flow spatio-temporal correlation graph (STCG) to model these correlations. In the STCG, each node represents a flow, with its features capturing the properties of the flow itself, and each edge with a weight vector represents the spatio-temporal correlation between two flows. Subsequently, we propose a graph neural network-based website fingerprinting method (STC-WF) by considering the inter-flow spatio-temporal correlations, in which the Graph Attention Network (GAT) and Self-Attention Graph Pooling (SAGPool) mechanisms are employed to acquire a comprehensive representation of the STCG. To evaluate the performance of STC-WF, we construct a real-world traffic dataset and conduct comprehensive evaluations. The experimental results demonstrate that STC-WF outperforms state-of-the-art methods in terms of accuracy and time consumption.

Video delivery over wireless networks with limited network resources and dynamically changing channel quality is an important challenge, and one of the most promising solutions for tackling this problem is to employ multicast transmissions, which improves network resource utilization efficiency. This article focuses on delivering video concurrently to multiple users over mobile networks leveraging Multicast Broadcast Multimedia Service (MBMS) and Mobile Edge Computing (MEC) technology. We propose a$k$-means clustering and cooperative bargaining game-based adaptive video multicast solution (KGS) over mobile edge networks, with the goal of providing high-quality video delivery service in an envisaged MBMS service area across multiple cell sites. By taking user subgrouping, resource allocation, and bitrate adaptation into account, we establish a Cooperative Bargaining Game (CBG) based joint optimization model for multiple Multicast Broadcast Synchronized Frequency Network (MBSFN) users in mobile edge networks. Then we transform this model into a two-stage convex optimization problem and a nonlinear integer programming problem. We propose a heuristic approach to solve them and achieve a Pareto optimal video delivery strategy for all users. Finally, the efficiency of the proposed scheme is evaluated through extensive simulations.

The proliferation of 360-degree video applications has brought significant challenges to existing networks. To meet the requirements of high transmission rate, low interaction latency, and high reliability, Mobile Edge Computing (MEC) has emerged as a promising technology that enables caching and processing at network edges. In this article, we present DACOD360, a deadline-aware content delivery system for the 360-degree video streaming over MEC networks. To address the challenges such as unpredictable viewports, uneven cached tiles, concurrent requests, and dynamic bandwidth, we formulate the deadline-aware delivery problem as a long-term integer program model to maximize the Quality of Experience (QoE) under the constraints of network bandwidth, cache capacity, and deadline. This optimization problem is a complex sequential decision that considers both deadline-constrained service quality at the temporal scale and multi-user resource allocation at the spatial scale. To solve it, we decompose the original problem into two sub-problems and solve them iteratively using Deep Reinforcement Learning (DRL) and Cooperative Bargaining Game (CBG). Comprehensive experiments are conducted in a wide variety of environments, and the results demonstrate that our proposed scheme outperforms the state-of-the-art schemes in terms of long-term QoE, traffic reduction, and other metrics.

The rapid growth of mobile network traffic poses major challenges for current wireless networks regarding bandwidth, delay, mobility, and stability. To overcome these obstacles, a new network architecture called Information-Centric Networking (ICN) has emerged, effectively addressing these issues and enhancing content delivery efficiency. However, with the in-network content cache, anyone including unauthorized users can access the content from intermediate network nodes. In response to this challenge, this paper proposes an efficient and lightweight ICN content access control framework based on a hybrid-coding mechanism that combines two or more encoding operations, which does not impose additional complexity on ICN routers. In the proposed scheme, the content is first divided into multiple original blocks, and these original blocks are encoded into encoded blocks using hybrid-coding operations. Each authorized user can obtain private decoding information from the content provider, and decode them into original content using its private decoding information. The proposed scheme can fully utilize ICN’s in-network cache capability and defend against a wide range of attacks. Furthermore, security analysis, ndnSIM-based simulation, and real-world experiments demonstrate the scheme’s security, performance, and scalability.

Variations in wireless network bandwidth will have a significant impact on the performance of mobile live video streaming. When multiple users have different network latency, the way of uploading a higher bitrate version of previously uploaded video segments may improve the quality of experience (QoE) of users with high network latency. In this paper, we propose an upload-download collaborative transmission scheme for mobile live video streaming with the goal of improving the overall QoE of all users. Moreover, we designed a frame-based transmission and scheduling mechanism to reduce the delay experienced by users watching live videos. Then, we design a joint upload-download transmission algorithm based on deep reinforcement learning (DRL) that takes into account the states of both the video upload and download sides. Through extensive simulation in multi-client mobile live video streaming scenarios, the proposed scheme outperforms existing solutions in terms of overall QoE, smoothness, and live video delay.

Driven by the prevalence of the computation-intensive and delay-intensive mobile applications, Mobile Edge Computing (MEC) is emerging as a promising solution. Traditional task offloading methods usually rely on centralized decision making, which inevitably involves a high computational complexity and a large state space. However, the MEC is a typical distributed system, where the edge servers are geographically separated, and independently perform the computing tasks. This fact inspires us to conceive a distributed cooperative task offloading system, where each edge server makes its own decision on how to allocate local computing resources and how to migrate tasks among the edge servers. To characterize diverse task requirements, we divide the arrival tasks into different priorities according to the tolerance time, which enables to dynamically schedule the local computing resources for reducing the task timeout. In order to coordinate the independent decision makings of geographically separate edge servers, we propose a priority driven cooperative task offloading algorithm based on multi-agent deep reinforcement learning, where the decision making of each edge server not only depends on its own state but also on the shared global information. We further develop a Variational Recurrent Neural Network (VRNN) based global state sharing model which significantly reduces the communication overhead among edge servers. The performance evaluation conducted on a movement trajectories dataset of mobile devices verifies that the proposed algorithm can reduce the task consumption time and improve the edge computing resources utilization.

Information-Centric Networking (ICN) is a commercially viable network architecture that enables content and location separation through in-network caching, reducing dupli-cate traffic and improving network resource utilisation. As with other networks, a reasonable pricing mechanism can facilitate the deployment of ICNs by encouraging operators to participate in the deployment of ICNs. A large number of studies on ICN pricing mechanisms have been conducted in which users pay for traffic on a per-unit basis, as opposed to the real-life method of paying for traffic on packages set by operators. In this paper, based on studying the interaction between users and ISPs and CPs and establishing the utility functions of each role, we analyse the caching and pricing strategies of each entity under NASH equilibrium and compare and analyse which charging model is more able to meet the needs of network entities in the ICN environment. It is found that the package billing model is more in line with the needs of operators and users in I CN s, and can achieve the objective of incentivising ICN development. This paper also contributes to the development of the best pricing strategy for ICN networks.

Information-Centric Networking (ICN) has the characteristics of in-network caching, which can reduce the transmission of duplicate traffic, reduce the load on the servers and improve the user experience. From a technical point of view, it is a very promising network architecture. A reasonable pricing mechanism can encourage internet service providers, content providers and users to participate in the operation and use of ICN, and convert ICN technical advantages into economic benefits, thereby promote the large-scale deployment of ICN. The current research focuses on ICN pricing to analyze the pricing mechanism on the internet service provider (ISP) side and the corresponding market equilibrium results. But the model of content providers (CPs) is usually relatively simple in this research. The model assumes the existence of one single CP operator, which will be very different from future deployment scenarios. Multiple CPs will introduce competition and stimulate end users to use ICN networks and ISPs to deploy ICN networks. Moreover, the relationship between CPs is not only competitive but also cooperative. This paper focuses on the complex relationship of competition and cooperation among multiple CPs, solves the non-cooperative game model based on game theory, and studies the interaction between cache and pricing strategies of ICN entities. The optimal cache share of ISPs and the optimal pricing of ISPs and CPs are obtained by establishing the optimal utility function of each entity. Finally, numerical analysis is performed to derive the utility function of ICN entities as the critical pricing and caching parameters change, while verifying the consistency with the equilibrium solution.

The performance of Dynamic Adaptive Streaming (DAS) in multi-client scenarios can be improved by taking advantage of the aggregation capability of Named Data Networking (NDN). In this paper, we propose a client-side game theory based (GB) ABR algorithm for NDN that can achieve proactive aggregation of requests among clients as much as possible without requiring coordinating with other clients or scheduling by a central controller. We model the interaction between a DAS client and network as an incomplete information non-cooperative game. Then, this game is transformed into a complete but imperfect information game by Harsanyi transformation, and each client can issue an appropriate bitrate request by solving the Bayesian Nash Equilibrium (BNE) problem respectively. By designing the payoff function pair elaborately, the equilibrium point of the game can correspond to the situation that multiple clients issuing the same video bitrate request, that is, requests aggregation, which will reduce the repeated traffic and also achieve fairness. Compared with the existing solutions, through simulation and real-world experiments in multi-client video distribution scenarios, the GB algorithm outperforms the comparison algorithms in terms of overall Quality of Experience (QoE), fairness, and network bandwidth utilization, etc.

Caching contents close to end-users can improve the network performance, while causing the problem of guaranteeing consistency. Specifically, solutions are classified into validation and invalidation, the latter of which can provide strong cache consistency strictly required in some scenarios. To date, little work on the analysis of cache invalidation has been covered. In this work, by using conditional probability to characterize the interactive relationship between existence and validity, we develop an analytical model that evaluates the performance (hit probability and server load) of four different invalidation schemes with LRU replacement under arbitrary invalidation frequency distribution. The model allows us to theoretically identify some key parameters that affect our metrics of interest and gain some common insights on parameter settings to balance the performance of cache invalidation. Compared with other cache invalidation models, our model can achieve higher accuracy in predicting the cache hit probability. We also conduct extensive simulations that demonstrate the achievable performance of our model.

The scenario that quantities of users subscribing to a same live video content cluster together in a spatially local area poses challenges to cellular operators even in 5G unicast networks. In this regard, we propose a network paradigm exploiting eMBMS and edge computing, which relieves resource starvation in both backbone and wired access networks by grouping users and distributing the desired content to each multicast group only once. However, problems are raised by operators to perform an optimal server-side decision: how to partition users with the heterogeneity and dynamic of channel conditions, how to fairly and optimally allot resources considering both unicast and multicast users, and how to maximize the overall QoE of this live video service. To cope with these coupled problems, we formulate an optimization model based on coalition game with QoE assured, which defines a fair allocation strategy according to respective contributions and a dynamic grouping method. Subsequently, we propose a heuristic algorithm with a low-time complexity that guarantees QoE for most users and shows conspicuous reduction of annoying stalling events. Noticeably, numerical simulations reveal the fairness and near-optimality of our algorithm compared with state-of-the-art approaches in multiple scenarios.

Promptly discovering unknown network attacks is critical for reducing the risk of major loss imposed on organizations and information infrastructure. This paper aims at developing an intelligent intrusion detection system capable of classifying known attacks as well as inferring unknown ones. To achieve this, we formulate the problem of fine-grained known/unknown intrusion detection as a two-stage minimization problem, where the first stage is to seek a score measure for minimizing the empirical risk of misclassifying the known attacks, while the second stage is to find another score measure for minimizing the identification risk of inferring unknown attacks. The hierarchical nature of problem formulation allows us to employ the class conditioned auto-encoders to construct a hierarchical intrusion detection framework. Since the reconstruction errors of unknown attacks are generally higher than that of the known attacks, we further employ extreme value theory in the second stage to model the distribution of reconstruction errors for differentiating known/unknown attack. To further reduce the false positive rate, we add a benign clustering module for learning the multimodal distribution of benign traffic. We conduct an experiment on two widely used datasets for assessing intrusion detection. The results show that the proposed method improves the detection rate of unknown attacks while keeping a low false positive rate.

Current schemes for Dynamic Adaptive Streaming over HTTP (DASH) are mainly client-driven. Thus, in the scenario of multiple users watching the same video, repeated subscription and data transmission results in an under-utilization of network bandwidth resources. Additionally, competition for limited network resources of individual users may motivate selfish behaviors, which leads to unfairness and sub-optimal utility of video services. In this paper, Multimedia Broadcast Multicast Service (MBMS) in mobile edge networks for multi-bitrate video sessions is applied to overcome these limitations. We formulate a non-linear integer programming (NLIP) model, which jointly optimize bitrate adaptation and resource allocation for multiple users. This model takes video quality, playback interruptions, and quality oscillations as linear constraints to maximize multicast users' Quality of Experience (QoE). Due to NP-Hardness of this problem, we propose a heuristic greedy algorithm, which can work out the optimal or near-optimal solution with low time complexity. The evaluation results demonstrate that our method can achieve Pareto Optimality of the system utility, and maximize users' QoE while ensuring fairness.

With the development of Virtual Reality (VR) technology, the growing number of VR users puts tremendous pressure on network bandwidth. The tile-based scheme is proposed to reduce the transmission size of 360° video and improve bandwidth utilization. However, when the Field of View (FoV) of the user changes unexpectedly, the tile-based scheme will cause video distortion and quality switching by unacceptable delay. Therefore, many methods are proposed to cache the tiles that users are most likely to playback in Cloud/Edge to decrease delay. However, the dynamic adaptive bitrate delivery and the caching decision is a complex joint optimization problem, which will be a dimensional explosion problem when the scale of users and videos is large. In this paper, we design a QoE-based 360° video adaptive bitrate delivery and caching scheme aiming to maximize the quality of experience (QoE) of multi-user and ensure the fairness of users. To solve this optimization problem which is proved to be NP-Hard, we propose a bitrate selection and caching decision algorithm by greedy strategy. Numerical simulation results demonstrate that our algorithm significantly improves cache hit rate and QoE performance compared with other algorithms with fairness guaranteed.

As a promising candidate for future Internet architecture, Named Data Networking (NDN) can achieve significant potential advantages over current TCP/IP based Internet in content distribution and mobility support, etc. However, the communication mode in NDN that one Interest packet for pulling one data packet is likely to incur Interest packets flooding and to cause extremely large-scale Pending Interest Table (PIT) of the NDN router, which may substantially degrade the performance of NDN. Moreover, the absence of predefined connections also induces another challenge for NDN to manage the successive and concurrent requests from consumers efficiently. In this paper, we propose flow-based NDN (f-NDN) architecture capable of supporting flow transmission mode for addressing the aforementioned challenges. In the context of f-NDN, a data flow is defined as an aggregate of data packets with the same name prefix, and a flow Interest packet is introduced to pull a data flow rather than a single data packet. The PIT, CS, and FIB are re-designed to enable f-NDN to operate at the granularity of flows. Bitmap structure aided error handling mechanism is further presented for f-NDN to deal with the flow transmission's uncertain failures. The built-in flow support in f-NDN allows us to conceive a flow-level multi-path transmission regime for balancing the traffic in NDN network and reducing the time consumed for pulling the entire content. Weight-based flow Interest splitting algorithm and optimal rate control algorithm are both proposed for optimizing multi-path transmission. We illustrate the capability of the proposed architecture in supporting flow transmission and multipath by implementing it in both simulator and prototype systems. The evaluation results are also presented to show its achievable performance.

Named Data Networking (NDN) is an emerging future network architecture that changes the network communication model from push mode to pull mode, which leads to the requirement of a new mechanism of congestion control. To fully exploit the capability of NDN, a suitable congestion control scheme must consider the characteristics of NDN, such as connectionless, in-network caching, content perceptibility, etc. In this paper, firstly, we redefine the congestion control objective for NDN, which considers requirements diversities for different contents. Then we design and develop an efficient congestion control mechanism based on deep reinforcement learning (DRL), namely DRL-based Congestion Control Protocol (DRL-CCP). DRL-CCP enables consumers to automatically learn the optimal congestion control policy from historical congestion control experience. Finally, a real-world test platform with some typical congestion control algorithms for NDN is implemented, and a series of comparative experiments are performed on this platform to verify the performance of DRL-CCP.

Explosive growth in video traffic volumes incurs a high percentage of redundancy in today's Internet, following the 80–20 rule. Fortunately, the advanced in-network cache is considered as an effective scheme for eliminating the repetitive traffic by caching the popular content in network nodes. Besides, the emerging software-defined networking (SDN) enables centralized control and management, as well as the collaboration between network devices and upper applications. Moreover, the Network Functions Virtualization is also developed to support for customized network functions, including caching and streaming. This inspires us to design an SDN-assisted multimedia streaming Video-on-Demand system, integrating in-network cache, to improve the quality of service. The designed architecture is capable of reducing the redundant traffic via the reusable duplications. In particular, it can achieve greater performance gains by deploying specific scheduling policy. We further propose a variable-length interval cache strategy for RTP streaming, which can realize the self-adaptive adjustment of the size of cached video segments based on their access patterns. Our goal is to efficiently utilize the limited storage resources and increase the cache hit ratio. We present the theoretical analysis to demonstrate the attainable performance of the proposed algorithm; furthermore, the integrated system design is implemented as a prototype to show its feasibility and applicability. Ultimately, emulation experiments are conducted to evaluate the achievable performance improvement more comprehensively.

Named Data Networking (NDN) architecture promises significant advantages over current Internet architecture by replacing its host-centric design with a content-centric one. In NDN, the mode that one Interest packet gets one Data packet can quite easily lead to Interest flooding and a huge number of the related entries. Moreover, the absence of predefined connections is a challenge for NDN to efficiently manage the successive requests from consumers or the innetwork concurrent requests. In this paper, we argue that it is necessary for NDN to support flow transmission mechanism aimed at improving transmission performance, and based on it we design flow-based NDN (f-NDN) architecture which can not only achieve overload decreasing by packing successive Interest packets but also performance improvement and load-balance by multi-path. Simulation of our architecture is carried out in different network scenarios to evaluate the performance of our architecture. Evaluation results show that the proposed architecture operates better than NDN architecture in many aspects such as transmission efficiency and system load including reducing the number of Interest packets and lookup operations performed on the related tables in routers.

With the explosive growth in the number of mobile terminals, the demand for visual communication with mobility is increasing. However, traditional solutions for mobility over IP network cannot always meet the demand of satisfying visual communication. Named Data Networking (NDN) is a new communication model that aims to replace IP model brings a different background to mobile visual communication problems. In this paper, we take advantage of the NDN model to realize seamless mobile visual communication. We introduce a delegate with calculation functions and a globally unique identifier (GUID) which can provide native identity indication into the NDN mechanism. The use of GUID benefits real-time applications like visual communication and further works with the delegate to decrease unnecessary routing update. We also specify the naming rule and design a FIB+ to support seamless mobile visual communication. To test the performance of our solutions, we build a proof-of-concept prototype and run experiments on it. The experiments demonstrate that our solution can provide real-time video communication with seamless mobility experience.

Smart grid is an emerging technology which is designed to integrate advanced communication technologies into electrical power grids. In smart grid, automatic communications between machine devices is necessary. The number of Machine-Type Communication (MTC) devices will increase exponentially as smart grid technologies are further developed and deployed. It is a critical issue to deal with the massive accesses from an enormous number of MTC devices while guarantee the desired quality of service (QoS). In this paper, we formulate this problem as a queuing problem. Then we propose an adaptive massive access management by applying a probability estimation measurement, which is based on large deviation theory. The results demonstrate that our algorithm can adaptively adjust allocation rate and provide a better QoS. The results also show that our algorithm can improve the spectral efficiency.

Multi-Channel Multi-Radio (MCMR) Wireless Mesh Networks (WMNs) have emerged as a new paradigm in multi-hop wireless networks. In a typical MCMR WMNs, each node has multiple radios with multiple available channels on each radio, which allows nodes to have simultaneous transmissions and receptions. Therefore, network performance is improved. As a key technology in WMNs, reliable broadcast can provide efficient data transmission. GreedyCode is a network coding based reliable broadcast protocol proposed by our group earlier, whose basic idea is to opportunistically select the forwarders with the highest transmission efficiency to transmit the encoded packets while the neighbors just listen. In this paper, we consider one-to-all broadcast scenarios and propose a novel GreedyCode based reliable broadcast protocol MCMR-GreedyCode, which is two-fold: channel assignment and link scheduling. Specially, we propose the Level Channel Assignment Strategy (LCAS) algorithm and determine the number of data packets to be sent each time according to the feedback information from one-hop neighbor nodes. In addition, any intermediate node that receives complete data can forward data to those nodes that don't. The process repeats until all destination nodes receive complete data. Simulation results show that MCMR-GreedyCode has lower network latency and greater throughput than some existing network protocols, such as GreedyCode, MCM, MLRM, etc.

Reliable broadcast is an important communication primitive for wireless mesh networks, which has many applications such as multimedia services and software upgrade. Recently, network coding is introduced into reliable broadcast to enhance the throughput of data transmissions. Existing network coding based reliable broadcast schemes, such as Pacifier and R-Code, fail to take advantage of the unique characteristic of reliable broadcast or the broadcast nature of wireless transmissions, which leads to redundant transmissions and performance degradation. In this paper, we propose a greedy strategy for network coding based reliable broadcast, which is called GreedyCode. GreedyCode opportunistically selects the forwarders with the highest transmission efficiency to transmit the encoded packets while the neighbors just listen. In order to measure the efficiency of broadcast transmission of a node, we also define a metric named One-hop Broadcast Throughput (OBT), which considers not only the current reception status of the destinations but also the quality of the broadcast link. Because GreedyCode only needs the information of its one-hop neighbors, so it can be distributed realized. The simulation results show that GreedyCode achieves 100% packet delivery ratio (PDR) and significantly reduces the number of transmissions and the broadcast delay.

Network security situation awareness (NSSA) is a hotspot in the network security research field, based on the security situation values, decision makers can be aware of the actual security situation of their networks and then make rational decision to make their networks safer. In this paper, we build a multi-level quantization model for NSSA firstly; this model is comprised of three levels, namely, special oriented level, essential oriented level and holistic level. We can not only perform a certain kind of situation awareness, but also an overall one using this model. Different from the previous methods which compute network security situation of whole network just by summing up the values of each asset's network security situation, we propose a novel algorithm based on exponential and logarithmic analysis, this novel method is more appropriate to obtain rational results. Our model and algorithm are proved to be feasible and effective through a series of experiments.