The increasing scale of graph datasets has significantly improved the performance of graph representation learning methods, but it has also introduced substantial training challenges. Graph dataset condensation techniques have emerged to compress large datasets into smaller yet information-rich datasets, while maintaining similar test performance. However, these methods strictly require downstream applications to match the original dataset and task, which often fails in cross-task and cross-domain scenarios. To address these challenges, we propose a novel causal-invariance-based and transferable graph dataset condensation method, named TGCC, providing effective and transferable condensed datasets. Specifically, to preserve domain-invariant knowledge, we first extract domain causal-invariant features from the spatial domain of the graph using causal interventions. Then, to fully capture the structural and feature information of the original graph, we perform enhanced condensation operations. Finally, through spectral-domain Enhanced contrastive learning, we inject the causal-invariant features into the condensed graph, ensuring that the compressed graph retains the causal information of the original graph. Experimental results on five public datasets and our novel FinReport dataset demonstrate that TGCC achieves up to a 13.41% improvement in cross-task and cross-domain complex scenarios compared to existing methods, and achieves state-of-the-art performance on 5 out of 6 datasets in the single dataset and task scenario.

Spiking Federated Learning (SFL) has been widely studied with the energy efficiency of Spiking Neural Networks (SNNs). However, existing SFL methods require model homogeneity and assume all clients have sufficient computational resources, resulting in the exclusion of some resource-constrained clients. To address the prevalent system heterogeneity in real-world scenarios, enabling heterogeneous SFL systems that allow clients to adaptively deploy models of different scales based on their local resources is crucial. To this end, we introduce SFedHIFI, a novel Spiking Federated Learning framework with Fire Rate-Based Heterogeneous Information Fusion. Specifically, SFedHIFI employs channel-wise matrix decomposition to deploy SNN models of adaptive complexity on clients with heterogeneous resources. Building on this, the proposed heterogeneous information fusion module enables cross-scale aggregation among models of different widths, thereby enhancing the utilization of diverse local knowledge. Extensive experiments on three public benchmarks demonstrate that SFedHIFI can effectively enable heterogeneous SFL, consistently outperforming all three baseline methods. Compared with ANN-based FL, it achieves significant energy savings with only a marginal trade-off in accuracy.

The "Fine-Tuning-as-a-Service" paradigm exposes large language models to catastrophic safety degradation from less harmful samples.Alignment-stage defenses address this by proactively injecting adversarial perturbations to bolster the model's inherent robustness against harmful drift.However, existing methods rely on perturbation directions that often conflict with harmful gradients, inadvertently facilitating the acquisition of malicious features rather than suppressing them.To address this issue, we propose Orthogonal and Adaptive Safety Alignment Strategy (OASIS) to mathematically decouple safety enforcement from harmful feature acquisition.By projecting perturbations orthogonal to harmful gradients and concentrating optimization on adaptively selected safetycritical layers, OASIS effectively resolves directional conflicts while maximizing parameter efficiency.Extensive experiments on four LLMs across three datasets (SST2, GSM8K, and AGNews) demonstrate that OASIS reduces the Harmful Score by approximately 60% compared to competitive baselines, while maintaining stable downstream task utility.Our code is publicly available at https://github. com/xiaoroyi/OASIS.

Revealing the underlying causal mechanisms in the real world is crucial for scientific and technological progress. Despite notable advances in recent decades, the lack of high-quality data and the reliance of traditional causal discovery algorithms (TCDA) on the assumption of no latent confounders, as well as their tendency to overlook the precise semantics of latent variables, have long been major obstacles to the broader application of causal discovery. To address this issue, we propose a novel causal modeling framework, TLVD, which integrates the metadata-based reasoning capabilities of large language models (LLMs) with the data-driven modeling capabilities of TCDA for inferring latent variables and their semantics. Specifically, we first employ a data-driven approach to construct a causal graph that incorporates latent variables. Then, we employ multi-LLM collaboration for latent variable inference, modeling this process as a game with incomplete information and seeking its Bayesian Nash Equilibrium (BNE) to infer the possible specific latent variables. Finally, to validate the inferred latent variables across multiple real-world web-based data sources, we leverage LLMs for evidence exploration to ensure traceability. We comprehensively evaluate TLVD on three de-identified real patient datasets provided by a hospital and two benchmark datasets. Extensive experimental results confirm the effectiveness and reliability of TLVD, with average improvements of 32.67% in Acc, 62.21% in CAcc, and 26.72% in ECit across the five datasets.

With the evolution of 6G, the procurement of Virtual Network Functions (VNFs) in Space-Ground Integrated Networks (SGINs) faces a complex challenge arising from three conflicting requirements. Specifically, the system must design truthful mechanisms to manage strategic providers with private costs, adapt to highly dynamic network topologies that render static allocation heuristics obsolete, and uphold robust empirical business privacy without subjecting the network to the variance and economic distortion inherent to noise-based cryptographic protection methods. Existing solutions fail to address these interconnected constraints simultaneously. In this paper, we propose MARLA-TGN, a novel framework for dynamic and privacy-preserving VNF auctions that addresses these challenges. Specifically, we model the strategic providers as autonomous agents trained with a Multi-Agent Reinforcement Learning (MARL) algorithm. To ensure privacy and scalability, agents learn from a mean-field signal that we enhance with bid-price standard deviation to accurately capture market volatility. For the auctioneer, we design a truthful winner-determination heuristic that leverages a Temporal Graph Network (TGN) to compute a price-independent quality score for each bid, enabling a provably monotonic and efficient greedy allocation. Rigorous theoretical analysis shows that the proposed mechanism guarantees truthfulness, individual rationality, and computational efficiency. Extensive simulation results verify that MARLA-TGN significantly outperforms state-of-the-art (SOTA) benchmarks in economic efficiency, achieving near-optimal social cost (SC) while upholding its desired economic properties.

Low-rank tensor recovery methods within the tensor singular value decomposition (t-SVD) framework have demonstrated considerable success by leveraging the inherent low-dimensional structures of multi-dimensional data. However, previous approaches in this framework often rely on linear transforms or, in some cases, nonlinear transforms constructed with fully connected networks (FCNs). These methods typically promote a global low-rank structure, which may not fully exploit the nature of multiple subspaces in real-world data. In this work, we propose a nonlinear transform to capture long-range dependencies and diverse patterns across multiple subspaces of the data within the t-SVD framework. This approach provides a richer and more nuanced representation compared to the localized processing typically seen in FCN-based transforms. In the transform domain, we construct a low-rank self-representation layer that fully exploits the multi-subspace structure inherent in tensor data. Instead of merely enforcing overall low-rankness, our method minimizes the nuclear norm of a self-representation tensor, allowing for a more precise and joint characterization of multiple subspaces. This results in a more accurate representation of the data's intrinsic low-dimensional structures, leading to superior recovery performance. This new framework, termed the DEep Low-rank Tensor representAtion (DELTA), is evaluated across several typical multi-dimensional data recovery applications, including tensor completion, robust tensor completion, and spectral snapshot imaging. Experiments on various real-world multi-dimensional data illustrate the superior performance of our DELTA.

Completing multidimensional color images is a fundamental challenge in image processing and computer vision. However, some tensor-based methods often treat RGB channels as independent modes, thereby neglecting their intrinsic correlations. To address this limitation, we represent RGB values as pure quaternions and organize them into a quaternion tensor for holistic modeling that preserves chromatic relationships. To better capture the nonlinear characteristics inherent in visual data and to improve the compactness of low-rank representations, we propose a nonlinear transformation within the quaternion domain. This design enables more expressive modeling compared to conventional linear approaches. In addition, we introduce two novel regularization terms that jointly encode global low-rankness and local smoothness, with the nonlinear transformation further enhancing the exploitation of structural priors. The overall model is optimized via a nonlinear alternating direction method of multipliers (ADMM), with theoretical guarantees of convergence. Extensive experiments on several datasets demonstrate that the proposed method significantly outperforms state-of-the-art low-rank tensor and quaternion tensor recovery techniques in multidimensional color image completion tasks.

The spiking federated learning (FL) is an emerging distributed learning paradigm that allows resource-constrained devices to train collaboratively at low power consumption without exchanging local data. It takes advantage of both the privacy computation property in FL and the energy efficiency in spiking neural networks (SNNs). However, existing spiking FL methods employ a random selection approach for client aggregation, assuming unbiased client participation. This neglect of statistical heterogeneity significantly affects the convergence and precision of the global model. In this work, we propose a credit assignment-based active client selection strategy for spiking federated learning, the SFedCA, to aggregate clients contributing to the global sample distribution balance judiciously. Specifically, the client credits are assigned by the firing intensity state before and after local model training, which reflects the difference in local data distribution from the global model. The comprehensive experiments are conducted on various non-identical and independent distribution (non-IID) scenarios. The experimental results demonstrate that the SFedCA outperforms the existing state-of-the-art spiking FL methods and requires fewer communication rounds.

The metro flow in Urban Rail Transit Systems (URTS) differs from other urban traffic flows because it is characterized by: (1) highly predetermined scheduling; and (2) interactively dynamic dependencies over the fixed physical infrastructure that vary with spatiotemporal and environmental factors. Notwithstanding the advances in graph neural networks, existing efforts fail to fully capture the characteristics and complex spatiotemporal dynamics specific to metro flow, as the innate graph-aware interactions underlying a metro flow are frequently affected by an amalgamation of: intrinsic connectivity, environmental associations, and flow-activated correlation, which usually dynamically evolve over time while containing redundant signals. We propose ReDyNet, a novel Responsive Dynamic Graph Neural Network to accurately understand the spatiotemporal dynamics of metro flow and external factors. Specifically, it employs a responsive mechanism that adapts to variations in metro flow and external influences, ensuring the construction of an appropriate dynamic graph. In addition, ReDyNet follows the merits of information bottleneck (IB) theory with redundancy disentanglement to enhance the clarity and precision of contextual spatial signals. Our experiments conducted on three real-world metro passenger flow datasets demonstrate that the proposed ReDyNet outperforms several representative baselines.

Few-shot Named Entity Recognition (NER) spotlights the tag of novel entity types in data-limited scenarios or lower-resource settings. Advances with Pre-trained Language Models (PLMs), including BERT, GPT, and their variants, have driven tremendous strategies to leverage context-dependent representations and exploit predefined relational cues, yielding significant gains in witnessing unseen entities. Nevertheless, a fundamental issue exists in prior efforts regarding their susceptibility to adversarial attacks in the intricate semantic environment. This vulnerability undermines the robustness of semantic representations, exacerbating the challenge of accurate entity identification, especially when transitioning across domains. To this end, we propose an Adversity-aware Augment Learning (AAL) solution for the few-shot NER task, dedicated to retrieving and reinforcing entity prototypes resilient to adversarial inference, thereby enhancing cross-domain semantic coherence. In particular, AAL employs a two-stage paradigm consisting of training and fine-tuning. The process initiates with augmentation learning by leveraging two kinds of prompt learning schemes, then identifies prototypes under the guidance of a variational manner. Furthermore, we devise a domain-oriented prototype refinement to optimize prototype learning under conditions of uncertainty attack, facilitating the effective transfer of common knowledge from source to target domains. The experimental results, encompassing the few-shot NER datasets under both certainty and uncertainty conditions, affirm the superiority of the proposed AAL over several representative baselines, particularly its capability against adversarial attacks.

Deep learning has been widely applied to various aspects of computer vision, but the emergence of adversarial attacks raises concerns about its reliability. Adversarial training (AT) is one of the most effective defense methods, which incorporates adversarial examples into the training data. However, AT is typically employed in a discriminative learning manner, i.e., learning the mapping (conditional probability) from samples to labels, it essentially reinforces this mapping without considering the underlying data distribution. It is notable that adversarial examples often deviate from the distribution of normal (clean) samples. Therefore, building upon existing adversarial defense schemes, we propose to further exploit the distribution of normal samples, partly from the generative learning perspective, resulting in a novel robustness enhancement paradigm. We train a simple autoencoder (AE) autoregressively on normal samples to learn their prior distribution, effectively serving as an image manifold. This AE is then used as a manifold projection operator to incorporate the distribution information of normal samples. Specifically, we organically integrate the pretrained AE into the training process of both AT and adversarial distillation (AD), a method aiming at improving the robustness of small models with low capacity. Since the AE captures the distribution of normal samples, it can adaptively pull adversarial examples closer to the normal sample manifold, weakening the attack strength of adversarial samples and easing the learning of mappings from adversarial samples to correct labels. From the Pearson correlation coefficient (PCC) between the statistics on normal and adversarial examples, it’s validated that the AE indeed pulls adversarial samples closer to normal samples. Extensive experiments illustrate that our proposed adversarial defense paradigm significantly improves the robustness compared with previous state-of-the-art AT and AD methods.

Personalized federated learning (PFL) is a new paradigm to address the statistical heterogeneity problem in federated learning. Most existing PFL methods focus on leveraging global and local information such as model interpolation or parameter decoupling. However, these methods often overlook the generalization potential during local client learning. From a local optimization perspective, we propose a simple and general PFL method, Federated learning with Flexible Sharpness-Aware Minimization (FedFSA). Specifically, we emphasize the importance of applying a larger perturbation to critical layers of the local model when using the Sharpness-Aware Minimization (SAM) optimizer. Then, we design a metric, perturbation sensitivity, to estimate the layer-wise sharpness of each local model. Based on this metric, FedFSA can flexibly select the layers with the highest sharpness to employ larger perturbation. Extensive experiments are conducted on four datasets with two types of statistical heterogeneity for image classification. The results show that FedFSA outperforms seven state-of-the-art baselines by up to 8.26% in test accuracy. Besides, FedFSA can be applied to different model architectures and easily integrated into other federated learning methods, achieving a 4.45% improvement.

Spectral super-resolution (SSR) from RGB images, which involves reconstructing hyperspectral images (HSIs) from color images, has recently received great attention. While convolutional neural network (CNN)-based methods have demonstrated strong performance, they often overlook the self-similarity across the spectral dimension of HSIs. Transformer-based approaches have addressed this limitation by leveraging self-attention mechanisms to capture spectral correlations. However, these methods encounter computational and memory overheads that scale quadratically with the size of the HSIs. To overcome these challenges, we introduce a novel Spectral-wise Low-Rank Attention (SLORA) mechanism that captures inter-spectral consistency in a low-dimensional space, thereby reducing both computational costs and model complexity. Additionally, we propose a flow-based refinement module to enhance generalization and performance on unseen HSIs. Experimental results from the NTIRE 2022 spectral reconstruction challenge and the spectral snapshot compression imaging task datasets validate the superiority of our method over state-of-the-art approaches.

The automatic and accurate detection of online fake news is crucial to society, drawing significant attention from both industry and academia. With news content becoming increasingly multimodal, assessing its truthfulness has become more challenging. Existing efforts to combat multimodal fake news primarily follow a target-egocentric paradigm, which makes predictions based solely on features extracted from the target news and its associated social context. However, their performance is constrained by the inherent knowledge paucity within the target news. To address this challenge, we propose ReTIP, a novel retrieval-enhanced framework for multimodal fake news detection. ReTIP enriches the knowledge of target news by retrieving relevant news content, along with potential diffusion participants. Specifically, ReTIP retrieves relevant content from a local content pool using a key vector generated through the joint modeling of text and images, and employs a communitybased strategy to retrieve potential participants from a historical user interaction pool. Additionally, ReTIP employs a hypergraphbased information enhancement module to align knowledge across modalities and instances at a fine-grained level by capturing higher-order correlations. Finally, an attention-based fusion layer is employed to aggregate the multi-element knowledge from retrieved instances, which is then concatenated with the target news knowledge for the final prediction. Extensive experiments on three real-world multimodal fake news datasets not only demonstrate the superior performance of ReTIP compared to state-of-the-art baselines but also confirm the effectiveness of its individual components. Our code is made publicly available at https://github.com/xytitor/ReTIP.

Revealing the underlying causal mechanisms in the real world is critical for scientific and technical progress. Despite advancements over the past decades, the lack of high-quality data and the inability of traditional causal discovery algorithms (TCDA) to fully comprehend the exact semantics of variables have long been major obstacles to the broader application of causal discovery. To address this issue, this paper proposes a novel causal modeling framework, LLM-CD, which integrates the metadata-based reasoning capabilities of large language models (LLMs) with the data-driven modeling abilities of TCDA for causal discovery. LLM-CD deeply couples the reasoning abilities of LLMs at various stages of TCDA, and enhances causal discovery through an iterative process. Due to the issues of overconfidence and hallucination in LLMs, LLM-CD quantifies and analyzes its uncertainty by incorporating evidence-based deep learning theory with the assumptions of TCDA. We utilize a large-scale de-identified real patient dataset provided by a hospital, a new dataset extracted from MIMIC-IV about the same disease (lung cancer), and two benchmark datasets to comprehensively evaluate LLM-CD. Extensive experimental results confirm the effectiveness and reliability of LLM-CD, with the highest improvement of 403.93% in the Recall and 25.77% in the Ratio metric across four datasets.

Sixth generation (6G) networks deploy unmanned aerial vehicles and mobile edge computing to provide collaborative computing and reliable connectivity for resource-limited mobile devices (MDs). However, due to the untrusted and broadcast nature of wireless transmission among communicating MDs and computing resource providers, ensuring the security of resource transactions will be challenging. Blockchain-based resource-sharing systems have been proposed to address security issues. However, these systems use existing consensus mechanisms like Proof-of-Work that consume massive amounts of system resources. In addressing this, some studies attempted to use single-agent deep reinforcement learning (DRL) in leader selection. Nevertheless, these solutions overlooked the intelligence and flexibility of blockchain configuration, and a single-point of failure can cause the system to fail. We propose a multiagent distributed deep deterministic policy gradient (MAD3PG)-assisted consensus mechanism for blockchain-based collaborative resource sharing to address these issues. First, we propose a stochastic game-based incentive-mechanism to encourage consensus nodes to participate in transaction validation. Then, we formulate the optimization problem of node selection and blockchain configuration as a Markov decision process and solve it with the MAD3PG algorithm. With MAD3PG, the agents select consensus nodes based on their experience and available resources and dynamically adjust blockchain settings. The simulation results show that MAD3PG outperforms the benchmarks in maximizing throughput and incentive while minimizing block production latency.

Multi-access edge computing (MEC)-based network slicing (MEC-NS) enables MEC network service providers (MEC-NSPs) to deploy autonomous virtual networks (slices) that deliver customized MEC services to edge Internet of Things devices (EIoTDs) with diverse quality-of-service (QoS) requirements, bringing flexibility to MEC resource management. However, developing an efficient slice-based computation task offloading and slice resource allocation (SCTOSRA) policy remains challenging due to constrained slice resources during slicing periods, evolving dynamics of the slice operating environment, and the difficulty of acquiring global information on connected EIoTDs. This paper proposes a novel adaptive and intelligent SCTOSRA scheme powered by personalized federated dueling double deep Q-learning (PerFedD3QL), which addresses these issues through three key innovations: (i) a dynamic regularization framework that enables robust adaptation across heterogeneous slice operating environments; (ii) an integrated knowledge distillation (KD) mechanism that mitigates non-IID data effects and curbs model drift; and (iii) a two-stage aggregation architecture combining parameter averaging and ensemble distillation to enhance model convergence and cross-slice generalization. PerFedD3QL constructs personalized local D3QL models at each slice and coordinates their training via federated learning to derive globally optimal SCTOSRA policies, which aim to reduce inference latency and energy consumption for connected EIoTDs while protecting data privacy and adapting to changing operational environment states of network slices over time. Simulation results demonstrate the effectiveness of the proposed PerFedD3QL-based SCTOSRA algorithm, which improves performance in reducing time delay and energy consumption compared to baseline methods while maintaining strong personalization and privacy preservation across varying slice scenarios.

The increasing demand for high-quality digital content has driven the growth of content exchange among mobile users (MUs) via device-to-device (D2D) communication. However, MUs often face challenges such as limited storage, low computational power, and short battery life, making it very difficult to meet the rising demands for content sharing. UAV-assisted D2D communication has emerged as a promising solution, integrating aerial and ground networks to enable efficient content caching and distribution while reducing latency and communication costs. However, the high mobility of MUs and increasing content size make it challenging to maintain stable communication links between MUs. This increases the complexity of content distribution, caching, and resource allocation in D2D content-sharing frameworks, resulting in higher latency, fluctuating resource demands, and lower QoS, ultimately affecting system efficiency and reliability. To address these challenges, we propose an adaptive and collaborative content-sharing and resource allocation framework integrating multi-agent twin delayed deep deterministic policy gradient (MATD3), blockchain, and a multiple-round distributed double auction (MDDA). MATD3 enables dynamic decision-making for content caching and resource allocation based on user behavior and mobility, while blockchain ensures secure, transparent, and tamper-proof content-sharing transactions. Furthermore, we propose the MDDA-based incentive scheme that allows content sellers, buyers, and the auctioneer to interact and establish optimal pricing strategies. This optimizes the content-sharing capability of MUs and edge devices, enhancing the cache hit rate and average system utility. Finally, the extensive simulation results demonstrate that our proposed scheme outperforms the benchmarks in enhancing cache hit rates, communication latency, and average system utility.

Recently, the fully-connected tensor network (FCTN) decomposition has shown a powerful capability of depicting intrinsic correlations between any pair of tensor modes. But there exists a challenging question in FCTN decomposition-based methods, i.e., the accurate determination of the complicated FCTN-rank, which contains${N(N-1)}/{2}$elements for$N$th-order tensors. In this paper, we design a structural sparsity regularization for the FCTN decomposition, which estimates the complicated FCTN-rank by adaptively pruning near-zero groups in FCTN factor. Based on this regularization, we propose a noisy tensor completion (NTC) model, aiming at the recovery of a tensor from its partial and noisy observation. Besides, we design a proximal alternating minimization (PAM)-based algorithm to solve the model. In theorem, we prove a guarantee for the global convergence of the developed algorithm. To further accelerate our method for large-scale data sets, we customize the randomized block sampling strategy for general tensor network decomposition methods by updating factors from small samples. Experiments demonstrate that our strategy can accurately estimate the FCTN-rank and achieve better reconstruction performances, and our methods outperform the state-of-the-art methods in the reconstruction of different types of real-world tensors.

Natural image matting is a widely used image processing technique that extracts foreground by predicting the alpha values of the unknown region based on the alpha values of the known foreground and background regions. However, existing image matting methods may not yield the most optimal results when applied to images containing transparent objects because the known foreground region is small or even absent. To address this shortcoming, in this paper, we propose a novel method named Transparent Object Matting using Predicted Definite Foreground and Background (TOM-PDFB), which can explore and utilize the definite foreground and background in the unknown region. For this purpose, a newly developed foreground-background confidence estimator is applied to predict the confidence level of the definite foreground and the definite background, thus providing the priors required for transparent object matting. Next, foreground-background guided progressive refinement network developed as a part of this work is adopted to incorporate the estimated definite foreground and background into the alpha matte refinement process. Extensive experimental results demonstrate that the TOM-PDFB outperforms state-of-the-art methods when applied to transparent objects. Project page:https://github.com/yihuiliang/TOM-PDFB.

Optimizing stock selection through stock ranking is one of the critical but intricate tasks in quantitative trading areas because of the non-stationary dynamics and complicated interdependencies behind stock markets. Recent studies have made efforts to model historical market movements to enhance stock selection. However, they primarily borrowed the spirit of time series modeling and sought to build a deterministic paradigm without considering the uncertain fluctuations. In addition, some of these studies tailor to explore stock correlations from a predefined (e.g., binary) graph structure and use explicitly simple relations (such as first-order relations) to guide evolving interactions. Nevertheless, aggregating predefined but shallow relationships to collaborate with stock movements may affect selection generalizability and increase the risk of portfolio failure. This study introduces a novelRelational stock selection framework via probabilisticStateSpaceLearning (orRSSL) for stock selection. Specifically, RSSL first attempts to build a tree-based structure to explicitly expose higher-order relations in the stock market, primarily by discovering a hierarchical delineation of ties between stocks. Whereafter, it couples with time-varying movements via an attention mechanism to smoothly explore the interactive correlations among different stocks. Inspired by recent state space models (SSM) in probabilistic Bayesian learning, we devise a Probabilistic Kalman Network (PKNet) with uncertainty estimates to recursively simulate ever-changing stock volatility, enabling more promising return-risk trade-offs. The experimental results on several real-world stock market datasets demonstrate that RSSL outperforms several representative baseline methods by a significant margin.

Large language models (LLMs) have gained a lot of attention and achievements recently because of their significant comprehension and generative abilities. However, the large-scale parameters of LLMs require considerable computational resources in the training and inference process, which restricts their wide application. To overcome this challenge, we propose an efficient mixed precision weight quantization (EMWQ) method for LLMs in this article. Specifically, we introduce a new outlier detection method by analyzing the weight distribution instead of the conventional weight magnitude. Then, we propose a dual-quantization strategy that quantizes both the outlier critical columns and the residual matrices with different precision. Besides, we introduce two effective EMWQ-based application frameworks, the EMWQ-R and EMWQ-O in our study. Comprehensive experiments are conducted on the Penn Treebank (PTB), C4, ARC-Easy datasets, and MMLU benchmark across various tasks. The comparison results demonstrate that the proposed EMWQ achieves state-of-the-art performance in mixed precision quantization and further reduces computational memory cost. Besides, it has higher generalizability compared with conventional methods.

Resource sharing in radio access networks (RAN) can be conceptualized as a resource trading process between infrastructure providers (InPs) and multiple mobile virtual network operators (MVNO), where InPs lease essential network resources, such as spectrum and infrastructure, to MVNOs. Given the dynamic nature of RANs, deep reinforcement learning (DRL) is a more suitable approach to decision-making and resource optimization that ensures adaptive and efficient resource allocation strategies. In RAN slicing, DRL struggles due to imbalanced data distribution and reliance on high-quality training data. In addition, the trade-off between the global solution and individual agent goals can lead to oscillatory behavior, preventing convergence to an optimal solution. Therefore, we propose a collaborative intelligent resource trading framework with a graph-based digital twin (DT) for multiple InPs and MVNOs based on Federated DRL. First, we present a customized mutual policy distillation scheme for resource trading, where complex MVNO teacher policies are distilled into InP student models and vice versa. This mutual distillation encourages collaboration to achieve personalized resource trading decisions that reach the optimal local and global solution. Second, the DT uses a graph-based model to capture the dynamic interactions between InPs and MVNOs to improve resource-trade decisions. DT can accurately predict resource prices and demand from MVNO to provide high-quality training data. In addition, DT identifies the underlying patterns and trends through advanced analytics, enabling proactive resource allocation and pricing strategies. The simulation results and analysis confirm the effectiveness and robustness of the proposed framework to an unbalanced data distribution.

The proliferation of Internet of Things (IoT) technologies and ubiquitous connectivity has led to uncrewed aerial vehicles (UAVs) playing key role as edge servers, revolutionizing the wireless communications landscape by facilitating computing and caching resources closer to ground users (GUs). This advancement significantly alleviates core network loads, reduces latency, and guarantees content availability even in congested or remote areas. However, jointly optimizing UAV caching strategies and trajectories gives rise to a multi-task optimization (MTO) problem. This paper introduces a novel multi-task geo-temporal caching (MT-GTC) framework that addresses the interplay between UAV caching mechanisms and trajectory optimization in a cohesive manner. Leveraging a proposed multi-task learning (MTL) model for joint optimization of UAV caching and trajectory design, we develop a federated learning cloud-assisted knowledge distillation (FL-CAKD) scheme to preserve data privacy and adapt to data heterogeneity. FL-CAKD transfers knowledge from a cloud model orchestrator (CMO), which houses a large and sophisticated teacher model, to a lightweight on-device MTL student models using soft target distributions instead of large model parameters, significantly reducing communication costs. MT-GTC optimizes caching and trajectories to maximize cache hits and minimize latency. Evaluations on real-world mobility datasets demonstrate up to 95% cache hit rates and 21% lower delays compared to baselines.

Federated Deep Reinforcement Learning (FDRL) for Radio Access Network (RAN) Slicing offers a promising approach for optimizing resource allocation and network performance, while also preserving data privacy for multiple tenants. However, the inherently non-independent and identically distributed (non-IID) nature of data, stemming from the diverse services and unique characteristics of RAN slices, poses significant challenges. This heterogeneity can disrupt the standard assumptions FDRL makes, leading to model training inefficiencies and potentially suboptimal slicing decisions. Addressing this non-IID challenge is imperative to harness the full potential of FDRL in RAN slicing and to ensure seamless, adaptive, and efficient resource sharing among the tenants. Hence, we propose FeDistSlice, a federated distillation slicing framework wherein multiple decision agents collaborate in real time, optimizing resource allocation tailored to each tenant's specific characteristics. Motivated by collaborative intelligence, we introduced a customized mutual policy distillation (MPD) strategy to foster collaboration across multiple tenants. This innovation allows for the creating of personalized models tailored to each agent's unique requirements and context. Through MPD, these models can collaboratively learn and refine their policies by leveraging insights from other agents within the network. Simulation results show that FeDistSlice converges more effectively and achieves increased robustness to non-IID data.

Recent advances in bio-inspired vision with event cameras and associated spiking neural networks (SNNs) have provided promising solutions for low-power consumption neuromorphic tasks. However, as the research of event cameras is still in its infancy, the amount of labeled event stream data is much less than that of the RGB database. The traditional method of converting static images into event streams by simulation to increase the sample size cannot simulate the characteristics of event cameras such as high temporal resolution. To take advantage of both the rich knowledge in labeled RGB images and the features of the event camera, we propose a transfer learning method from the RGB to the event domain in this paper. Specifically, we first introduce a transfer learning framework named R2ETL (RGB to Event Transfer Learning), including a novel encoding alignment module and a feature alignment module. Then, we introduce the temporal centered kernel alignment (TCKA) loss function to improve the efficiency of transfer learning. It aligns the distribution of temporal neuron states by adding a temporal learning constraint. Finally, we theoretically analyze the amount of data required by the deep neuromorphic model to prove the necessity of our method. Numerous experiments demonstrate that our proposed framework outperforms the state-of-the-art SNN and artificial neural network (ANN) models trained on event streams, including N-MNIST, CIFAR10-DVS and N-Caltech101. This indicates that the R2ETL framework is able to leverage the knowledge of labeled RGB images to help the training of SNN on event streams.

The emergence of network slicing as a flagship technology in 5G networks has not only enhanced network expansion and flexibility in resource management for service continuity, but also provided an avenue for establishing a viable market for resource sharing. To optimize the network's resource usage, stakeholders are encouraged to take pragmatic steps toward dynamic resource sharing. This paper designs a techno-economic model for the strategic interactions among multiple competing mobile virtual network operators (MVNOs) and their users in a trading marketplace. We formulate the dynamic pricing problem as a two-stage Stackelberg game, where the MVNOs are leaders, and the users are followers. In the first stage, the MVNOs compete to set their differentiated unit prices using a negotiation mechanism while considering system-level network load. Then, the users decide their purchasing volumes to match the prices of the MVNOs. We transform the game-based optimization problem into a stochastic Markov decision process (MDP) problem and propose a multi-agent deep Q-network (MADQN) method that obtains an optimal solution for the formulated game. Simulation results and analysis reveal that the proposed algorithm achieves convergence under the competitive pricing scheme (CPS) and independent pricing scheme (IPS) while enhancing MVNOs and users’ utilities at acceptable levels.

Traditional spiking learning algorithm aims to train neurons to spike at a specific time or on a particular frequency, which requires precise time and frequency labels in the training process. While in reality, usually only aggregated labels of sequential patterns are provided. The aggregate-label (AL) learning is proposed to discover these predictive features in distracting background streams only by aggregated spikes. It has achieved much success recently, but it is still computationally intensive and has limited use in deep networks. To address these issues, we propose an event-driven spiking aggregate learning algorithm (SALA) in this article. Specifically, to reduce the computational complexity, we improve the conventional spike-threshold-surface (STS) calculation in AL learning by analytical calculating voltage peak values in spiking neurons. Then we derive the algorithm to multilayers by event-driven strategy using aggregated spikes. We conduct comprehensive experiments on various tasks including temporal clue recognition, segmented and continuous speech recognition, and neuromorphic image classification. The experimental results demonstrate that the new STS method improves the efficiency of AL learning significantly, and the proposed algorithm outperforms the conventional spiking algorithm in various temporal clue recognition tasks.

A large quantity of labeled data is required to train high-performance deep spiking neural networks (SNNs), but obtaining labeled data is expensive. Active learning is proposed to reduce the quantity of labeled data required by deep learning models. However, conventional active learning methods in SNNs are not as effective as that in conventional artificial neural networks (ANNs) because of the difference in feature representation and information transmission. To address this issue, we propose an effective active learning method for a deep SNN model in this article. Specifically, a loss prediction module ActiveLossNet is proposed to extract features and select valuable samples for deep SNNs. Then, we derive the corresponding active learning algorithm for deep SNN models. Comprehensive experiments are conducted on CIFAR-10, MNIST, Fashion-MNIST, and SVHN on different SNN frameworks, including seven-layer CIFARNet and 20-layer ResNet-18. The comparison results demonstrate that the proposed active learning algorithm outperforms random selection and conventional ANN active learning methods. In addition, our method converges faster than conventional active learning methods.

Radio Access Network (RAN) slicing enables resource sharing among multiple tenants and is an essential feature for next-generation mobile networks. Usually, a centralized controller aggregates available resource pools from multiple tenants to increase spectrum availability. In dynamic resource allocation, a tenant could behave strategically by adjusting its preferences based on perceived conditions to maximize its utility. Slice tenants may lie about the resources needed to gain greater utility. Such behavior could lead to poor resource utilization due to excess resources acquired by lying tenants and resource shortages because slice tenants choose not to purchase high-priced resources to save costs. Furthermore, in a scenario with many slice tenants, the centralized controller can become overwhelmed by the number of requests. This, in turn, can lead to slower response times and higher latency, resulting in poor resource utilization and QoS performance of slice tenants. Therefore, this paper proposes a peer-to-peer (P2P) approach to resource trading, where slice tenants communicate directly instead of relying on a centralized orchestrator. This design is motivated by the need for slice tenants to collaborate effectively. We model the interaction between tenants in a Stackelberg multi-leader and multi-follower game and solve the game with multi-agent deep reinforcement learning with an incentive-reward model to achieve the Stackelberg equilibrium. Furthermore, we propose a decentralized resource trading framework by integrating blockchain technology and federated deep reinforcement learning, enabling network tenants to perform inter-slice resource sharing securely. The simulation results show that the proposed mechanism has significant performance improvements over existing implementations.

Text summarization is to generate a brief version of a given article while maintaining its essential meaning. Most existing solutions typically relied on the standard attention-based encoder-decoder framework, where each token in the source article, including redundancy, would be contributed to the de-coder through the attention mechanism. It follows that how to filter out the redundant content becomes an important issue in the text summarization task. In this study, we propose a localness attention network, with simplicity and feasibility in mind, which circles different local regions in the source article as contributors in different decoding steps. To further strengthen the localness model, we share the semantic space of the encoder and decoder. The experimental results conducted on two benchmark datasets demonstrate the effectiveness and applicability of the proposed method in relation to several well-practiced works.

Fifth-generation (5G) wireless networks enable gigabit-per-second data speeds, minimal latency, and reliable Internet of Things (IoT) connectivity. Thus, network slicing (NS) has gained enormous interest due to its ability to improve resource allocation. Due to the exponential growth of IoT data, it is difficult for the infrastructure providers (InPs) to determine the appropriate resource to allocate to mobile virtual network operators (MVNOs). In addition, MVNOs and IoT devices may use self-serving tactics that cause MVNOs to violate service level agreements (SLAs). Therefore, a fundamental problem in NS is capturing the interaction between MVNOs and IoT devices and ensuring efficient use of InP resources. This article proposes a two-tier resource allocation technique for NS involving a monopolistic market between an InP, multiple MVNOs, and IoT devices. First, we model the upper tier problem as a Markov decision problem (MDP) and design a federated deep reinforcement learning-based resource allocation algorithm (FDRL-RA) to explore the optimization solution. At the lower tier, we model a trading market between MVNOs and IoT devices as a two-stage Stackelberg game, where MVNOs set their unit prices and IoT devices set their purchase quantities. We use the backward induction method to analyze the proposed Stackelberg game under a competitive pricing scheme (CPS) and independent pricing scheme (IPS), which ensures high MVNOs’ profit and users’ utility at acceptable levels. Simulation results show that our proposed algorithm converges to the optimal solution and effectively maximizes utility under different pricing schemes while providing a high degree of privacy.

As the third-generation neural networks, spiking neural networks (SNNs) have great potential on neuromorphic hardware because of their high energy efficiency. However, deep spiking reinforcement learning (DSRL), that is, the reinforcement learning (RL) based on SNNs, is still in its preliminary stage due to the binary output and the nondifferentiable property of the spiking function. To address these issues, we propose a deep spiking Q -network (DSQN) in this article. Specifically, we propose a directly trained DSRL architecture based on the leaky integrate-and-fire (LIF) neurons and deep Q -network (DQN). Then, we adapt a direct spiking learning algorithm for the DSQN. We further demonstrate the advantages of using LIF neurons in DSQN theoretically. Comprehensive experiments have been conducted on 17 top-performing Atari games to compare our method with the state-of-the-art conversion method. The experimental results demonstrate the superiority of our method in terms of performance, stability, generalization and energy efficiency. To the best of our knowledge, our work is the first one to achieve state-of-the-art performance on multiple Atari games with the directly trained SNN.

Network slicing (NS) is envisioned as an emerging paradigm for accommodating different virtual networks on a common physical infrastructure. Considering the integration of blockchain and NS, a secure decentralized spectrum trading platform can be established for autonomous radio access network (RAN) slicing. Moreover, the realization of proper incentive mechanisms for fair spectrum trading is crucial for effective RAN slicing. This paper proposes a novel hierarchical framework for blockchain-empowered spectrum trading for NS in RAN. Specifically, we deploy a consortium blockchain platform for spectrum trading among spectrum providers and buyers for slice creation, and autonomous slice adjustment. For slice creation, the spectrum providers are infrastructure providers (InPs) and buyers are mobile virtual network operators (MVNOs). Then, underloaded MVNOs with extra spectrum to spare, trade with overloaded MVNOs, for slice spectrum adjustment. For proper incentive maximization, we propose a three-stage Stackelberg game framework among InPs, seller MVNOs, and buyer MVNOs, for joint optimal pricing and demand prediction strategies. Then, a multi-agent deep reinforcement learning (MADRL) method is designed to achieve a Stackelberg equilibrium (SE). Security assessment and extensive simulation results confirm the security and efficacy of our proposed method in terms of players’ utility maximization and fairness, compared with other baselines.

We investigate the computing resource allocation in multi-access edge network slicing (NS) in the context of revenue and multi-access edge computing (MEC) resource management. The significant variety of slice resource utilization levels across slice tenants (i.e., Mobile Virtual Network Operators (MVNOs)) challenges MEC resource management in NS with MEC, leading to virtual machine resource (VMR) (i.e., computing resource) wastage or scarcity. As a result, for efficient MEC resource management, the infrastructure provider (InP) encourages dynamic resource sharing and trading (DRST) of unutilized slice VMR quotas. Nevertheless, cellular network security and privacy issues deter MVNOs from collaborating on effective DRST. The security characteristics inherent in blockchain have recently gained much interest for secure resource trading. Thus, this paper proposes a unique hierarchical blockchain-based inter-slice computing resource trading (ISCRT) scheme for peer-to-peer (P2P) MVNOs in an autonomous multi-sliced MEC-based 5G network. For secure ISCRT transactions, a consortium blockchain network with hyperledger smart contracts (SC) is designed. We model the demand and pricing problems of buyer and seller MVNOs for the unutilized VMRs using a two-stage Stackelberg game. Then, to obtain the Stackelberg equilibrium (SE), an enhanced dueling double deep Q-network (D3QN) algorithm is proposed, which intelligently determines the optimal demand and pricing policies of MVNOs for the unutilized VMRs during ISCRT transactions at negotiation intervals. Simulation analysis shows that the proposed enhanced D3QN algorithm outperforms benchmark schemes in terms of the MVNO slice-level satisfaction and VMR utilization while reducing double-spending attacks in ISCRT settings by 16% and increasing both players’ utility.

Pedestrian trajectory prediction is one of the critical research issues in road traffic, which helps autonomous vehicles foresee the future paths of pedestrians and accordingly avoid crashes in time. However, the randomness and uncertainty of trajectories is a challenge caused by numerous social rules, various surroundings, and individual intentions of pedestrians. In this paper, we propose a method based on adaptive graph convolutional neural network (AGCN) to process these factors, named social interactions, from spatial and temporal perspectives. Specifically, we employ an LSTM encoder-decoder framework and adopt the AGCN to model the pedestrian spatial interactions per time step from all trajectories. Then, in order to capture the temporal interactions and reduce error accumulation, we introduce an attention mechanism to help focus more on those important moments and integrate the historical trajectory features with a distance-based loss function. We evaluate the performance of our proposed method on various benchmark datasets, and the results show our method achieves better performance compared with several existing methods.

The emergence of new business and technological models for urban-related transportation has revealed the need for transportation network companies (TNCs). Most research works on TNCs optimize the interests of drivers and passengers, and the operator assuming vehicle resources remain unchanged, but ignore the optimization of resource utilization and satisfaction from the perspective of flexible and controllable vehicle resources. In fact, the load of the scene is variable in time, which necessitates the flexible control of resources. Drivers wish to effectively utilize their vehicle resources to maximize profits. Passengers desire to spend minimum time waiting and the platform cares about the commission they can accrue from successful trips. In this article, we propose an adaptive intelligent cruise guidance and vehicle resource management model to balance vehicle resource utilization and request success rate, while improving platform revenue. We propose an advanced deep reinforcement learning (DRL) method to autonomously learn the statuses and guide the vehicles to hotspot areas where they can pick orders. We assume the number of online vehicles in the scene is flexible and the learning agent can autonomously change the number of online vehicles in the system according to the real-time load to improve effective vehicle resource utilization. An adaptive reward mechanism is enforced to control the importance of vehicle resource utilization and request success rate at decision steps. The simulation results and analysis reveal that our proposed DRL-based scheme balances vehicle resource utilization and request success rate at acceptable levels while improving the platform revenue, compared with other baseline algorithms.

As the third generation of neural networks, spiking neural networks (SNNs) have gained much attention recently because of their high energy efficiency on neuromorphic hardware. However, training deep SNNs requires many labeled data that are expensive to obtain in real-world applications, as traditional artificial neural networks (ANNs). In order to address this issue, transfer learning has been proposed and widely used in traditional ANNs, but it has limited use in SNNs. In this article, we propose an effective transfer learning framework for deep SNNs based on the domain in-variance representation. Specifically, we analyze the rationality of centered kernel alignment (CKA) as a domain distance measurement relative to maximum mean discrepancy (MMD) in deep SNNs. In addition, we study the feature transferability across different layers by testing on the Office-31, Office-Caltech-10, and PACS datasets. The experimental results demonstrate the transferability of SNNs and show the effectiveness of the proposed transfer learning framework by using CKA in SNNs.

The advent of radio access network (RAN) slicing is envisioned as a new paradigm for accommodating different virtualized networks on a single infrastructure in 5G and beyond. Consequently, infrastructure providers (InPs) desire virtualized networks to share their subleased resources for effective resource management. Nonetheless, security and privacy challenges in the wireless network deter operators from collaborating with one another for resource trading. Lately, blockchain technology has received overwhelming attention for secure resource trading thanks to its security features. This paper proposes a novel hierarchical framework for blockchain-based resource trading among peer-to-peer (P2P) mobile virtual network operators (MVNOs), for autonomous resource slicing in 5G RAN. Specifically, a consortium blockchain network that supports hyperledger smart contract (SC) is deployed to set up secure resource trading among seller and buyer MVNOs. With the aim of designing a fair incentive mechanism, we model the pricing and demand problem of the seller and buyers as a two-stage Stackelberg game, where the seller MVNO is the leader and buyer MVNOs are followers. To achieve a Stackelberg equilibrium (SE) for the formulated game, a dueling deep Q-network (Dueling DQN) scheme is designed to achieve optimal pricing and demand policies for autonomous resource allocation at negotiation interval. Comprehensive simulation results analysis prove that the proposed scheme reduces double spending attacks by 12% in resource trading settings, and maximizes the utilities of players. The proposed scheme also outperforms deep Q-Network (DQN), Q-learning (QL) and greedy algorithm (GA), in terms of slice and system level satisfaction and resource utilization.

In the fifth-generation (5G) wireless networks, Edge-Internet-of-Things (EIoT) devices are envisioned to generate huge amounts of data. Due to the limitation of computation capacity and battery life of devices, all tasks cannot be processed by these devices. However, mobile-edge computing (MEC) is a very promising solution enabling offloading of tasks to nearby MEC servers to improve quality of service. Also, during emergency situations in areas where network failure exists, unmanned aerial vehicles (UAVs) can be deployed to restore the network by acting as Aerial Base Stations and computational nodes for the edge network. In this article, we consider a central network controller who trains observations and broadcasts the trained data to a multi-UAV cluster network. Each UAV cluster head acts as an agent and autonomously allocates resources to EIoT devices in a decentralized fashion. We propose model-free deep reinforcement learning (DRL)-based collaborative computation offloading and resource allocation (CCORA-DRL) scheme in an aerial to ground (A2G) network for emergency situations, which can control the continuous action space. Each agent learns efficient computation offloading policies independently in the network and checks the statuses of the UAVs through Jain’s Fairness index. The objective is minimizing task execution delay and energy consumption and acquiring an efficient solution by adaptive learning from the dynamic A2G network. Simulation results reveal that our scheme through deep deterministic policy gradient, effectively learns the optimal policy, outperforming A3C, deep$Q$-network and greedy-based offloading for local computation in stochastic dynamic environments.

Person re-identification is the task of recognizing an individual across heterogeneous non-overlapping camera views. It has become a crucial capability needed by many applications in public space video surveillance. However, it remains a challenging task due to the subtle inter-class similarity and large intra-class variation found in person images. Current CNN-based approaches have focused and investigated traditional identification or verification frameworks. Such approaches typically use the whole input image including the background and fail to pay attention to specific body parts, deviating the feature representation learning from informative parts. In this article, we introduce a self-attention mechanism coupled with cross-resolution to improve the feature representation learning of person re-identification task. The proposed self-attention module reinforces the most informative parts from a high-resolution image using its internal representation at the low-resolution. In particular, the model is fed with a pair of images on a different scale and consists of two branches. The upper branch processes the high-resolution image and learns high dimensional feature representation while the lower branch processes the low-resolution image and learns a filtering attention heatmap. The feature maps on the lower branch are subsequently weighted to reflect the importance of each patch of the input image using a softmax operation; whereas, on the upper branch, we apply a max pooling operation to downsample the high-resolution feature map before element-wise multiplied with the attention heatmap. Our attention module helps the network learn the most discriminative visual features of multiple regions of the image and is specifically optimized to attend and enforce feature representation at different scales. Extensive experiments on three large-scale datasets show that network architectures augmented with our self-attention module systematically improve their accuracy and outperform various state-of-the-art models by a large margin.

The introduction of 5G will see exponential growth in the amount of data generated in mobile networks. This huge growth in data volume will put great pressure on not only the wireless access network but also the backhaul. In-network caching as a key component of 5G targets faster download speeds and reduction in latency through efficient content placement to avoid contents being transmitted repeatedly. In addition, the reduction in latency will require an effective resource allocation scheme to improve radio resource utilization. This paper investigates the problem of delay-aware content distribution in a multi-tenant network. We propose a content placement scheme to minimize the average visiting time of all users and a novel heuristic graph-partitioning algorithm via cell clustering to maximize the user transmission rates. Finally, simulations are conducted to evaluate the proposed scheme with QoE satisfaction and resource utilization for multi-tenants.

Local binary pattern (LBP) is a simple, yet efficient coding model for extracting texture features. To improve texture classification, this paper designs a median sampling regulation, defines a group of gradient LBP (gLBP) descriptors, proposes a training-based feature model mapping method, and then develops a texture classification frame using the multiresolution feature fusion of four gLBP descriptors. Cooperated by median sampling, four descriptors encode a pixel respectively by central gradient, radial gradient, magnitude gradient and tangent gradient to generate initial gLBP patterns. The feature mapping models of gLBP descriptors are constructed by the maximal relative-variation rate (mr2) of rotation-invariant patterns, and then prestored as mapping lookup files. By mapping, initial patterns can be transformed into low-dimensional ones. And then it generates multiresolution texture features via the joint and concatenation of gLBP descriptors on different sampling parameters. A trained nearest neighbor classifier with chi-square distance is applied to classify textures by feature histograms. The experimental results of simulation on five public texture databases show that the proposed method is reliable and efficient in texture classification. In comparison with nine other similar approaches, including two state-of-the-art ones, the proposed method runs faster than most of them and also outperforms all of them in terms of classification accuracy and noise robustness. It achieves higher accuracy and has also better robustness to the Salt&Pepper and Gaussian noise added artificially into texture images.

In recent times, there has been an increase in the number of mobile devices to access a variety of services on radio access network, and the trend is expected to continue. In addition, ultra-low latency services require much bandwidth and often characterized by having extremely short delay constraints. Hence, satisfying required strong QoS requirement becomes challenging task. Existing scheduling methods to solve this problem exhibit very poor performance in terms of transmission latency. In this paper, a scheduling-based resource reservation mechanism is proposed for cloud UE. Unlike other methods, the proposed algorithm in this paper considers various traffic parameters to calculate the effective bandwidth of the flow and always gives priority to delay sensitive flows under a software defined network framework. Simulation results show that the proposed scheduling algorithm improves the average throughput of ultra-low latency flows.

The ultra-low latency transmission for emergency services needs an effective resources management scheme to deliver content in few milliseconds. The traditional solutions can't guarantee the ultra-low latency performance required by such traffic. In this paper, we propose an air-interface slice based dynamic resource reservation schema for a massive number of sensors with emergency flows in the context of the next generation cellular networks. The proposed schema allows ultra-low latency flows to be transported by guaranteed-rate radio link connection with a content name as identifier and it achieves air-interface latency in few milliseconds. The dynamic bit-map update, silence probability and window-based re-transmission are introduced based on the Frame Slotted Aloha, which can schedules the delay-sensitive flows immediately from one or many groups of connected terminals. Furthermore, a probability theory based analytic model is provided and evaluated with Monte-Carlo simulation results.

To address the challenge of unprecedented growth in mobile data traffic, ultra-dense network deployment is a cost efficient solution to offload the traffic over some small cells. The overlapped coverage areas of small cells create more than one candidate access points for one mobile station. Signal strength based user association in IEEE 802.11 results in a significantly unbalanced load distribution among access points. However, the bandwidth demand of each user actually differs vastly due to their different preferences on mobile applications. In this paper, we formulate a non-linear integer programming model for joint user association and user bandwidth demand guarantee problem. In this model, we try to maximize the system capacity and guarantee the effective bandwidth demand for each user by soft-association control. Finally, we evaluate the proposed algorithm performance for the edge users with dynamic and heterogeneous bandwidth demands. Simulation results show that the proposed soft-association control performs better than the distributed ones and improves the individual quality of user experience with a little price on system throughput.