Federated learning (FL) is revolutionizing machine learning by enabling multiple decentralized clients to collaboratively train a shared model. In mobile scenarios, client devices exchange model parameters with a central server via wireless channels. However, designing efficient wireless FL (WFL) is challenging due to limited energy and channel capacity. To fully utilize communication resources, over-the-air (OTA) computation has been introduced, allowing direct aggregation of analog parameter signals. However, it conceals clients’ information from the server, making advanced client selection strategies, e.g., cluster-based client selection, and sparsification compression algorithms like TopK inapplicable. To address these limitations, we propose the WFL with Voting-based Clustering (WFL-VC) algorithm, which can integrate advanced client selection and the TopK model sparsification algorithm with OTA. WFL-VC consists of two phases: 1) Phase 1: clients vote on significant parameters based on local models, allowing the server to select clients and identify the global Topk parameters; and 2) Phase 2: the selected clients upload model update parameters with globally aligned indices for over-the-air computation at the server. By combining OTA computation with cluster-based client selection and TopK sparsification, WFL-VC substantially reduces the energy consumption of OTA-based WFL. Extensive experiments on real-world datasets show that WFL-VC outperforms competitive baselines while consuming considerably less energy.

To address growing concerns about data privacy on mobile devices, the federated learning (FL) paradigm enables clients to collaboratively train models while sharing only local model updates. However, privacy risks remain in FL, as adversaries can still infer sensitive information from these updates. To enhance secure aggregation in FL, various protection mechanisms combining encryption and multi-party computation (MPC) have been proposed. These approaches, however, often introduce substantial communication and computational overhead, making secure aggregation impractical on resource-constrained devices, e.g., smart phones. To tackle these efficiency challenges, we are among the first to propose the integration of differential privacy (DP) with encryption and MPC for secure aggregation. Our proposed protocol, Federated Learning with Noise-based Secure Aggregation (FedNSA), injects noise through DP to obfuscate individual model updates. Encryption is employed to correlate the noise across different clients, while MPC ensures perfect noise cancellation at the server side. Finally, we theoretically analyze its advantages and conduct extensive experiments on public datasets to demonstrate the superiority of our approach across multiple dimensions in comparison with the state-of-the-art baselines.

With the advent of the Industry 4.0 era, Federated Learning (FL) provides robust data privacy protection for smart manufacturing and supply chain optimization, while facilitating collaborative intelligent optimization across enterprises and devices. However, the complex and overparameterized deep neural networks used in FL result in significant computational overhead for Industrial Internet of Things (IIoT) devices, leading to low energy efficiency and hindering the practical deployment of FL on IIoT devices. Moreover, the widespread data and device heterogeneity in the IIoT exacerbates the decrease in energy efficiency caused by inconsistent computational efficiency across nodes. This article proposes an energy-efficient dynamic model pruning method for FL, named EDPrune-FL, to address the aforementioned challenges. Compared to existing methods, this approach offers greater flexibility and efficiency by utilizing a dynamic pruning rate allocation mechanism. This mechanism updates the pruning rate for each participating client in every communication round, allowing the pruning upper bound to adapt to the varying importance of different learning stages in FL. EDPrune-FL ensures the global model’s performance while reducing the training energy consumption of clients in heterogeneous environments. To guarantee that dynamic pruning maintains the stability and effectiveness of the model in heterogeneous environments, we also demonstrated the convergence of EDPrune-FL and discussed the relationship between pruning rates and convergence, providing a qualitative analysis. Experimental results demonstrate that our method outperforms the state-of-the-art technique across four real-world datasets. With tests conducted on 100 clients, our approach reduces energy consumption by 10% while maintaining comparable accuracy.

Recently, federated learning (FL) has gained momentum because of its capability in preserving data privacy. To conduct model training by FL, multiple clients exchange model updates with a parameter server over the Internet. To accelerate the communication speed, it has been explored to deploy a programmable switch (PS) in lieu of the parameter server to coordinate clients. The challenge to deploy the PS in FL lies in its scarce memory space, prohibiting running memory consuming aggregation algorithms on the PS, like TopK. To overcome this challenge, we propose Federated Learning In-Network Aggregation with Global TopK Model Updates (Fed-GTopK) algorithm, consisting of two phases: voting and aggregating. In the voting phase, clients efficiently upload their votes for top model updates to the PS for selecting global top ones. Note that the voting phase consumes little memory or communication resources by only exploring the sparsity of top model updates without transmitting any model update values. In the aggregating phase, clients can reach the consensus to upload global top model updates such that the PS can swiftly aggregate global top model updates in a streamline manner without consuming much memory cost. Compared with existing works, our study is the first one accelerating in-network aggregation for FL by sparsifying model updates, and hence achieving the highest compression rate and the best learning performance. Finally, we conduct extensive experiments by using public datasets to demonstrate that Fed-GTopK remarkably surpasses the state-of-the-art baselines in terms of both model accuracy and communication traffic.

Online video streaming has evolved into an integral component of the contemporary Internet landscape. Yet, the disclosure of user requests presents formidable privacy challenges. As users stream their preferred online videos, their requests are automatically seized by video content providers, potentially leaking users’ privacy. Unfortunately, current protection methods are not well-suited to preserving user request privacy from content providers while maintaining high-quality online video services. To tackle this challenge, we introduce a novel Privacy-Preserving Video Fetching (PPVF) framework, which utilizes trusted edge devices to pre-fetch and cache videos, ensuring the privacy of users’ requests while optimizing the efficiency of edge caching. More specifically, we design PPVF with three core components: 1)Online privacy budget scheduler, which employs a theoretically guaranteed online algorithm to select non-requested videos as candidates with assigned privacy budgets. Alternative videos are chosen by an online algorithm that is theoretically guaranteed to consider both video utilities and available privacy budgets. 2)Noisy video request generator, which generates redundant video requests (in addition to original ones) utilizing correlated differential privacy to obfuscate request privacy. 3)Online video utility predictor, which leverages federated learning to collaboratively evaluate video utility in an online fashion, aiding in video selection in 1) and noise generation in 2). Finally, we conduct extensive experiments using real-world video request traces from Tencent Video and Netflix. The results demonstrate that PPVF effectively safeguards user request privacy while upholding high video caching performance.

To mitigate the rising concern on privacy infringement in recommenders, federated recommender (FRec) boosted with local differential privacy (LDP) has been proposed, in which each user privately retains his (or her) dataset, and only exposes model parameters distorted by LDP to a parameter server (PS). Although such kind of solution can largely preserve data privacy, it overlooks the potential unfairness issue. For example, a user can become a free-rider by setting an extremely small privacy budget to completely distort his (or her) model such that the model training is mainly contributed by other users. To guarantee fairness, we design a fairness-aware federated recommender framework called FFRec when users adopt heterogeneous privacy budgets. In FFRec, fairness implies that clients with higher utilities will have more opportunities to participate in training. We integrate fairness into FFRec by taking both sample size and privacy requirements into account. Specifically, a user is selected by the PS to participate in model training based on his (or her) sample size and potential noise influence by LDP and cumulative participation counts. We formulate the fairness-aware client selection problem as an integer programming problem and then reduce it into a submodular problem which has been proven to be NP-hard. To resolve this problem, we employ an approximation algorithm with an approximation ratio of 1 − e−12. Extensive experiments with popular field-measured datasets (i.e., Movielens and Netflix) demonstrate that FFRec can improve fairness by 10.3%-38.5% while guaranteeing high recommendation accuracy

As users request their preferred metaversal media, e.g., 360-degree video, user requests tracked by metaverse content providers (MCPs) pose significant privacy leakage risks. Unfortunately, existing privacy-enhancing techniques are largely ineffective in protecting user privacy for metaversal content requests since these requests cannot be easily altered or concealed by users and must remain visible to MCPs to ensure accurate content delivery. To safeguard user privacy in metaverse multimedia services (MMS), one practical approach is pre-fetching multimedia content (e.g., short videos, video patches in 360$^\circ$videos) that is not directly related to users' interests, thereby preventing MCPs from accurately inferring user preferences. However, plain pre-fetching strategies encounter a critical trade-off between privacy protection and edge caching performance given that MCPs often rely on edges for distributing metaverse content. In this paper, we propose acache-friendly and privacy-awarecontent pre-fetching(CRACE) algorithm for user devices (UDs) along with a complementary caching algorithm for edge caches (ECs). CRACE effectively mitigates privacy leakage in metaverse content requests while minimally impacting caching performance. Specifically, we introduce a novel privacy model to guide pre-fetching decisions and formulate a Stackelberg game to analyze strategic interactions between UDs and ECs. We derive optimal strategies that maximize their respective utilities and demonstrate the existence and uniqueness of the Stackelberg equilibrium. Extensive experiments conducted with real-world data demonstrate that CRACE significantly enhances privacy protection, reducing privacy disclosure by up to 59.03% compared to baseline algorithms, with negligible impact on the edge caching performance.

The scarcity data of medical field brings the collaborative training in medical vision-language pre-training (VLP) cross different clients. Therefore, the collaborative training in medical VLP faces two challenges: First, the medical data requires privacy, thus can not directly shared across different clients. Second, medical data distribution across institutes is typically heterogeneous, hindering local model alignment and representation capabilities. To simultaneously overcome these two challenges, we propose the framework called personalized model selector with fused multimodal information (PMS-FM). The contribution of PMS-FM is two-fold: 1) PMS-FM uses embeddings to represent information in different formats, allowing for the fusion of multimodal data. 2) PMS-FM adapts to personalized data distributions by training multiple models. A model selector then identifies and selects the best-performing model for each individual client. Extensive experiments with multiple real-world medical datasets demonstrate the superb performance of PMS-FM over existing federated learning methods on different zero-shot classification tasks.

Personal Knowledge Graphs (PKGs) organize an individual user's information into a structured format comprising entities, attributes, and relationships. By leveraging this structured and semantically rich data, PKGs have become essential for securing personal data management and delivering personalized services. To unlock their potential in personalized recommendations, prior research has explored the construction of PKGs and recommendation methods built upon them. However, these studies often overlook challenges associated with distributed PKGs across different users, such as joint training and privacy protection. To address these challenges, we propose PKGRec, a federated graph recommendation method specifically designed for PKGs, which utilizes a federated learning framework to ensure user privacy and data security during joint learning. Furthermore, to accommodate the user-centric graph structure of PKGs, our approach categorizes entities into three types: users, items, and other entities. It then applies a novel staged graph convolution method to model various entities based on these entity categories during local training. To enable efficient graph information sharing among distributed PKGs without requiring additional data transfer or aggregation, PKGRec performs graph expansion on the trained gradients by federated aggregation. Extensive experiments conducted on four publicly available datasets demonstrate that our method consistently outperforms the existing federated recommendation approaches.

We release a large-scale dataset that captures interactions between human users and CommentRobert, an LLM-based social media agent on Weibo. The dataset contains Weibo posts in which users actively mention the LLM agent account @CommentRobert, indicating that the users are interested in interacting with the platform-empowered LLM agent. The dataset contains 557,645 interactions from 304,400 unique users over 17 months. We detail our data collection methodology, user attributes, and content characteristics, underscoring the dataset's value in examining real-world human-LLM agent interactions. Our analysis offers insights into the demographic and behavioral traits of users interested in the selected LLM agent, interaction dynamics between humans and the agent, and linguistic patterns in comments. These interactions provide a unique lens through which to explore how humans perceive, trust, and communicate with LLMs. This dataset enables further research into modeling human intent understanding, improving LLM agent design, and studying the evolution of human-LLM agent relationships. Potential applications also include long-term user engagement prediction and AI-generated comment detection on social platforms. This constructed dataset is available at https://zenodo.org/records/16921462.

Fairness in Federated Learning (FL) is emerging as a critical factor driven by heterogeneous clients’ constraints and balanced model performance across various scenarios. In this survey, we delineate a comprehensive classification of the state-of-the-art fairness-aware approaches from a multifaceted perspective, i.e., model performance-oriented and capability-oriented. Moreover, we provide a framework to categorize and address various fairness concerns and associated technical aspects, examining their effectiveness in balancing equity and performance within FL frameworks. We further examine several significant evaluation metrics leveraged to measure fairness quantitatively. Finally, we explore exciting open research directions and propose prospective solutions that could drive future advancements in this important area, laying a solid foundation for researchers working toward fairness in FL.

Large language models (LLMs) typically require fine-tuning for domain-specific tasks, and LoRA offers a computationally efficient approach by training low-rank adaptors. LoRA is also communication-efficient for federated LLMs when multiple users collaboratively fine-tune a global LLM model without sharing their proprietary raw data. However, even the transmission of local adaptors between a server and clients risks serious privacy leakage. Applying differential privacy (DP) to federated LoRA encounters a dilemma: adding noise to both adaptors amplifies synthetic noise on the model, while fixing one adaptor impairs the learnability of fine-tuning. In this paper, we propose FedASK (Differentially Private Federated Low Rank Adaptation with Double SKetching) , a novel federated LoRA framework to enable effective updating of both low-rank adaptor matrices with robust differential privacy. Inspired by randomized SVD, our key idea is a two-stage sketching pipeline. This pipeline first aggregates carefully sketched, privacy-preserving local updates, and then reconstructs the global matrices on the server to facilitate effective updating of both adaptors. We theoretically prove FedASK's differential privacy guarantee and its exact aggregation property. Comprehensive experiments demonstrate that FedASK consistently outperforms baseline methods across a variety of privacy settings and data distributions.

The data stream generated by users on web applications is often collected using a local differential privacy (LDP) approach to ensure privacy. This approach offers rigorous theoretical guarantees and low computational overhead, albeit at the expense of data utility. Data utility encompasses both the value of individual data points and the temporal relevance that exists between them, but existing studies primarily focus on enhancing the former utility while neglecting the latter. Furthermore, the collected data often requires cleaning, and we have demonstrated through a case study that data stream lacking time relevance poses a significant risk to users' privacy during the cleaning process. In this paper, for the first time we present an online LDP publishing mechanism while preserving the inherent temporal relevance for the infinite stream, called the Sampling Period Perturbation Algorithm (SPPA). Specifically, we model the temporal relevance between data points as the Fourier interpolation function, resulting in a computational complexity reduction from O(n2) to O(n log n) when compared with the conventional Markov approach in the offline setting. To strike a better balance between privacy and utility, we add noise to the sampling period due to its minimal impact on sensitivity, which is analyzed by our novel concepts of (ε,τ)-temporal indistinguishability and (ε,w,τ)-event LDP. Through extensive experiments, SPPA exhibits superior performance in terms of both data utility and privacy preservation compared to the state-of-the-art baselines. In particular, when ε=1, compared with the state-of-the-art baseline, SPPA diminishes the MSE by up to 64.2%, and raises the event monitoring efficiency by up to 21.4%.

Federated learning (FL) in symbiotic Internet of Things (IoT) networks is a promising collaborative paradigm that utilizes IoT devices to co-train machine learning models, promising to accelerate edge intelligence for 6G. Existing studies on heterogeneous FL in IoT networks focus mainly on the differences in link capacity, ignoring the fundamental impact of channel fluctuation on model transmission and communication energy consumption. FL in symbiotic IoT networks still faces the challenges of heterogeneous and dynamic wireless links and inter-round competition of limited resource allocation, significantly impacting energy efficiency and learning performance. To address this issue, we first model wireless channel fading and dynamics for FL over symbiotic IoT networks and develop a joint optimization model for energy efficiency and learning performance. Then, we propose a novel heterogeneous-aware client sampling scheme to achieve energy-efficient training by exploiting prompt channel state tracking to predict energy consumption and update the deviation of the energy budget of each client promptly to select the optimal set of clients for each training phase. Finally, extensive experiments show that our proposed client sampling scheme significantly outperforms the existing methods and improves energy efficiency by up to$1.6\times $.

Predicting future heart rate (HR) not only helps in detecting abnormal heart rhythms but also provides timely support for downstream health monitoring services. Existing methods for HR prediction encounter challenges, especially concerning privacy protection and data heterogeneity. To address these challenges, this paper proposes a novel HR prediction framework, PCFedH, which leverages personalized federated learning and prototypical contrastive learning to achieve stable clustering results and more accurate predictions. PCFedH contains two core modules: a prototypical contrastive learning-based federated clustering module, which characterizes data heterogeneity and enhances HR representation to facilitate more effective clustering, and a two-phase soft clustered federated learning module, which enables personalized performance improvements for each local model based on stable clustering results. Experimental results on two real-world datasets demonstrate the superiority of our approach over state-of-the-art methods, achieving an average reduction of 3.1% in the mean squared error across both datasets. Additionally, we conduct comprehensive experiments to empirically validate the effectiveness of the key components in the proposed method. Among these, the personalization component is identified as the most crucial aspect of our design, indicating its substantial impact on overall performance.

With the rapid development of big data, Federated learning (FL) has found numerous applications, enabling machine learning (ML) on edge devices while preserving privacy. However, FL still faces crucial challenges, such as single point of failure and poisoning attacks, which motivate the integration of blockchain-enabled FL (BeFL). Beyond that, the efficiency issue still limits the further application of BeFL. To address these issues, we propose a novel decentralized framework: Accelerating Blockchain-Enabled Federated Learning with Clustered Clients (ABFLCC), who utilize actual training time for clustering clients to achieve hierarchical FL and solve the single point of failure problem through blockchain. Additionally, the framework clusters edge devices considering their actual training times, which allows for synchronous FL within clusters and asynchronous FL across clusters simultaneously. This approach guarantees that devices with a similar training time have a consistent global model version, improving the stability of the converging process, while the asynchronous learning between clusters enhances the efficiency of convergence. The proposed framework is evaluated through simulations on three real-world public datasets, demonstrating a training efficiency improvement of 30% to 70% in terms of convergence time compared to existing BeFL systems.

To preserve the data privacy, the federated learning (FL) paradigm emerges in which clients only expose model gradients rather than original data for conducting model training. To enhance the protection of model gradients in FL, differentially private federated learning (DPFL) is proposed which incorporates differentially private (DP) noises to obfuscate gradients before they are exposed. Yet, an essential but largely overlooked problem in DPFL is the heterogeneity of clients' privacy requirement, which can vary significantly between clients and extremely complicates the client selection problem in DPFL. In other words, both the data quality and the influence of DP noises should be taken into account when selecting clients. To address this problem, we conduct convergence analysis of DPFL under heterogeneous privacy, a generic client selection strategy, popular DP mechanisms and convex loss. Based on convergence analysis, we formulate the client selection problem to minimize the value of loss function in DPFL with heterogeneous privacy, which is a convex optimization problem and can be solved efficiently. Accordingly, we propose the DPFL-BCS (biased client selection) algorithm. The extensive experiment results with real datasets under both convex and non-convex loss functions indicate that DPFL-BCS can remarkably improve model utility compared with the SOTA baselines.

Federated learning (FL) has attracted tremendous attention due to its capability to preserve data privacy. In FL, a parameter server (PS) without accessing clients' raw data can assist decentralized clients in completing model training by aggregating and distributing model parameters for multiple iterations. From the clients' perspective, exposing model parameters to the PS can still result in privacy leakage. To further enhance privacy protection, differentially private federated learning (DPFL) is invented, in which clients add differentially private (DP) noises to distort their parameters to be exposed. However, the main challenge of DPFL lies in inferior model accuracy due to DP noises. To overcome this challenge, in this paper we propose a novel algorithmic framework for DPFL, which is called CODP, by cascading and offsetting DP noises between iterations. In existing works, each DPFL client only considers how to protect its model parameters based on the number of iterations to expose parameters overlooking the underlying relation of model parameters in consecutive iterations. The novelty of CODP lies in cascading DP noises from each iteration to its subsequent iteration so that DP noises can be offset in the subsequent iteration, and hence model accuracy can be improved. Additionally, we theoretically prove that CODP can substantially improve the convergence rate of DPFL without compromising privacy preservation by leveraging the most widely used Laplace and Gaussian mechanisms, respectively. We conduct comprehensive experiments using MNIST, Fashion-MNIST, and Lending Club datasets to demonstrate that the model accuracy of DPFL can be remarkably improved by CODP with a fixed privacy budget.

The traditional recommendation system provides web services by modeling user behavior characteristics, which also faces the risk of leaking user privacy. To mitigate the rising concern on privacy leakage in recommender systems, federated learning (FL) based recommendation has received tremendous attention, which can preserve data privacy by conducting local model training on clients. However, devices (e.g., mobile phones) used by clients in a recommender system may have limited capacity for computation and communication, which can severely deteriorate FL training efficiency. Besides, offloading local training tasks to the cloud can lead to privacy leakage and excessive pressure to the cloud. To overcome this deficiency, we propose a novel federated cloud-edge recommendation framework, which is called FCER, by offloading local training tasks to powerful and trusted edge servers. The challenge of FCER lies in the heterogeneity of edge servers, which makes the parameter server (PS) deployed in the cloud face difficulty in judiciously selecting edge servers for model training. To address this challenge, we divide the FCER framework into two stages. In the first pre-training stage, edge servers expose their data statistical features protected by local differential privacy (LDP) to the PS so that edge servers can be grouped into clusters. In the second training stage, FCER activates a single cluster in each communication round, ensuring that edge servers with statistical homogenization are not repeatedly involved in FL. The PS only selects a certain number of edge servers with the highest data quality in each cluster for FL. Effective metrics are proposed to dynamically evaluate the data quality of each edge server. Convergence rate analysis is conducted to show the convergence of recommendation algorithms in FCER. We also perform extensive experiments to demonstrate that FCER remarkably outperforms competitive baselines by$3.85\%-9.14\%$on HR@10 and$1.46\%-11.77\%$on NDCG@10.

Deep learning (DL) has found extensive application in supporting various mobile applications. The efficient execution of DL tasks is paramount for ensuring the effectiveness of AI-driven mobile applications. While previous research has predominantly focused on minimizing the completion time of DL tasks, the associated cost of execution has often been overlooked. Nonetheless, cost becomes a critical factor, particularly when utilizing DL infrastructure rented from third-party cloud service providers. In this paper, we propose a cost-aware resource scheduling framework named CRS for orchestrating DL task execution in mobile cloud systems. Our aim is to minimize server rental costs by strategically orchestrating DL jobs with diverse deadlines and workload scales across rented cloud servers. We formally define the problem and prove its NP-hardness by reducing it to a multiple knapsack problem (MKP). To solve this problem, we devise an approximation algorithm with a guaranteed upper bound performance ratio of$1+\frac{1}{e-1}$. We evaluate CRS against state-of-the-art baselines through simulations of various job arrival scenarios in a real elastic mobile cloud system. The results demonstrate that CRS, on average, reduces rental costs by 45.1% compared to other baselines, while simultaneously achieving a shorter average job completion time (JCT) and maximum job completion time (i.e., makespan).

Traditional federated learning (FL) frameworks rely on a central server for model coordination among distributed mobile terminals (MTs). The centralization faces two critical challenges, i.e., single point of failure and potential privacy leakage. Differentially private decentralized FL (DP-DFL) has been proposed to address these challenges, wherein the MTs exchange models in a decentralized manner and maintain the differential privacy (DP) guarantee by adding noise to local models before model interaction. However, existing DP-DFL frameworks confront difficulty in achieving the expected privacy and convergence performance, simultaneously. To address this issue, we propose a novel DP-DFL framework (called randomized DP-DFL) that employs a randomized model interaction scheme to lower the model exposure frequency and hence reduce privacy budget consumption. Specifically, the scheme includes two sequential steps, i.e., randomized terminal assignment and randomized model transmission. In Step 1), the model interaction phase of DFL is further divided into several sequential substages. MTs are randomly assigned to each sub-stage. In Step 2), each MT sequentially transmits either a model previously received from its neighbors or its own local model according to the assigned sub-stage order. The proposed scheme enhances the MTs' privacy of DFL since the exposure probabilities of the MTs' local models are significantly reduced via these two randomized steps. Besides, we theoretically analyze the convergence and privacy performance of randomized DP-DFL. In particular, properly tuning the number of sub-stages in randomized DP-DFL can achieve an optimal balance between privacy and convergence. Experimental results show that randomized DP-DFL consistently outperforms traditional frameworks. Compared with baselines, randomized DP-DFL reduces 40.9% privacy loss under the same target accuracy while improving 9.5% learning accuracy under the same privacy loss on EMNIST and CIFAR-10, respectively

It is known that federated learning (FL) incurs heavy communication overhead for model training by exchanging model updates between clients and the parameter server (PS) over the Internet for multiple rounds. Compressing model updates is an effective approach to alleviating communication overhead in FL. Yet the tradeoff between compression and model accuracy in the networked environment remains unclear and, for simplicity, most implementations adopt a fixed compression rate only during the entire learning process. In this paper, we for the first time systematically examine this tradeoff, explicitly quantifying the relation between the compression error, the final model accuracy and the learning rate. Specifically, we factor the compression error of each global iteration into the convergence rate analysis under non-convex loss for both unbiased and biased compression algorithms. We then present an adaptation framework to maximize the final model accuracy by strategically adjusting the compression rate in each iteration. We further discuss key implementation issues of our framework in practical networks with classical compression algorithms. Experiments over the most representative MNIST, CIFAR-10 and CIFAR-100 datasets demonstrate that our solutions effectively shrink network traffic volume while maintain high model accuracy in FL.

Hierarchical federated learning (HFL) improves the scalability and efficiency of traditional federated learning (FL) by incorporating a hierarchical topology into the FL framework. In a typical HFL system, clients are divided into multiple tiers, and the training process involves both local and global model aggregation. However, existing HFL approaches have several significant drawbacks. Firstly, the root parameter server (PS) is vulnerable to single-point failure and also acts as a bottleneck for global aggregation. Additionally, frequent global aggregation over the wide area network (WAN) incurs substantial communication costs, which negatively affect training efficiency. In this paper, we propose a novel HFL algorithm called CPFedAvg to address the aforementioned challenges. CPFedAvg introduces a root-free hierarchical topology, where the top tier consists of multiple PSes, effectively resolving the issues associated with the root PS. Additionally, we substitute the expensive global aggregation with parameter mixing operations between the PSes in the top tier. We analyze the convergence rate of CPFedAvg under non-convex loss. Based on this analysis, we formulate a convex optimization problem to optimize the frequency of executing local aggregations between consecutive parameter mixing operations. To simulate real-world communication networks, we develop FedNetSimulator to simulate a diverse range of FL communication processes. Finally, we conduct extensive experiments using real datasets (i.e., CIFAR-10 and CIFAR-100). The experimental results demonstrate that CPFedAvg can improve model accuracy by up to 18% and the speedup can be as high as 6 compared with the state-of-the-art baselines.

To mitigate the rising concern of privacy leakage, the federated recommender (FR) paradigm emerges as a potential solution, in which decentralized clients co-train the recommendation model without exposing their raw user-item rating data. The differentially private federated recommender (DPFR) further enhances the FR by injecting differentially private (DP) noises into clients' data. Yet, current DPFRs, suffering from noise distortion, cannot achieve the desired satisfactory accuracy. Various efforts have been dedicated to improving DPFRs by adaptively allocating the privacy budget over the learning process. However, due to the intricate relation between privacy budget allocation and model accuracy, existing attempts are still far from maximizing the DPFR accuracy. To address this challenge, we develop a BGTplanner (Budget Planner) to strategically allocate the privacy budget for each round of the DPFR training, improving overall training performance. Specifically, we leverage the Gaussian process regression and historical information to predict the change in the recommendation accuracy with a certain allocated privacy budget. Additionally, Contextual Multi-Armed Bandit (CMAB) is harnessed to make privacy budget allocation decisions by reconciling the current improvement and long-term privacy constraints. Our extensive experimental results on real datasets demonstrate that theBGTplannerachieves an average improvement of 6.76% in training performance compared to the state-of-the-art baselines.

Federated learning (FL) is a distributed machine learning paradigm in which clients collaboratively train models in a privacy-preserving manner. While centralized FL (CFL) suffers from single points of failure and performance bottlenecks, decentralized FL (DFL), which depends on inter-client communication, has emerged to eliminate the need of a central entity. However, due to lack of the coordination of a central server, heterogeneous data distribution across clients makes local models in DFL inclined to diverge towards their local objectives, resulting in poor model accuracy. Moreover, each client in DFL needs to communicate with multiple neighbors, yielding a heavy communication load. To tackle these challenges, we propose a novel DFL framework called DFLStar, which can improve DFL from two perspectives. First, to avoid significantly diverging towards local data, DFLStar incorporates self-knowledge distillation to enhance the local model training by assimilating knowledge from the aggregated model. Second, clients in DFLStar identify and only select the most informative neighbors (based on the last layer model similarity) for parameter exchange, thereby minimizing the communication overhead. Our experimental results on two real datasets demonstrate that DFLStar significantly improves both communication overhead and training time compared to traditional DFL algorithms while achieving a specific target accuracy. Furthermore, within a fixed training duration, DFLStar constantly obtains the highest model accuracy compared to the baselines.

Federated graph learning (FGL) has risen as a promising paradigm for collaboratively training graph neural networks while safeguarding data privacy. Nevertheless, the distributed nature of FGL also renders it susceptible to backdoor attacks. Although backdoor attacks are recognized as a significant threat to both centralized graph learning and federated learning (FL), the study of such attacks in FGL remains very limited. Current research on FGL backdoor attacks often merely adapts centralized graph backdoor attacks or FL backdoor attacks designed for image classification tasks to the FGL context, leaving key issues such as the effectiveness of triggers and the stealthiness of malicious models largely unexplored. To bridge this research gap, in this paper, we propose a novel backdoor attack, named FGLBA, targeting the FGL paradigm. Specifically, we design an input-aware trigger generator that generates a customized trigger for each target node based on its feature vector and neighborhood information, making that poisoned nodes injected with triggers are more likely misclassified into the category specified by the attacker. Additionally, we develop a stealthy federated backdoor training strategy that leverages collaborative optimization among multiple malicious clients to circumvent existing server-side defenses. The trigger generator and malicious clients' local models are iteratively optimized through a bilevel optimization framework, enabling the malicious models to achieve optimal attack performance under the optimal trigger generator. Extensive experiments on 4 real-world datasets demonstrate the effectiveness and superiority of our attack, outperforming all baseline attacks and successfully bypass 6 state-of-the-art and classical FL backdoor defenses.

Unsupervised hashing methods have gained wide-spread popularity for remote sensing (RS) image retrieval due to their high efficiency. Existing methods heavily rely on similarity matrix generated by pre-trained models as supervised signals. However, such pre-trained models obtained from natural images fail to comprehensively extract features in RS images, yielding unreliable similarity relationships. Considering complex features in RS images, we propose a novel adaptive global-local fusion network based deep unsupervised hashing (AFDUH) method. The advantages of AFDUH lie in the use of large kernel convolution and bi-level routing attention mechanism for learning local and global features simultaneously. AFDUH fuses these features under different resolutions in an interactive fashion to deduce high-quality similarity matrix based on contrastive learning. Besides, we embed a sample selection strategy in AFDUH, which can filter out unreliable supervised signals to further improve retrieval accuracy. Extensive experiments on two RS datasets demonstrate that AFDUH outperforms the state-of-the-art baselines.

Federated learning (FL) is an emerging machine learning paradigm for preserving data privacy. However, diverse client hardware often has varying computation resources. Such system heterogeneity limits the participation of resource-constrained clients in FL, and hence degrades the global model accuracy. To enable heterogeneous clients to participate in and contribute to FL training, previous works tackle this problem by assigning customized sub-models to individual clients with model pruning, distillation, or low-rank based techniques. Unfortunately, the global model trained by these methods still encounters performance degradation due to heterogeneous sub-model aggregation. Besides, most methods are heuristic-based and lack convergence analysis. In this work, we propose the FedLMT framework to bridge the performance gap, by assigning clients with a homogeneous pre-factorized low-rank model to substantially reduce resource consumption without conducting heterogeneous aggregation. We theoretically prove that the convergence of the low-rank model can guarantee the convergence of the original full model. To further meet clients’ personalized resource needs, we extend FedLMT to pFedLMT, by separating model parameters into common and custom ones. Finally, extensive experiments are conducted to verify our theoretical analysis and show that FedLMT and pFedLMT outperform other baselines with much less communication and computation costs.

Recently, federated learning (FL) has gained momentum because of its capability in preserving data privacy. To conduct model training by FL, multiple clients exchange model updates with a parameter server via Internet. To accelerate the communication speed, it has been explored to deploy a programmable switch (PS) in lieu of the parameter server to coordinate clients. The challenge to deploy the PS in FL lies in its scarce memory space, prohibiting running memory consuming aggregation algorithms on the PS. To overcome this challenge, we propose Federated Learning in-network Aggregation with Compression (FediAC) algorithm, consisting of two phases: client voting and model aggregating. In the former phase, clients report their significant model update indices to the PS to estimate global significant model updates. In the latter phase, clients upload global significant model updates to the PS for aggregation. FediAC consumes much less memory space and communication traffic than existing works because the first phase can guarantee consensus compression across clients. The PS easily aligns model update indices to swiftly complete aggregation in the second phase. Finally, we conduct extensive experiments by using public datasets to demonstrate that FediAC remarkably surpasses the state-of-the-art baselines in terms of model accuracy and communication traffic.

In Federated Learning (FL) paradigm, a parameter server (PS) concurrently communicates with distributed participating clients for model collection, update aggregation, and model distribution over multiple rounds, without touching private data owned by individual clients. FL is appealing in preserving data privacy; yet the communication between the PS and scattered clients can be a severe bottleneck. Model compression algorithms, such as quantization and sparsification, have been suggested but they generally assume a fixed code length, which does not reflect the heterogeneity and variability of model updates. In this paper, through both analysis and experiments, we show strong evidences that variable-length is beneficial for compression in FL. We accordingly present Fed-CVLC (Federated Learning Compression with Variable-Length Codes), which fine-tunes the code length in response of the dynamics of model updates. We develop optimal tuning strategy that minimizes the loss function (equivalent to maximizing the model utility) subject to the budget for communication. We further demonstrate that Fed-CVLC is indeed a general compression design that bridges quantization and sparsification, with greater flexibility. Extensive experiments have been conducted with public datasets to demonstrate that Fed-CVLC remarkably outperforms state-of-the-art baselines, improving model utility by 1.50%-5.44%, or shrinking communication traffic by 16.67%-41.61%.

In Ethereum, the ledger exchanges messages along an underlying Peer-to-Peer (P2P) network to reach consistency. Understanding the underlying network topology of Ethereum is crucial for network optimization, security and scalability. However, the accurate discovery of Ethereum network topology is non-trivial due to its deliberately designed security mechanism. Consequently, existing measuring schemes cannot accurately infer the Ethereum network topology with a low cost. To address this challenge, we propose the Distributed Ethereum Network Analyzer (DEthna) tool, which can accurately and efficiently measure the Ethereum network topology. In DEthna, a novel parallel measurement model is proposed that can generate marked transactions to infer link connections based on the transaction replacement and propagation mechanism in Ethereum. Moreover, a workload offloading scheme is designed so that DEthna can be deployed on multiple distributed probing nodes so as to measure a large-scale Ethereum network at a low cost. We run DEthna on Goerli (the most popular Ethereum test network) to evaluate its capability in discovering network topology. The experimental results demonstrate that DEthna significantly outperforms the state-of-the-art baselines. Based on DEthna, we further analyze characteristics of the Ethereum network revealing that there exist more than 50% low-degree Ethereum nodes that weaken the network robustness.

Gaze prediction is essential for enhancing user experiences of virtual reality (VR) applications. However, existing methods seldom considered the privacy nature of gaze data, which may reveal both psychological and physiological characteristics of VR users. Moreover, the commonly adopted one-sizefits-all prediction model cannot well capture behavioral patterns of different VR users. In this paper, we propose a privacyaware personalized gaze prediction framework called GazeFed, which can train a personalized gaze prediction model for each user in a collaborative manner. In GazeFed, only intermediate computations are exchanged between users and the server. The raw gaze data samples are locally preserved to protect user privacy. The global model is shared among all users, which can be further trained with local gaze data to generate a personalized prediction model for each individual user. We also propose a deep neural network tailored for VR gaze prediction called GazeNet, which can effectively extract features from VR contents, gaze data and other user behaviors, and improve the accuracy of gaze prediction. Moreover, the technique of differential privacy (DP) is also integrated to provide more privacy protection, and we theoretically prove that GazeFed can well converge and satisfy the requirement of differential privacy in the meanwhile. Last, we conduct extensive experiments to evaluate the effectiveness of our proposed GazeFed on real datasets and various VR scenarios. The experimental results demonstrate that GazeFed outperforms the state-of-the-art approaches.

Federated learning (FL) emerges as an attractive collaborative machine learning framework that enables training of models across decentralized devices by merely exposing model parameters. However, malicious attackers can still hijack communicated parameters to expose clients’ raw samples resulting in privacy leakage. To defend against such attacks, differentially private FL (DPFL) is devised, which incurs negligible computation overhead in protecting privacy by adding noises. Nevertheless, the low model utility and communication efficiency makes DPFL hard to be deployed in the real environment. To overcome these deficiencies, we propose a novel DPFL algorithm called FedDP-SA (namely, federated learning with differential privacy by splitting Local data sets and averaging parameters). Specifically, FedDP-SA splits a local data set into multiple subsets for parameter updating. Then, parameters averaged over all subsets plus differential privacy (DP) noises are returned to the parameter server. FedDP-SA offers dual benefits: 1) enhancing model accuracy by efficiently lowering sensitivity, thereby reducing noise to ensure DP and 2) improving communication efficiency by communicating model parameters with a lower frequency. These advantages are validated through sensitivity analysis and convergence rate analysis. Finally, we conduct comprehensive experiments to verify the performance of FedDP-SA compared with other state-of-the-art baseline algorithms.

Metaverse (especially 360-degree) video streaming allows broadcasting virtual events in the metaverse to a broad audience. To reduce the huge bandwidth consumption, quite a few super-resolution (SR)-enhanced 360-degree video streaming systems have been proposed. However, there is very limited work to investigate how the granularity of SR model affects the system performance, and how to choose a proper SR model for different video contents under diverse environmental conditions. In this paper, we first conduct a dedicated measurement study to unveil the impact of different granularities of SR models. It is found that the scene of a video largely determines the effectiveness of SR models in different granularities. Based on our observations, we propose a novel 360-degree video streaming framework with saliency-driven dynamic super-resolution, called SDSR. To maximize user QoE, we formally formulate an optimization problem and adopt the model predictive control (MPC) theory for bitrate adaptation and SR model selection. To improve the effectiveness of SR model, we leverage the saliency information, which well reflects users’ view interests, for model training. In addition, we reuse an SR model for similar chunks based on temporal redundancy of a video. Finally, we conduct extensive experiments on real traces and the results show that SDSR outperforms the state-of-the-art algorithms with an improvement up to 32.78% in terms of the average QoE.

Federated learning (FL) has been extensively exploited in the training of machine learning models to preserve data privacy. In particular, wireless FL enables multiple clients to collaboratively train models by sharing model updates via wireless communication without exposing raw data. The state-of-the-art wireless FL advocates efficient aggregation of model updates from multiple clients by over-the-air computing. However, a significant deficiency of over-the-air aggregation lies in the infeasibility of TopK model compression given that top model updates cannot be aggregated directly before they are aligned according to their indices. In view of the fact that TopK can greatly accelerate FL, we design a novel wireless FL with voting based TopK algorithm, namely WFL-VTopK, so that top model updates can be aggregated by over-the-air computing directly. Specifically, there are two phases in WFL-VTopK. In Phase 1, clients vote their top model updates, based on which global top model updates can be efficiently identified. In Phase 2, clients formally upload global top model updates so that they can be directly aggregated by over-the-air computing. Furthermore, the convergence of WFL-VTopK is theoretically guaranteed under non-convex loss. Based on the convergence of WFL-VTopK, we optimize model utility subjecting to training time and energy constraints. To validate the superiority of WFL-VTopK, we extensively conduct experiments with real datasets under wireless communication. The experimental results demonstrate that WFL-VTopK can effectively aggregate models by only communicating 1%-2% top models updates, and hence significantly outperforms the state-of-the-art baselines. By significantly reducing the wireless communication traffic, our work paves the road to train large models in wireless FL.

Federated Learning (FL) enables geo-distributed clients to collaboratively train a learning model without exposing their private data. By only exposing local model parameters, FL well preserves data privacy of clients. Yet, it remains possible to recover raw samples from over frequently exposed parameters resulting in privacy leakage. Differentially private federated learning (DPFL) has recently been suggested to protect these parameters by introducing information noises. In this way, even if attackers get these parameters, they cannot exactly infer true parameters from these noisy information. Directly incorporating Differentially Private (DP) into FL however can severely affect model utility. In this paper, we present an optimized sparse response mechanism (OSRM) that seamlessly incorporates DP into FL to diminish privacy budget consumption and improve model accuracy. Through OSRM, each FL client only exposes a selected set of large gradients, so as not to waste privacy budgets in protecting valueless gradients. We theoretically derive the convergence rate of DPFL with OSRM under non-convex loss. Then, OSRM is optimized by minimizing the loss of the convergence rate. Based on analysis, we present an effective algorithm for optimizing OSRM. Extensive experiments are conducted with public datasets, including MNIST, Fashion-MNIST and CIFAR-10. The results suggest that OSRM can achieve the average improvement of accuracy by 18.42% as compared to state-of-the-art baselines with a fixed privacy budget.

Although Federated Learning (FL) prevents the exposure of original data samples when collaboratively training machine learning models among decentralized clients, it has been revealed that vanilla FL is still susceptible to adversarial attacks if model parameters are leaked to malicious attackers. To enhance the protection level of FL, Differential Private Federated Learning (DPFL) has been proposed in recent years. DPFL injects zero-mean noises randomly generated by differential private (DP) mechanisms on local model parameters before they are disclosed. Nevertheless, DP noises can significantly deteriorate model utility jeopardizing the practicality of DPFL. In this paper, we are among the first to explore how to improve the model utility of DPFL by tuning the number of local iterations (LIs) on DPFL clients. Our work shows that such a local iteration tuning approach can well mitigate the adverse influence of DP noises on the final model utility. Formally, we derive the sensitivity (a measure of the maximum change of the output given two adjacent inputs) with respect to the number of LIs conducted on DPFL clients for the Laplace mechanism, and the aggregated variances of Laplace noises at the server side. We further conduct convergence rate analysis to quantify the influence of the Laplace noises on the final model accuracy and determine how to optimally set the number of LIs. Finally, to verify our theoretical findings, we perform extensive experiments using three real-world datasets, namely, Lending Club, MNIST and Fashion-MNIST. The results not only corroborate our analysis, but also demonstrate that our approach significantly improves the practicality of DPFL.

The differentially private federated learning (DPFL) paradigm emerges to firmly preserve data privacy from two perspectives. First, decentralized clients merely exchange model updates rather than raw data with a parameter server (PS) over multiple communication rounds for model training. Secondly, model updates to be exposed to the PS will be distorted by clients with differentially private (DP) noises. To incentivize clients to participate in DPFL, various incentive mechanisms have been proposed by existing works which reward participating clients based on their data quality and DP noise scales assuming that all clients are honest and genuinely report their DP noise scales. However, the PS cannot directly measure or observe DP noise scales leaving the vulnerability that clients can boost their rewards and lower DPFL utility by dishonestly reporting their DP noise scales. Through a quantitative study, we validate the adverse influence of dishonest clients in DPFL. To overcome this deficiency, we propose a robust incentive mechanism called client selection with reverse auction (CSRA) for DPFL. We prove that CSRA satisfies the properties of truthfulness, individual rationality, budget feasibility and computational efficiency. Besides, CSRA can prevent dishonest clients with two steps in each communication round. First, CSRA compares the variance of exposed model updates and claimed DP noise scale for each individual to identify suspicious clients. Second, suspicious clients will be further clustered based on their model updates to finally identify dishonest clients. Once dishonest clients are identified, CSRA will not only remove them from the current round but also lower their probability of being selected in subsequent rounds. Extensive experimental results demonstrate that CSRA can provide robust incentive against dishonest clients in DPFL and significantly outperform other baselines on three real public datasets.

Given that devices (i.e., clients) participating in federated edge learning (FEL) are autonomous and resource-constrained in nature, it is critical to design effective incentive mechanisms to encourage client participation so as to improve the performance of FEL. In this article, we aim to boost the FEL training efficiency by answering how much compute resource should clients autonomously contribute to maximize their utilities. To this end, we develop AutoFL, an autonomous client participation decision framework for federated learning at the network edge without assuming that each client possesses complete information. We first model the problem of autonomous client participation as a Bayesian game with incomplete information, where each player in the game is associated with a set of types according to network conditions. We optimize an individual client's decision based on the dynamics of the population estimated following the Bayes rule. We prove that AutoFL can converge to a unique Bayesian Nash equilibrium point. Empirical results on three real datasets show that AutoFL achieves a higher model accuracy with only 15.5-24.5% model aggregation time per global training round, and its energy cost saving on mobile devices is 82.2-86.8% compared to the state-of-the-art algorithms. Moreover, we can achieve a 2.75-3.2x long-term fairness compared to classical solutions.

In order to schedule resources efficiently or maintain applications' continuity for mobile customers, edge platforms often need to adaptively migrate the applications on them. However, our measurement shows that existing migration solutions cannot solve the issue of migrating video analytics applications in edge computing because the memory states of video analytics applications have different characteristics from other applications. We conduct a breakdown analysis of the memory states of video analytics applications, and propose to treat three types of states separately with three different techniques,i.e., warm-up, sync, and replay, to minimize the negative influence of migrations on application performance. Based on this idea, we implement a prototype system in which two new components,i.e.,state storeandsidecar, are designed to achieve near-transparent live migration with minimal application code modifications. Evaluation experiments demonstrate that the time of application interruption caused by migrating a video analytics application with our solution is less than 405ms, and our solution does not consume much resources.

Recently, federated learning (FL) has received intensive research because of its ability in preserving data privacy for scattered clients to collaboratively train machine learning models. Decentralized federated learning (DFL) is upgraded from FL which allows clients to aggregate model parameters with their neighbours directly. DFL is particularly feasible for dynamic systems, in which the neighbour set of each client is dynamic. However, due to the restrictions of client trajectories and communication distances, it is hard for individual clients to sufficiently exchange models with others, resulting in poor model accuracy. To address this challenge, we propose the DFL-DMS (DFL with Diversified Model Sources) algorithm to diversify sources for model aggregation, and improve model utility. Specifically, models exchanged between DFL-DMS clients are jointly determined by their staleness scores and the bandwidth constraint. An asynchronous learning mode is adopted so that DFL-DMS clients can temporarily store and relay fresh models collected from different client sources to accelerate the propagation of rare models. A state vector is maintained to track the contribution weight of each source to its model aggregation, and an entropy based metric (EBM) is optimized by clients in a distributed manner. Finally, the superiority of DFL-DMS is evaluated by extensive experiments (with MNIST and CIFAR-10 datasets) which demonstrate that DFL-DMS can accelerate the convergence of DFL and improve the model accuracy significantly compared with the state-of-the-art baselines.

The publicly released machine learning (ML) models are susceptible to malicious attacks (e.g., gradient leakage attacks), which may expose sensitive training data of data-sharing platforms to untrusted third-parties. To preserve the privacy of training data, differential privacy (DP) is exploited to limit the amount of leaked privacy with a predefined budget, which in fact is a non-recoverable resource. Considering DP, allocating privacy budgets to ML queries is a non-trivial but crucial problem because a certain amount of non-recoverable privacy budget will be consumed if a datablock is assigned to a query once. Meanwhile, both datablocks and ML queries are continuously generated, which further complicates the problem. Most existing works simply relied on greedy-based algorithms to make myopic allocation decisions, far away from the optimal decision. In this paper, we propose a novelHistory-awarePrivacyBudgetAllocation (HPBA) algorithm for data-sharing platforms to address the above challenges. Different from existing works, HPBA leverages historical query records to approximate global ML query patterns so as to overcome the drawback of shortsighted greedy-based algorithms. Moreover, the performance of HPBA is theoretically guaranteed by competitive analysis. A lightweight version called S-HPBA is proposed to further reduce computation overhead by using fewer historical records. Experimental results demonstrate that, compared to the state-of-the-art baselines, HPBA and S-HPBA improve the average performance by 32.8% and 16.2% in terms of model accuracy, respectively.

Federated learning (FL) is an emerging paradigm using a parameter server (PS) to coordinate multiple decentralized clients for training a common model without exposing their raw data. Despite its amazing capability in preserving data privacy, FL confronts two significant challenges that have not been sufficiently addressed by existing works, which are: 1) heterogeneous data distributed on clients given that the PS cannot alter data locations. 2) limited computation resources of FL clients who may conduct model training with mobile devices. To tackle these challenges, we propose a novel federated one-shot transferring of intermediate features (FedOTF) algorithm. More specifically, FedOTF consists of two stages: feature extraction and model reconstruction. To overcome challenge 1, clients in stage 1 only collaboratively train a small model in a federated manner, with the objective to extract features. In stage 2, intermediate features (generated by the small model trained by stage 1) are transferred from clients to the PS so that a large model can be trained to overcome challenge 2. In addition, the design of FedOTF is robust, which can flexibly diminish the number of communication rounds of stage 1 when network capacity is limited, and reduce the amount of exposed features when privacy is concerned. To verify the superiority of FedOTF, we conduct comprehensive experiments with real datasets. The experiment results demonstrate that FedOTF can significantly improve the model utility of FL because a better model can be finally obtained by the PS without incurring heavy computational load on clients. Besides, we conduct robustness evaluation of FedOTF which can achieve stable performance when varying network capacity and privacy requirement.

Volumetric video is fully three-dimensional and provides users with highly immersive and interactive experience. However, it is difficult to stream volumetric video over the Internet due to sheer video size and limited network bandwidth. Existing solutions suffered from poor perceptual quality and low coding efficiency. In this paper, we first conduct a comprehensive user study to understand the effectiveness of popular perceptual quality metrics for volumetric video. It is observed that those metrics cannot well capture the impact of user viewing behaviors. Considering the findings that users are more sensitive to the distortion of 2D image rendered from 3D point cloud, a new metric called Volu-FMAF is proposed to better represent perceptual quality of volumetric video. Next, we propose a novel neural-based volumetric video streaming framework RenderVolu and design a distortion-aware rendered image super-resolution network, called RenDA-Net, to further improve user perceptual quality. Last, we conduct extensive experiments with real datasets to validate our proposed method, and the results show that our method can boost the perceptual quality of volumetric video by 171% to 190%, and achieves a speedup of 108x in terms of decoding efficiency compared to the state-of-the-art approaches.

Different from conventional federated learning, personalized federated learning (PFL) is able to train a customized model for each individual client according to its unique requirement. The mainstream approach is to adopt a kind of weighted aggregation method to generate personalized models, in which weights are determined by the loss value or model parameters among different clients. However, such kinds of methods require clients to download others' models. It not only sheer increases communication traffic but also potentially infringes data privacy. In this paper, we propose a new PFL algorithm called FedDWA (Federated Learning with Dynamic Weight Adjustment) to address the above problem, which leverages the parameter server (PS) to compute personalized aggregation weights based on collected models from clients. In this way, FedDWA can capture similarities between clients with much less communication overhead. More specifically, we formulate the PFL problem as an optimization problem by minimizing the distance between personalized models and guidance models, so as to customize aggregation weights for each client. Guidance models are obtained by the local one-step ahead adaptation on individual clients. Finally, we conduct extensive experiments using five real datasets and the results demonstrate that FedDWA can significantly reduce the communication traffic and achieve much higher model accuracy than the state-of-the-art approaches.

In traditional machine learning, it is trivial to conduct model evaluation since all data samples are managed centrally by a server. However, model evaluation becomes a challenging problem in federated learning (FL), which is called federated evaluation in this work. This is because clients do not expose their original data to preserve data privacy. Federated evaluation plays a vital role in client selection, incentive mechanism design, malicious attack detection, etc. In this paper, we provide the first comprehensive survey of existing federated evaluation methods. Moreover, we explore various applications of federated evaluation for enhancing FL performance and finally present future research directions by envisioning some challenges.

Federated learning (FL) is a prospective distributed machine learning framework that can preserve data privacy. In particular, cross-silo FL can complete model training by making isolated data islands of different organizations collaborate with a parameter server (PS) via exchanging model parameters for multiple communication rounds. In cross-silo FL, an incentive mechanism is indispensable for motivating data owners to contribute their models to FL training. However, how to allocate the reward budget among different rounds is an essential but complicated problem largely overlooked by existing works. The challenge of this problem lies in the opaque feedback between reward budget allocation and model utility improvement of FL, making the optimal reward budget allocation complicated. To address this problem, we design an online reward budget allocation algorithm using Bayesian optimization named BARA (Budget Allocation for Reverse Auction). Specifically, BARA can model the complicated relationship between reward budget allocation and final model accuracy in FL based on historical training records so that the reward budget allocated to each communication round is dynamically optimized so as to maximize the final model utility. We further incorporate the BARA algorithm into reverse auction-based incentive mechanisms to illustrate its effectiveness. Extensive experiments are conducted on real datasets to demonstrate that BARA significantly outperforms competitive baselines by improving model utility with the same amount of reward budget.

Nowadays, the training data and neural network models are getting increasingly large. The training time of deep learning will become unbearably long on a single machine. To reduce the computation and storage burdens, distributed deep learning has been put forward to collaboratively train a large neural network model with multiple computing nodes in parallel. The unbalanced development of computation and communication capabilities has led to training time being dominated by communication time, making the communication overhead a major challenge toward efficient distributed deep learning. Communication compression is an effective method to alleviate communication overhead, and it has evolved from simple random sparsification or quantization to versatile strategies or data structures. In this survey, existing communication compression techniques are reviewed and classified to provide a bird’s eye view. The main properties of each class of compression methods are analyzed, and their applications or theoretical convergence are described if necessary. This survey is potentially helpful for researchers and engineers to understand the up-to-date achievements on the communication compression techniques that accelerate the training of large deep learning models.

Federated learning (FL) empowers distributed clients to collaboratively train a shared machine learning model through exchanging parameter information. Despite the fact that FL can protect clients’ raw data, malicious users can still crack original data with disclosed parameters. To amend this flaw, differential privacy (DP) is incorporated into FL clients to disturb original parameters, which however can significantly impair the accuracy of the trained model. In this work, we study an imperative question which has been vastly overlooked by existing works: what are the optimal numbers of queries and replies in FL with DP so that the final model accuracy is maximized. In FL, the parameter server (PS) needs to query participating clients for multiple global iterations to complete training. Each client responds a query from the PS by conducting a local iteration. We consider FL that will uniformly and randomly select participating clients to conduct local iterations with the FedSGD algorithm. Our work investigates how many times the PS should query clients and how many times each client should reply the PS by incorporating two most extensively used DP mechanisms (i.e., the Laplace mechanism and Gaussian mechanisms). Through conducting convergence rate analysis, we can determine the optimal numbers of queries and replies in FL with DP so that the final model accuracy can be maximized. Finally, extensive experiments are conducted with publicly available datasets: MNIST and FEMNIST, to verify our analysis and the results demonstrate that properly setting the numbers of queries and replies can significantly improve the final model accuracy in FL with DP.

Empowered by 5G, it has been extensively explored by existing works on the deployment of differentially private federated learning (DPFL) in the Industrial Internet of Things (IIoT). Through federated learning, decentralized IIoT devices can collaboratively train a machine learning model by merely exchanging model gradients with a parameter server (PS) for multiple global iterations. Differentially private (DP) mechanisms will be incorporated by IIoT devices (also called clients) to prevent the leakage of privacy due to the exposure of gradients because original gradients will be distorted DP noises. Yet, learning with distorted gradients can seriously deteriorate model accuracy, making DPFL unusable in reality. To address this problem, we propose a novel DPFL with sparse responses (DPFL-SR) algorithm, which applies the sparse vector technique to reduce the privacy budget consumption in each global iteration. Specifically, DPFL-SR evaluates the value of each gradient, and only distorts and uploads significant gradients to the PS because significant gradients are more essential for model training. Since insignificant gradients are not disclosed, the reserved privacy budget can be used to return significant gradients for more iterations so that DPFL-SR can achieve higher model accuracy without lowering the privacy protection level. Extensive experiments are conducted with the MNIST and Fashion-MNIST datasets to demonstrate the practicability and superiority of DPFL-SR in IIoT systems.

Given the rapid growth of user-generated videos, internet traffic has been heavily dominated by online video streaming. Caching videos on edge servers in close proximity to users has been an effective approach to reduce the backbone traffic and the request response time, as well as to improve the video quality on the user side. Video popularity, however, can be highly dynamic over time. The cost of cache replacement at edge servers, particularly that related to service interruption during replacement, is not yet well understood. This paper presents a novel lightweight video caching algorithm for edge servers, seeking to optimize the hit rate with real-time decisions and minimized cost. Inspired by recent advances in deep Q-learning, our DQN-based online video caching (DQN-OVC) makes effective use of the rich and readily available information from users and networks. We decompose the Q-value function as a product of the video value function and the action function, which significantly reduces the state space. We instantiate the action function for cost-aware caching decisions with low complexity so that the cached videos can be updated continuously and instantly with dynamic video popularity. We used video traces from Tencent, one of the largest online video providers in China, to evaluate the performance of our DQN-OVC and to compare it with state-of-the-art solutions. The results demonstrate that DQN-OVC significantly outperforms the baseline algorithms in the edge caching context.

The high-definition (HD) live video streaming has gained significant popularity due to the rapid growth of 4 G/5 G and social media. However, for devices with constrained bandwidth, they still have no sufficient bandwidth to support HD live video streaming. In this paper, we propose a neural-enhanced HD live video streaming framework calledLiveSRto provide universal HD live video streaming for both bandwidth-constrained and bandwidth-rich devices. For bandwidth-constrained devices, LiveSR delivers low-quality video streams and then boosts video quality at the device side with super-resolution (SR) techniques. The difficulty lies in how to train the SR model with low cost and conduct quality enhancement in real time. To address these challenges, we design a crowdsourced online training method by exploiting computation resources and HD video data on bandwidth-rich devices in the same video channel. We also propose an imitation learning-based decision making algorithm to make downloading decisions for video chunks and SR models under limited bandwidth. We implement and evaluate our proposed LiveSR framework using real network traces, and the experiment results show that LiveSR outperforms all the other baseline approaches, with 65.5% improvement in terms of the average QoE and 5.7% in terms of video quality (i.e., PSNR), and the achieved frame rate can be as high as 30 frames per second.

In vanilla Federated Learning (FL) systems, a centralized parameter server (PS) is responsible for collecting, aggregating and distributing model parameters with decentralized clients. However, the communication link of a single PS can be easily overloaded by concurrent communications with a massive number of clients. To overcome this drawback, multiple PSes can be deployed to form a parallel FL (PFL) system, in which each PS only communicates with a subset of clients and its neighbor PSes. On one hand, each PS conducts iterations with clients in its subset. On the other hand, PSes communicate with each other periodically to mix their parameters so that they can finally reach a consensus. In this paper, we propose a novel parallel federated learning algorithm called Fed-PMA, which optimizes such parallel FL under constrained communications by conducting parallel parameter mixing and averaging with theoretic guarantees. We formally analyze the convergence rate of Fed-PMA with convex loss, and further derive the optimal number of times each PS should mix with its neighbor PSes so as to maximize the final model accuracy within a fixed span of training time. Theoretical study manifests that PSes should mix their parameters more frequently if the connection between PSes is sparse or the time cost of mixing is low. Inspired by our analysis, we propose the Fed-APMA algorithm that can adaptively determine the near-optimal number of mixing times with non-convex loss under dynamic communication conditions. Extensive experiments with realistic datasets are carried out to demonstrate that both Fed-PMA and its adaptive version Fed-APMA significantly outperform the state-of-the-art baselines.

Due to the high cost of cellular networks, vehicle users would like to offload elastic traffic through vehicular networks as much as possible. This demand prompts researchers to consider how to make the vehicular network system achieve better performance for requests coming online, such as maximizing throughput. The traffic in vehicular networks is transferred through opportunistic contacts between vehicles and infrastructures. When making scheduling decisions, the scheduler must be aware of vehicles’ future trajectories. Vehicles’ future trajectories are usually predicted by trajectory prediction algorithms when users are unwilling to report their future trips. Unfortunately, no trajectory prediction algorithm can be completely accurate, and these inaccurate prediction results will degrade the throughput achieved by scheduling algorithms. In this paper, we focus on reducing the negative impact of inaccurate predictions. Specifically, we measure two data-driven trajectory prediction algorithms that have been widely used for trajectory predictions and understand the characteristics of the accuracy of predicted contacts. Based on the enlightenment from the measurement, we design a system, i.e., i-Offload, to offload elastic traffic under imperfect trajectory predictions. The experimental results show that our system has good throughput and high scheduling efficiency even under imperfect trajectory predictions. Compared with existing scheduling algorithms, our method improves the throughput by about one time.

Recently, federated learning (FL) has received tremendous attention in both academia and industry, in which decentralized clients collaboratively complete model training by exchanging model updates with a parameter server through the Internet. Its distributed nature well utilizes the localized data and preserves clients’ privacy, but also incurs heavy communication overhead. Existing studies on model update have mostly focused on the bandwidth constraint of the communication channels. Today's Internet however is highly unreliable. Simply using Transmission Control Protocol (TCP) would lead to low network utilization under frequent losses. In this paper, we closely examine the optimal transmission strategies in FL over the realistic lossy Internet. We systematically integrate model compression, forward error correction (FEC) and retransmission towards Federated Learning with Lossy Communications (FedLC). We derive the convergence rate of FedLC under non-convex loss with the optimal transmission. We then decompose this non-convex problem and present effective practical solutions. Public datasets are exploited for performance evaluation by varying the packet loss rate from 10% to 50%. In a fixed training time budget, FedLC can improve model accuracy by 3.91% on average or reduce the communication traffic by 34.27%-47.57% in comparison with state-of-the-art baselines.

Recently, federated learning (FL) has received intensive research because of its ability in preserving data privacy for scattered clients to collaboratively train machine learning models. Commonly, a parameter server (PS) is deployed for aggregating model parameters contributed by different clients. Decentralized federated learning (DFL) is upgraded from FL which allows clients to aggregate model parameters with their neighbours directly. DFL is particularly feasible for vehicular networks as vehicles communicate with each other in a vehicle-to-vehicle (V2V) manner. However, due to the restrictions of vehicle routes and communication distances, it is hard for individual vehicles to sufficiently exchange models with others. Data sources contributing to models on individual vehicles may not diversified enough resulting in poor model accuracy. To address this problem, we propose the DFL-DDS (DFL with diversified Data Sources) algorithm to diversify data sources in DFL. Specifically, each vehicle maintains a state vector to record the contribution weight of each data source to its model. The Kullback–Leibler (KL) divergence is adopted to measure the diversity of a state vector. To boost the convergence of DFL, a vehicle tunes the aggregation weight of each data source by minimizing the KL divergence of its state vector, and its effectiveness in diversifying data sources can be theoretically proved. Finally, the superiority of DFL-DDS is evaluated by extensive experiments (with MNIST and CIFAR-10 datasets) which demonstrate that DFL-DDS can accelerate the convergence of DFL and improve the model accuracy significantly compared with state-of-the-art baselines.

Federated Learning (FL) incurs high communication overhead, which can be greatly alleviated by compression for model updates. Yet the tradeoff between compression and model accuracy in the networked environment remains unclear and, for simplicity, most implementations adopt a fixed compression rate only. In this paper, we for the first time systematically examine this tradeoff, identifying the influence of the compression error on the final model accuracy with respect to the learning rate. Specifically, we factor the compression error of each global iteration into the convergence rate analysis under both strongly convex and non-convex loss functions. We then present an adaptation framework to maximize the final model accuracy by strategically adjusting the compression rate in each iteration. We have discussed the key implementation issues of our framework in practical networks with representative compression algorithms. Experiments over the popular MNIST and CIFAR-10 datasets confirm that our solution effectively reduces network traffic yet maintains high model accuracy in FL.

Eye tracking techniques have been widely adopted by a wide range of devices (e.g., AR/VR headsets, smartphones) to enhance user experiences. However, eye gaze data is private in nature, which can reveal users’ psychological and physiological features. Privacy protection techniques can be incorporated to preserve privacy of eye tracking information. Yet, most existing solutions based on Differential Privacy (DP) mechanisms cannot well protect privacy for individual users without sacrificing user experience. In this paper, we are among the first to propose a novel gaze model-based privacy control framework called Otus for eye tracking applications, which incorporates local DP (LDP) mechanisms to preserve user privacy and improves user experience in the meanwhile. First, we conduct a measurement study on real traces to illustrate that direct noise injection on raw gaze trajectories can significantly lower the utility of gaze data. To preserve utility and privacy simultaneously, Otus injects noises in two steps: (1) Extracting model features from raw data to depict gaze trajectories on individual users; (2) Adding LDP noises into model features so as to protect privacy. On one hand, established models can be used to recover user gaze data in order to improve service quality of eye tracking applications. On the other hand, we only need to add LDP noises to distort a small number of model parameters rather than every point on a trajectory to preserve privacy, which has less impact on the utility of gaze data given the same privacy budget. By applying the tile view graph model in step (1), we illustrate the entire workflow of Otus and prove its privacy protection level. For evaluation, we conduct extensive experiments using real gaze traces and the results show that Otus can effectively protect privacy for individual users without significantly compromising gaze data utility.

Edge computing architectures can help us quickly process the data collected by Internet of Things (IoT) and caching files to edge nodes can speed up the response speed of IoT devices requesting files. Blockchain architectures can help us ensure the security of data transmitted by IoT. Therefore, we have proposed a system that combines IoT devices, edge nodes, remote cloud, and blockchain. In the system, we designed a new algorithm in which blockchain-assisted compressed algorithm of federated learning is applied for content caching, called CREAT to predict cached files. In the CREAT algorithm, each edge node uses local data to train a model and then uses the model to learn the features of users and files, so as to predict popular files to improve the cache hit rate. In order to ensure the security of edge nodes’ data, we use federated learning (FL) to enable multiple edge nodes to cooperate in training without sharing data. In addition, for the purpose of reducing communication load in FL, we will compress gradients uploaded by edge nodes to reduce the time required for communication. What is more, in order to ensure the security of the data transmitted in the CREAT algorithm, we have incorporated blockchain technology in the algorithm. We design four smart contracts for decentralized entities to record and verify the transactions to ensure the security of data. We used MovieLens data sets for experiments and we can see that CREAT greatly improves the cache hit rate and reduces the time required to upload data.

Federated learning (FL) is an emerging paradigm through which decentralized devices can collaboratively train a common model. However, a serious concern is the leakage of privacy from exchanged gradient information between clients and the parameter server (PS) in FL. To protect gradient information, clients can adopt differential privacy (DP) to add additional noises and distort original gradients before they are uploaded to the PS. Nevertheless, the model accuracy will be significantly impaired by DP noises, making DP impracticable in real systems. In this work, we propose a novel noise information secretly sharing (NISS) algorithm to alleviate the disturbance of DP noises by sharing negated noises among clients. We theoretically prove that: 1) if clients are trustworthy, DP noises can be perfectly offset on the PS and 2) clients can easily distort negated DP noises to protect themselves in case that other clients are not totally trustworthy, though the cost lowers model accuracy. NISS is particularly applicable for FL across multiple Internet of Things (IoT) systems, in which all IoT devices need to collaboratively train a model. To verify the effectiveness and the superiority of the NISS algorithm, we conduct experiments with the MNIST and CIFAR-10 data sets. The experimental results verify our analysis and demonstrate that NISS can improve model accuracy by 19% on average and obtain better privacy protection if clients are trustworthy.

Recently, blockchain-based federated learning (BFL) has attracted intensive research attention due to that the training process is auditable and the architecture is serverless avoiding the single point failure of the parameter server in vanilla federated learning (VFL). Nevertheless, BFL tremendously escalates the communication traffic volume because all local model updates (i.e., changes of model parameters) obtained by BFL clients will be transmitted to all miners for verification and to all clients for aggregation. In contrast, the parameter server and clients in VFL only retain aggregated model updates. Consequently, the huge communication traffic in BFL will inevitably impair the training efficiency and hinder the deployment of BFL in reality. To improve the practicality of BFL, we are among the first to propose a fast blockchain-based communication-efficient federated learning framework by compressing communications in BFL, called BCFL. Meanwhile, we derive the convergence rate of BCFL with non-convex loss. To maximize the final model accuracy, we further formulate the problem to minimize the training loss of the convergence rate subject to a limited training time with respect to the compression rate and the block generation rate, which is a bi-convex optimization problem and can be efficiently solved. To the end, to demonstrate the efficiency of BCFL, we carry out extensive experiments with standard CIFAR-10 and FEMNIST datasets. Our experimental results not only verify the correctness of our analysis, but also manifest that BCFL can remarkably reduce the communication traffic by 95–98% or shorten the training time by 90–95% compared with BFL.

A Covert Electricity-Theft Cyberattack Against Machine Learning-Based Detection Models

With the growth of participating clients, the centralized parameter server (PS) will seriously limit the scale and efficiency of Federated Learning (FL). A straightforward approach to scale up the FL system is to construct a Parallel FL (PFL) system with multiple parallel PSes. However, it is unclear whether PFL can really accelerate FL or reduce the training time of FL. Even if the answer is yes, it is non-trivial to design a highly efficient parameter average algorithm for a PFL system. In this paper, we propose a completely parallelizable FL algorithm called P-FedAvg under the PFL architecture. P-FedAvg extends the well-known FedAvg algorithm by allowing multiple PSes to cooperate and train a learning model together. In P-FedAvg, each PS is only responsible for a fraction of total clients, but PSes can mix model parameters in a dedicatedly designed way so that the FL model can well converge. Different from heuristic-based algorithms, P-FedAvg is with theoretical guarantees. To be rigorous, we theoretically analyze the convergence rate of P-FedAvg in terms of the number of conducted iterations, the communication cost of each global iteration and the optimal weights for each PS to mix parameters with its neighbors. Based on theoretical analysis, we conduct a case study on five typical overlay topolgoies formed by PSes to further examine the communication efficiency under different topologies, and investigate how the overlay topology affects the convergence rate, communication cost and robustness of a PFL system. Lastly, we perform extensive experiments with real datasets to verify our analysis and demonstrate that P-FedAvg can significantly speed up FL than traditional FedAvg and other competitive baselines. We believe that our work can help to lay a theoretical foundation for building more efficient PFL systems.

Today’s Internet traffic has been dominated by video contents. To efficiently serve online videos for millions of users, it is essential to cache frequently requested videos on various devices such as edge servers, personal computers, etc. Existing caching algorithms are mainly designed by leveraging the information of past request records. However, it is insufficient to only use simple statistics of past request records,e.g., video popularity which ignores the occurrence time of past video request events and the discrepancies among individual devices. In this paper, we propose a radically different learning-based video caching algorithm for Internet devices, calledPPVC (Point Process based Video Cache)which can take exact video request patterns into account. By utilizing Hawkes Process (HP), we are able to link the future video request rates of a device with three kinds of historical records, namely, historical requests for the same video launched by the device itself, historical requests for the same video from other devices with a similar request pattern, and historical requests for other similar videos from the same device. The video request patterns can be efficiently computed via Singular Value Decomposition (SVD). Parameters linking these historical events can be determined by maximizing the likelihood of historical events. To further reduce the computation load, we also propose an online version of PPVC, which can timely update cached videos with the incremental arrival of events. One nice property of PPVC is that it can adapt to the change of video popularity very fast and is also applicable to Internet devices at different levels, ranging from the high-level servers that serve a large population of users to the low-level user devices (e.g., personal computers). Finally, we conduct extensive simulations with real traces and the results show that PPVC can always achieve the best video caching performance in terms of the hit rate on all levels of Internet devices, and quickly adapt to the dynamics of video requests.

Federated learning (FL) emerges as a promising paradigm to enable a federation of clients to train a machine learning model in a privacy-preserving manner. Most existing works assumed that the central parameter server (PS) determines the participation of clients implying that clients cannot make autonomous participation decisions. The above assumption is unrealistic because the participation in FL training may incur various cost and clients also have strong desire to be rewarded for participation. To address this problem, we design a novel autonomous client participation scheme to incentivize clients. Specifically, the PS provides a certain reward shared among participating clients for each training round. Clients decide whether to participate each FL training round or not based on their own utilities (i.e., reward minus cost). The process can be modeled as a minority game (MG) with incomplete information and clients end up in the minority side win after each training round because the reward of each participating client may not cover its cost if too many clients participate and vice verse. The challenge of autonomous participation schemes lies in lowering thevolatilityof participating clients in each round due to the lack of coordination among clients. Through solid analysis, we prove that: 1) The volatility of participating clients in each round is very high under the standard MG scheme. 2) The volatility of participating clients can be reduced significantly under the stochastic MG scheme. 3) A coalition based MG is proposed, which can further reduce the volatility in each round. By conducting extensive experiments in real settings, we demonstrate that the stochastic MG-based scheme outperforms other state-of-the-art algorithms in terms of utility and volatility, and the coalition MG-based client participation scheme can further boost the utility by 39%-48% and reduce the volatility by 51%–100%. Moreover, our algorithms can achieve almost the same model accuracy as that obtained by centralized client participation algorithms.

It is always a challenging problem to deliver a huge volume of videos over the Internet. To meet the high bandwidth and stringent playback demand, one feasible solution is to cache video contents on edge servers based on predicted video popularity. Traditional caching algorithms (e.g., LRU, LFU) are too simple to capture the dynamics of video popularity, especially long-tailed videos. Recent learning-driven caching algorithms (e.g., DeepCache) show promising performance, however, such black-box approaches are lack of explainability and interpretability. Moreover, the parameter tuning requires a large number of historical records, which are difficult to obtain for videos with low popularity. In this paper, we optimize video caching at the edge using a white-box approach, which is highly efficient and also completely explainable. To accurately capture the evolution of video popularity, we develop a mathematical model calledHRSmodel, which is the combination of multiple point processes, including Hawkes’ self-exciting, reactive and self-correcting processes. The key advantage of the HRS model is its explainability, and much less number of model parameters. In addition, all its model parameters can be learned automatically through maximizing the Log-likelihood function constructed by past video request events. Next, we further design an online HRS-based video caching algorithm. To verify its effectiveness, we conduct a series of experiments using real video traces collected from Tencent Video, one of the largest online video providers in China. Experiment results demonstrate that our proposed algorithm outperforms the state-of-the-art algorithms, with 15.5% improvement on average in terms of cache hit rate under realistic settings.

In today's Internet, recommender systems play an indispensable role in helping users discover items of interests, such as products, books, movies and so on. However, a higher recommendation accuracy is commonly at the cost of more disclosure of user privacy. Thus, a wider adoption of recommender systems poses significant security and privacy concerns to users. In this article, we propose a light-weight privacy-preserving framework calledRAPfor recommender systems, which can protect user privacy while still ensuring a high recommendation accuracy. Instead of directly sending users’ private ratings to the recommender, users first conduct a local perturbation operation on private ratings, and then send the perturbed ratings to the recommender. The recommender can run recommendation algorithms directly over the perturbed ratings and return the results to users. Different from crypto-based methods, our perturbation and de-perturbation methods are linear operations. Thus,RAPis light-weight and highly efficient in privacy protection. To be more rigorous, we formally prove that the order of recommendation accuracy will not decrease when ourRAPframework is applied to any MF (Matrix Factorization)-based recommender systems. We also derive the closed-form expression for the degree of privacy preservation of our framework. Finally, we conduct extensive evaluations using large-scale real-world datasets to verify the effectiveness of ourRAPframework and compare with other baseline algorithms. The results show that ourRAPframework can improve the degree of privacy preservation from zero to over 0.5 for theMovielensdataset and 4 for theJesterdataset, and still maintain the approaching level of recommendation accuracy.

With the growth of participating clients, the centralized parameter server (PS) will seriously limit the scale and efficiency of Federated Learning (FL). A straightforward approach to scale up the FL system is to construct a Parallel FL (PFL) system with multiple PSes. However, it is unclear whether PFL can really achieve a faster convergence rate or not. Even if the answer is yes, it is non-trivial to design a highly efficient parameter average algorithm for a PFL system. In this paper, we propose a completely parallelizable FL algorithm called P-FedAvg under the PFL architecture. P-FedAvg extends the well-known FedAvg algorithm by allowing multiple PSes to cooperate and train a learning model together. In P-FedAvg, each PS is only responsible for a fraction of total clients, but PSes can mix model parameters in a dedicatedly designed way so that the FL model can well converge. Different from heuristic-based algorithms, P-FedAvg is with theoretical guarantees. To be rigorous, we conduct theoretical analysis on the convergence rate of P-FedAvg, and derive the optimal weights for each PS to mix parameters with its neighbors. We also examine how the overlay topology formed by PSes affects the convergence rate and robustness of a PFL system. Lastly, we perform extensive experiments with real datasets to verify our analysis and demonstrate that P-FedAvg can significantly improve convergence rates than traditional FedAvg and other competitive baselines. We believe that our work can help to lay a theoretical foundation for building more efficient PFL systems.

The rapidly evolving technology of edge computing shows great potential in revolutionizing the market of cloud gaming. Edge computing can significantly lower the latency for better gaming experiences by performing computation at the proximity of game players. However, the acceleration of latency-sensitive cloud gaming services at the edge is challenging due to the heterogeneity of edge servers, different requirements among game players, and so on. In this paper, we propose an efficient latency-sensitive scheduling algorithm called EGSA to satisfy latency constraints for cloud gaming services at the edge. We first formulate the problem as a weighted congestion game, which takes a number of key factors (e.g., game genres, user strategies, latency constraint and device heterogeneity) into account. Based on the weighted congestion game model, we further design an efficient latency-sensitive scheduling algorithm, which can approximate the pure Nash equilibrium under Shapley cost-sharing method. We also perform theoretic analysis to prove that our proposed algorithm converges in polynomial steps. Finally, we conduct a set of experiments and the results show that our algorithm outperforms alternative strategies with up to 46% performance improvement.

The prevalence of personal devices motivates the rapid development of crowdsourced livecast in recent years. However, there exists huge diversity of upstream bandwidth among amateur broadcasters. Moreover, the highest video quality that can be streamed is limited by the hardware configuration of broadcaster devices (e.g., 540p for low-end mobile devices). The above factors pose significant challenges to the ingestion of high-resolution live video streams, and result in poor quality-of-experience (QoE) for viewers. In this paper, we propose a novel live video ingest approach called CrowdSR for crowdsourced livecast. CrowdSR can transform a low-resolution video stream uploaded by weak devices into a high-resolution video stream via super-resolution, and then deliver the stream to viewers. CrowdSR can exploit crowdsourced high-resolution video patches from similar broadcasters to speedup model training. Different from previous work, our approach does not require any modification at the client side, and thus is more practical and easy to implement. Finally, we implement and evaluate CrowdSR by conducting a series of real-world experiments. The results show that CrowdSR significantly outperforms the baseline approaches by 0.42-1.09 dB in terms of PSNR and 0.006-0.014 in terms of SSIM.

Variable Bitrate (VBR) video encoding can provide much high quality-to-bits ratio compared to the widely adopted Constant Bitrate (CBR) encoding, and thus receives significant attentions by content providers in recent years. However, it is challenging to design efficient adaptive bitrate algorithms for VBR-encoded videos due to the sharply fluctuating chunk size and the resulting bitrate burstiness. In this paper, we propose a neural adaptive streaming framework called Vibra for VBR-encoded videos, which can well accommodate the high fluctuation of video chunk sizes and improve the quality-of-experience (QoE) of end users significantly. Our framework takes the characteristics of VBR-encoded videos into account, and adopts the technique of deep reinforcement learning to train a model for bitrate adaptation. We also conduct extensive trace-driven experiments, and the results show that Vibra outperforms the state-of-the-art ABR algorithms with an improvement of 8.17% -- 29.21% in terms of the average QoE.

With recent advances in autonomous driving, voice control systems have become increasingly adopted as human-vehicle interaction methods. This technology enables drivers to use voice commands to control the vehicle and will be soon available in Advanced Driver Assistance Systems (ADAS). Prior work has shown that Siri, Alexa and Cortana, are highly vulnerable to inaudible command attacks. This could be extended to ADAS in real-world applications and such an inaudible command threat is difficult to detect due to microphone nonlinearities. In this paper, we aim to develop a more practical solution by using camera views to defend against inaudible command attacks where ADAS are capable of detecting their environment via multi-sensors. To this end, we propose a novel multimodal deep learning classification system to defend against inaudible command attacks. Our experimental results confirm the feasibility of the proposed defense methods and the best classification accuracy reaches 89.2%. Code is available at https://github.com/ITSEG-MQ/Sensor-Fusion-Against-VoiceCommand-Attacks.

In federated learning, machine learning and deep learning models are trained globally on distributed devices. The state-of-the-art privacy-preserving technique in the context of federated learning is user-level differential privacy. However, such a mechanism is vulnerable to some specific model poisoning attacks such as Sybil attacks. A malicious adversary could create multiple fake clients or collude compromised devices in Sybil attacks to mount direct model updates manipulation. Recent works on novel defense against model poisoning attacks are difficult to detect Sybil attacks when differential privacy is utilized, as it masks clients' model updates with perturbation. In this work, we implement the first Sybil attacks on differential privacy based federated learning architectures and show their impacts on model convergence. We randomly compromise some clients by manipulating different noise levels reflected by the local privacy budget ε of differential privacy with Laplace mechanism on the local model updates of these Sybil clients. As a result, the global model convergence rates decrease or even leads to divergence. We apply our attacks to two recent aggregation defense mechanisms, called Krum and Trimmed Mean. Our evaluation results on the MNIST and CIFAR-10 datasets show that our attacks effectively slow down the convergence of the global models. We then propose a method to keep monitoring the average loss of all participants in each round for convergence anomaly detection and defend our Sybil attacks based on the training loss reported from randomly selected sets of clients as the judging panels. Our empirical study demonstrates that our defense effectively mitigates the impact of our Sybil attacks.

Edge computing is promising to solve the latency issue in the Internet of Vehicles (IoV). However, due to decentralization, traditional edge computing suffers in management, deployment, and security. Containerization relaxes resource deployment and migration problems, but current container scheduling policies are inefficient to process complicated tasks based on directed acyclic graph or DAG structures. In this article, we design a containerized edge computing platform CUTE, which provides low-latency computation services for the Internet of Vehicles. The centralized controller is empowered with resource management and orchestration, and containers are scheduled to appropriate edge servers to optimize the computation delay. CUTE is also integrated with blockchain to improve network security. We formulate the vehicle task offloading and container scheduling problems and develop a heuristic container scheduling algorithm for DAG-based computation tasks submitted by vehicles remotely. We implement and deploy CUTE into the China Mobile Network, and conduct comprehensive experiments and a case study. The experiment results show that CUTE can provide low-latency computation services for vehicular applications and that the heuristic algorithm outperforms traditional container scheduling policies.

Federated learning (FL) is a promising new technology in the field of IoT intelligence. However, exchanging model-related data in FL may leak the sensitive information of participants. To address this problem, we propose a novel privacy-preserving FL framework based on an innovative chained secure multiparty computing technique, named chain-PPFL. Our scheme mainly leverages two mechanisms: 1) single-masking mechanism that protects information exchanged between participants and 2) chained-communication mechanism that enables masked information to be transferred between participants with a serial chain frame. We conduct extensive simulation-based experiments using two public data sets (MNIST and CIFAR-100) by comparing both training accuracy and leak defence with other state-of-the-art schemes. We set two data sample distributions (IID and NonIID) and three training models (CNN, MLP, and L-BFGS) in our experiments. The experimental results demonstrate that the chain-PPFL scheme can achieve practical privacy preservation (equivalent to differential privacy with ∈ approaching zero) for FL with some cost of communication and without impairing the accuracy and convergence speed of the training model.

Federated Learning (FL) is an emerging decentralized learning framework through which multiple clients can collaboratively train a learning model. However, a major obstacle that impedes the wide deployment of FL lies in massive communication traffic. To train high dimensional machine learning models (such as CNN models), heavy communication traffic can be incurred by exchanging model updates via the Internet between clients and the parameter server (PS), implying that the network resource can be easily exhausted. Compressing model updates is an effective way to reduce the traffic amount. However, a flexible unbiased compression algorithm applicable for both uplink and downlink compression in FL is still absent from existing works. In this work, we devise the Model Update Compression by Soft Clustering (MUCSC) algorithm to compress model updates transmitted between clients and the PS. In MUCSC, it is only necessary to transmit cluster centroids and the cluster ID of each model update. Moreover, we prove that: 1) The compressed model updates are unbiased estimation of their original values so that the convergence rate by transmitting compressed model updates is unchanged; 2) MUCSC can guarantee that the influence of the compression error on the model accuracy is minimized. Then, we further propose the boosted MUCSC (B-MUCSC) algorithm, a biased compression algorithm that can achieve an extremely high compression rate by grouping insignificant model updates into a super cluster. B-MUCSC is suitable for scenarios with very scarce network resource. Ultimately, we conduct extensive experiments with the CIFAR-10 and FEMNIST datasets to demonstrate that our algorithms can not only substantially reduce the volume of communication traffic in FL, but also improve the training efficiency in practical networks.

360-degree video streaming shows great potential to revolutionize the streaming market, by providing much better immersive experience than standard video streams. However, its wide adoption is hindered by the surging demand of network bandwidth due to multi-screen video transmission. To reduce the bandwidth cost, one promising approach is to predict a user’s field of view (FoV), and then prefetch video tiles that a user will view a few seconds ahead. The challenge lies in that user behaviors cannot be properly captured with very limited information, especially the viewing time spent on each tile and the FoV switching behavior are hard to predict. In this paper, we propose a novel 360-degree video streaming algorithm calledTVG-Streamingto optimize user experiences by learning user view behaviors. Different from previous approaches, our idea is to exploit tile-view graphs (TVGs) generated by real user behaviors and accurately estimate the probability that each tile falls in the FoV. With the tile view probability, we can determine the bitrate of each tile for delivery and buffering with limited bandwidth budget so as to maximize users’ quality of experience (QoE). For evaluation, we conduct extensive experiments using real traces and the results show that our proposedTVG-Streamingalgorithm significantly outperforms other algorithms by at least 20% improvement in terms of users’ QoE.

Nowadays, surveillance cameras have been pervasively equipped with vehicles in public transport systems. For the sake of public security, it is crucial to upload recorded surveillance videos to remote servers timely for backup and necessary video analytics. However, continuously uploading video content generated by tens of thousands of vehicles can be extremely bandwidth consuming. In this work, we investigate the video uploading problem for moving buses by proposing to deploy dedicated access points (AP) at bus stops to facilitate video uploading. We define the harmonic objective for our problem, which includes minimizing the video uploading delay and minimizing the AP deployment cost. This problem is with two fundamental challenges. Firstly, it is difficult to balance the bandwidth capacity allocated to many buses because a bus obtains bandwidth resource from a series of APs deployed at stops along its route. Secondly, due to the randomness of bus movement and the complexity of bus routes, it is hard to predict the workload of an AP. Hence, it is challenging to estimate the delay of uploading video content through an AP. To cope with these challenges, we propose a water filling placement (WFP) algorithm, aiming to balance the aggregated bandwidth allocated to each bus. A queuing model is established to analyze the uploading delay of video content. We further resort to machine learning models to factor the influence of bus routes into our queuing model. Finally, a convex problem is formulated to optimize the harmonic objective, which can be optimally solved with the gradient descent (GD) based algorithm. We validate the correctness of our theoretical analysis and demonstrate the effectiveness of our method by carrying out extensive experiments using bus traces collected in Shenzhen city of China. In comparison with benchmark algorithms, our solution can always achieve the best performance.

With the development of smart cities, Internet services will be pervasively accessible for moving vehicles. It is envisioned that the video content demand of vehicles will explode in the near future. However, the strategy to efficiently distribute video content in large-scale vehicular networks is still absent due to challenges arising from the huge video population, heavy bandwidth consumption, heterogeneous user devices, and vehicles’ mobility. In this work, we propose to collaboratively replicate video content on Roadside Units (RSUs) to enhance video distribution services based on the fact that the contact period between moving vehicles and a single RSU is not long enough to complete video downloading. In our design, a video file is split into multiple chunks. Each RSU replicates a small number of original chunks and chunks encoded by network coding. Replicating encoded chunks can reduce redundancy of chunks on different RSUs so that RSUs can complement each other better, whereas original chunks can be transrated to chunks with lower bitrates flexibly to fit in users’ devices. Therefore, we replicate both original and encoded chunks on RSUs to take advantages of both sides. Stochastic models are employed to analyze chunk download processes and a convex optimization problem is formulated to determine the optimal partition of space allocated to each kind of chunks. Furthermore, we extend our strategy to support video streaming services and empirically prove that the influence caused by limitations of network coding is moderate. In the end, we conduct extensive simulations which not only validate the accuracy of our models but also demonstrate that our strategy can effectively boost video distribution services.

In 360-degree video streaming, users commonly watch a video scene within aField of View (FoV). Such observation provides an opportunity to save bandwidth consumption by predicting and then prefetching video tiles within the FoV. However, existing FoV prediction methods seldom consider the diversity among user behaviors and the impact of different video genres. Thus, previous one-size-fits-all models cannot make accurate prediction for users with different behavior patterns. In this paper, we propose a user-aware viewport prediction algorithm calledSparkle, which is a practical whitebox approach for FoV prediction. Instead of training a single learning model to predict the behaviors for all users, our proposed algorithm is tailored to fit each individual user. In particular, unlike other learning models, our prediction model is completely explainable and all the parameters have their physical meanings. We first conduct a measurement study to analyze real user behaviors and observe that there exists sharp fluctuation of view orientation and user posture has significant impact on the viewport movement of users. Moreover, cross-user similarity is diverse across different video genres. Inspired by these insights, we further design a user-aware viewport prediction algorithm by mimicking a user's viewport movement on the tile map, and determine how a user will change the viewport angle based on his (or her) trajectory and other similar users’ behaviors in the past time window. Extensive evaluations with real datasets demonstrate that, our proposed algorithm significantly outperforms the state-of-the-art benchmark methods (e.g., LSTM-based methods) by over$\text{5}\%$, and the prediction accuracy is much more stable on various types of 360-degree videos than previous methods.

Recently, Federated Learning (FL) has drawn tremendous attentions due to its ability to protect client's privacy. In FL, clients collaboratively train machine learning models by merely sharing intermediate computations, i.e., gradients of model parameters. However, training a complicated model involves multiple rounds of interactions between clients and the server via the Internet. Consequently, communication is a primary bottleneck of FL attributed to the poor network conditions and the large amount of interchanged computations. To overcome the communication bottleneck, we propose the ClusterGrad algorithm to compress gradients which can considerably reduce the volume of communicated computations. Our design is based on the fact that there is only a small fraction of gradients whose values are far away from the origin in each round of interaction in FL. We first identify these essential gradients that are far away from 0 using the K-means algorithm. These gradient values are approximated by a novel clustering based quantization algorithm. Then, the rest gradients lying close to 0 are approximated with a single value. We can prove that ClusterGrad outperforms the latest FL gradient compression algorithms: Probability Quantization (PQ) and Deep Gradient Compression (DGC). We conduct extensive experiments with the CIFAR-10 datasets which further demonstrate that ClusterGrad can achieve compression ratio (used interchangeably with compression rate) 123 on average in comparison with PQ and DGC with compression ratios 16 and 60 respectively.

The following topics are dealt with: cloud computing; Internet of Things; learning (artificial intelligence); virtual machines; mobile computing; security of data; resource allocation; scheduling; software engineering; regression analysis.

Person Re-Identification (Re-ID) aims to match the persons contained in surveillance videos, and is usually run on powerful servers in a supervised mode. However, centralized processing of massive video from thousands of cameras in a city is very costly and causes serious problems of privacy protection. Moreover, the labeling of numerous data for supervised training is also infeasible in this scenario. To address this problem, we propose a novel Self-optimizing Memory Network model, namely SoMem, which runs person Re-ID on edge devices in a totally unsupervised and distributed way. Specifically, SoMem adopts a random walk based collaborative training procedure to optimize the visual model on each camera based on locally collected images, and builds a distributed memory network to memorize and match the observed persons by using a distributed mutual ranking algorithm. Based on the cross-camera person matching results learned by the memory network, the visual models on edge devices are further optimized in a self-organized manner. Comprehensive experiments are conducted on several real person Re-ID datasets and deployed on edge devices to show the effectiveness and efficiency of this novel distributed Re-ID model.

Recent years have witnessed great success of mobile short-form video apps. However, most current video streaming strategies are designed for long-form videos, which cannot be directly applied to short-form videos. Especially, short-form videos differ in many aspects, such as shorter video length, mobile friendliness, sharp popularity dynamics, and so on. Facing these challenges, in this paper, we perform an in-depth measurement study on Douyin, one of the most popular mobile short-form video platforms in China. The measurement study reveals that Douyin adopts a rather simple strategy (called Next-One strategy) based on HTTP progressive download, which uses a sliding window with stop-and-wait protocol. Such a strategy performs poorly when network connection is slow and user scrolling is fast. The results motivate us to design an intelligent adaptive streaming scheme for mobile short-form videos. We formulate the short-form video streaming problem and propose an adaptive short-form video streaming strategy called LiveClip using a deep reinforcement learning (DRL) approach. Trace-driven experimental results prove that LiveClip outperforms existing state-of-the-art approaches by around 10%-40% under various scenarios.

Federated collaborative filtering (Fed-CF) is a variant of federated learning (FL) models, which can protect user privacy in recommender systems. In Fed-CF, the recommendation model is collectively trained across multiple decentralized clients by exchanging gradients only. However, the decentralized nature of Fed-CF makes it vulnerable to shilling attacks, which can be realized by inserting fake ratings of target items to distort recommendation results. Unfortunately, previous detection algorithms cannot work well in the FL framework, as all original data samples are not disclosed at all. In this paper, we are the first to systematically study the problem of shilling attacks in the context of federated learning, and propose an effective detection method called Federated Shilling Attack Detector (FSAD) to detect shilling attackers in Fed-CF. We first show the feasibility of shilling attacks in Fed-CF. Next, we dedicatedly design four novel features based on exchanged gradients among clients. By incorporating these gradient-based features, we train a semi-supervised Bayes classifier to identify shilling attackers effectively. Finally, we conduct extensive experiments based on real-world datasets to evaluate the performance of our proposed FSAD method. The experimental results show that FSAD can detect shilling attackers in Fed-CF with high accuracy, with the F1value as high as 0.90 on the Netflix dataset, which approaches the performance of the optimal detector that utilizes complete private user information for detection.

Task offloading is one of key operations in edge computing, which is essential for reducing the latency of task processing and boosting the capacity of end devices. However, the heterogeneity among tasks generated by various users makes it challenging to design efficient task offloading algorithms. In addition, the assumption of complete information for offloading decision-making does not always hold in a distributed edge computing environment. In this article, we formulate the problem of heterogeneous task offloading in a distributed environment as a minority game (MG), in which each player must make decisions independently in each turn and the players who end up on the minority side win. The multi-player MG incentivizes players to cooperate with each other in the scenarios with incomplete information, where players don't have full information about other players (e.g., the number of tasks, the required resources). To address the challenges incurred by task heterogeneity and the divergence of naive MG approaches, we propose an MG based scheme, in which tasks are divided into subtasks and instructed to form into a set of groups as possible, and the left ones are scheduled to perform decision adjustment in a probabilistic manner. We prove that our proposed algorithm can converge to a near-optimal point, and also investigate its stability and price of anarchy in terms of task processing time. Finally, we conduct a series of simulations to evaluate the effectiveness of our proposed scheme and the results indicate that our scheme can achieve around 30% reduction of task processing time compared with other approaches. Moreover, our proposed scheme can converge to a near-optimal point, which cannot be guaranteed by naive MG approaches.

Confronting with the sheer amount of Over-the-Top (OTT) videos, personalized recommendation is especially important for users to locate videos of interest. However, previous approaches seldom considered the influence of watching time when designing video recommendation algorithms. In this paper, we first conduct a detailed measurement study on a leading OTT video service provider in China and our results show that user view preferences are substantially influenced by watching time. Based on the above results, we further propose a personalized time-aware video recommendation algorithm called TAMF for OTT videos. The basic idea of our proposed TAMF algorithm is to utilize matrix factorization to unveil how watching time affects user view interests and cluster time slots with similar influence. In this way, we can collaboratively learn users' personal interests if their views belong to the same cluster, and precisely capture user view preferences with watching time. Finally, we also conduct extensive experiments using real traces to evaluate the performance of our algorithm, and the experimental results show that our proposed algorithm can improve video recommendation performance by 4.83% and 4.42% in terms of WMRR and WMAP respectively and significantly boost user engagement.

As more and more applications need to analyze unbounded data streams in a real-time manner, data stream processing platforms, such as Storm, have drawn the attention of many researchers, especially the scheduling problem. However, there are still many challenges unnoticed or unsolved. In this paper, we propose and implement an adaptive online scheme to solve three important challenges of scheduling. First, how to make a scaling decision in a real-time manner to handle the fluctuant load without congestion? Second, how to minimize the number of affected workers during rescheduling while satisfying the resource demand of each instance? We also point out that the stateful instances should not be placed on the same worker with stateless instances. Third, currently, the application performance cannot be guaranteed because of resource contention even if the computation platform implements an optimal scheduling algorithm. In this paper, we realize resource isolation using Cgroup, and then the performance interference caused by resource contention is mitigated. We implement our scheduling scheme and plug it into Storm, and our experiments demonstrate in some respects our scheme achieves better performance than the state-of-the-art solutions.

Edge computing emerges as a promising paradigm to decentralize computation power to the edge of the network and thus improve user experience by task offloading. A user can perfectly schedule his tasks to be executed on edge servers if the execution time of all tasks can be known beforehand. However, it is difficult to know the task execution time (TET) before performing actual offloading, which normally varies on edge servers with different software and hardware configurations. Moreover, such configuration information is not always available to end users due to security concerns. In this paper, we first propose a learning-driven algorithm to accurately predict TETs of all tasks in such an asymmetrically informed edge computing environment. The basic idea is to predict unknown TETs using only a small sampled set of TETs by exploiting the underlying correlation between TETs and edge server configurations. Next, we formulate the problem of task offloading into a constrained optimization problem, which is unfortunately proved to be NP-hard. To address the above challenge, we design a task offloading algorithm, called Maximum Efficiency First Ordered (MEFO), to achieve near-optimal efficiency. Field measurements and experiments have been conducted to demonstrate that our proposed learning-driven algorithm can predict TETs more accurately than other algorithms as long as the fraction of sampled TETs is larger than a small predefined threshold, and our proposed MEFO algorithm achieves a much higher success rate of task offloading and a shorter processing delay with very limited information of edge servers.

With increasing development of Synthetic Aperture Radar (SAR) jamming technology, the jamming effects analysis for some innovative SAR, such as Multi-channel and Multi Input and Multi Output (MIMO) SAR, is significant. The geometry model of Multi-Channel and MIMO SAR are introduced. Meanwhile, the jamming blanket factor is researched and the barrage jamming is simulated. Simulations demonstrated jamming effects for both Multi-Channel and MIMO SAR.

With the advance of cloud computing and 4G/5G technology, video contents recorded by in-car cameras (i.e., vehicular digital video recorders) can be uploaded to the cloud to facilitate accident analysis, online surveillance, video sharing, etc. However, the cost of uploading such huge volume of video contents via unstable vehicular access networks (including cellular base stations and road-side units) can be considerable by considering the increasing video quality requirement, time constraint, and limited local buffer space. In this paper, we propose an adaptive video recording and uploading scheme to maximize the overall utility of cloud-based in-car video uploading over vehicular access networks. Specifically, the utility function is defined as the weighted sum of bandwidth cost and video quality and we formulate the problem into a constrained Markov decision process (MDP). Based on the theoretic foundation of MDP, we design and implement an algorithm to obtain an adaptive chunk uploading policy for video contents over vehicular access networks. Extensive simulations have been conducted to demonstrate that our policy can achieve the best performance compared with other alternative strategies.

Recently, crowdsourcing-based content delivery networks (CDN) emerge as a promising technology that can distribute massive video content to a vast number of Internet users by crawling bandwidth and storage resources from Internet end devices. Any ordinary Internet users with excessive resources can be recruited into such systems as mini-servers. Different from edge servers equipped with dedicated resources in traditional CDNs, the resource of a single mini-server is scarce and volatile that can vary severely with time, since its bandwidth is shared by many different applications. How to build a robust high performance crowdsourcing-based CDN system has attracted contributions from both academia and industry, but how to solve the drawback caused by unstable uploading bandwidth is still a challenging problem. So far, a prevalent methodology is to migrate the strategies implemented by traditional CDNs into crowdsourcing-based CDN systems based on the fact that these two kinds of systems share many similarities. In this paper, our argument is that the content delivery time can be reduced by replicating coded content on mini-servers (which is almost useless for edge servers in traditional CDNs) to enable downloading users to automatically adapt their downloading progress with oscillating bandwidth capacity from different mini-servers. Theoretical model is created to derive the performance improvement (evaluated in term of average file downloading time) achieved by our strategy, which is further validated via simulation. This paper not only provides system designers a more efficient content replication solution, but also can push forward the development of the crowdsourcing-based CDNs.

Delivering high-definition (HD) videos to a large number of Internet users is a challenging research problem due to its heavy bandwidth consumption and inelastic quality-of-service (QoS) requirement. Different from the traditional content delivery networks and overlay peer-to-peer networks, crowdsourcing-based platforms, e.g., Thunder Crystal, deliver HD videos by renting agents' bandwidth and storage resources. Cash will be rewarded to agents based on agents' upload traffic. Online video providers, i.e., Tencent Video and YouKu, pay Thunder Crystal for its video distribution service. Therefore, a critical problem for Thunder Crystal is to evaluate the performance it can achieve, which determines Thunder Crystal's competitiveness and its bargaining power with online video providers. Although previously studied proportional video replication and random request scheduling strategy are implemented by Thunder Crystal, the system performance cannot be evaluated by simply using existing models, because of its novel business model. To address this problem, this paper proposes a theoretical framework that can analyze the performance for synchronized streaming, video-on-demand (VoD) streaming, and video downloading, which are all supported by Thunder Crystal. A differentiated bandwidth allocation is designed to boost Thunder Crystal's streaming performance by assigning downloading users more fluctuating bandwidth, which only slightly degrades downloading performance. Finally, simulation is conducted to validate the accuracy of our theoretical results.

The knowledge of view preferences of users is crucial for online video providers to improve their system operations and video recommendations. However, it is challenging to accurately acquire this knowledge by merely relying on a single online video system. In this paper, we conduct a joint statistical study using the cross-platform information obtained from Douban, the largest online video database with video rating functionality in China, and Youku, one of the largest online video streaming systems in China. The Douban dataset includes feedbacks (e.g., movie ratings, comments, and reviews) from all users of different online video systems, and movie metadata (e.g., release date, actors, and directors), based on which we can statistically explore effective and significant factors attributing to video view counts. Meanwhile, our study unveils user behaviors that are latent when only observing a single video system. Finally, a multiple correlation analysis reveals that factors extracted from Douban can significantly increase our ability to predict video view counts. Our study can benefit video caching, video procurement, and advertisement campaign for online video providers.

All large-scale online video systems, for example, Netflix and Youku, make a significant investment on video recommendations that can dramatically affect video information diffusion processes among users. However, there is a lack of efficient methodology to interpret how various recommendation mechanisms affect information diffusion processes resulting in the difficulty to evaluate video recommendation efficiency. In this paper, we propose to quantify and explain video recommendation mechanisms by using epidemic models to mine video view count traces. It is well known that an epidemic model is an efficient approach to model information diffusion processes; while view count traces can be viewed as the results of video information diffusion driven by video recommendations. Thus, we propose a framework based on extended epidemic models to quantify and interpret two recommendation mechanisms, that is, direct and word-of-mouth (WOM) recommendations, by fitting video view count traces collected from Tencent Video, a large-scale online video system in China. Our approach is a novel methodology to evaluate video recommendation mechanisms, and a new perspective to interpret how recommendation mechanisms drive view count evolution.

Understanding streaming user behavior is crucial to the design of large-scale Video-on-Demand (VoD) systems. In this article, we begin with the measurement of individual viewing behavior from two aspects: the temporal characteristics and user interest. We observe that active users spend more hours on each active day, and their daily request time distribution is more scattered than that of the less active users, while the inter-view time distribution differs negligibly between two groups. The common interest in popular videos and the latest uploaded videos is observed in both groups. We then investigate the predictability of video popularity as a collective user behavior through early views. In the light of the limitations of classical approaches, the Autoregressive-Moving-Average (ARMA) model is employed to forecast the popularity dynamics of individual videos at fine-grained time scales, thus achieving much higher prediction accuracy. When applied to video caching, the ARMA-assisted Least Frequently Used (LFU) algorithm can outperform the Least Recently Used (LRU) by 11--16%, the well-tuned LFU by 6--13%, and the LFU is only 2--4% inferior to the offline LFU in terms of hit ratio.

The online video streaming service is of huge market values with billions of worldwide users. For online video providers, e.g., Netflix, Youku, the crucial question is how to understand users' view behaviors and preferences because this knowledge is important for their business operation. Existing solutions mainly rely on analyzing users' historical view records, which however are not always available, especially for new videos and unprovided videos. Different from existing solutions, we propose to infer user behaviors and preferences by jointly analyzing data collected from multiple platforms (e.g., video streaming systems, video databases, etc.). In particular, we use the movie data crawled from a leading video streaming system (i.e., Youku), and a well-known video database in China (i.e., Douban) for this study. Our investigation points out that movie quality (evaluated in terms of Douban scores) and release date jointly influence viewers' preferences. In addition, we reveal a series of user behaviors, e.g., users are reluctant to post comments or ratings for movies they do not like, and user eyeballs are heavily captured by new movies. Understanding of these user behaviors covered by this study is essential for video recommendation and video popularity prediction which can benefit video procurement and advertisement campaign.

Understanding streaming user behavior is crucial to the design of large-scale video-on-demand (VoD) systems. However, existing studies usually treat all the users as an entire entity to analyze the collective user behavior. In this paper, we measure the individual viewing behavior of 10 million sampled users from two perspectives: the temporal characteristics and the user interest, and present our results by dividing users into the active and inactive groups. We observe that the active users spend more hours on each active day, and their daily request time distribution is more scattered than that of the inactive users, while the inter-viewing time distribution differs negligible between two groups. We exhibit the similar viewing behaviors of the active and inactive users, e.g. the common interests in popular videos and the latest uploaded videos. We further propose a modified Weibull distribution to fit users' view completion rate, which can deal with different video categories well. To identify users with similar viewing behaviors, we cluster them into 24 classes using their daily request timestamp or 11 classes using the watched video category. The analysis of cluster centroid manifests the efficacy of the clustering, which enables us to step closer to the understanding of user behavior in large-scale VoD systems.

With the astronomical growth of online video population, the video recommendation system is crucial for users to locate videos fitting their interests. The collaborative filtering (CF) realizing personalized recommendation by analyzing users' historical view records is currently the most prevalent algorithm adopted by existing systems. Nevertheless, video recommendation performance can be improved for most cases, particularly the one with cold-start problem, if additional information is involved. In this article, we propose an AP-based Context-Aware (APCA) recommendation scheme on top of the traditional factor-based CF algorithm by utilizing the information of access points which has not been explored yet by existing works. The underlying principle is that users' view preferences are expected to be different with different contexts, e.g., hotel, home, public areas, which can be inferred by mining the information of access points via which users launch video requests. With the data collected from Tencent Video, a leading online video provider in China, we present a measurement study to show user view preferences in different contexts before our APCA algorithm is introduced. Experiments are executed driven by the trace data collected from Tencent Video to validate the effectiveness of our scheme in improving recommendation performance.

Unraveling quality of experience (QoE) of video streaming is very challenging in bandwidth shared wireless networks. It is unclear how QoE metrics such as starvation probability and buffering time interact with dynamics of streaming traffic load. In this paper, we collect view records from one of the largest streaming providers in China over two weeks and perform an in-depth measurement study on flow arrival and viewing time that shed light on the real traffic pattern. Our most important observation is that the viewing time of streaming users fits a hyper-exponential distribution quite well. This implies that all the views can be categorized into two classes, short and long views with separated time scales. We then map the measured traffic pattern to bandwidth shared cellular networks and propose an analytical framework to compute the closed-form starvation probability on the basis of ordinary differential equations (ODEs). Our framework can be naturally extended to investigate practical issues including the progressive downloading and the finite video duration. Extensive trace-driven simulations validate the accuracy of our models. Our study reveals that the starvation metrics of the short and long views possess different sensitivities to the scheduling priority at base station (BS). Hence, a better QoE tradeoff between the short and long views has a potential to be leveraged by offering them different scheduling weights. The flow differentiation involves tremendous technical and non-technical challenges because video content is owned by content providers but not the network operators and the viewing time of each session is unknown beforehand. To overcome these difficulties, we propose an online Bayesian approach to infer the viewing time of each incoming flow with the “least” information from content providers.

Video popularity (measured by view count) over time is an essential reference for both online video providers and users. According to state-of-the-art works, video popularity is useful for system optimization, load generation, video caching, and video recommendation. Thus, deeper understanding of video popularity evolution is very helpful for improving video service quality and providers' operating efficiency. The core question to be explored in this paper is what key factors govern online video popularity evolution? Through collaboration with our industry partner, Tencent Video, we obtain historical data of video view counts over a period of time, and observe their patterns. We then propose a stochastic fluid model, named as EvoModel, which captures two processes giving rise to different evolution patterns of a given video: (a) the information spreading process and (b) the user reaction process. The driving forces for process (a) can be either via recommendation from the system directly, or word-of-mouth; the extent of the spread is governed by the intrinsic popularity of the video. The factor affecting the second process can be modeled by a user reaction rate. These processes together determine different video popularity evolution patterns. We validate our model by fitting the historical data obtained from a real-world system. Furthermore, we discuss the feasibility of estimating model parameters and predicting popularity.

In the era of mobile Internet, mobile game apps (i.e., applications) enable users to play games on mobile devices anywhere at any time. Such a change has brought a dramatic revolution to the traditional gaming industry. In this paper, we aim at having a comprehensive understanding of the ecosystem of mobile game apps. To this purpose, we conduct a large-scale measurement study over all game apps hosted by four leading app stores in China, which cover both Android and iOS platforms. We collect information of over 75000 mobile game apps in a period of three months (October 2014-January 2015). With obtained datasets, we study the scale, evolution, and overlap of game apps in different app stores from a macroscopic level. We find that none of major app stores can provide a complete set of all game apps. We also investigate download patterns of mobile game apps and the impacts brought by user comments and ratings. We observe clear Pareto effect and power-law effect for game app downloads, and there is no strong positive correlation between app score and the number of its downloads. Last, we characterize the features of popular and unpopular game apps and confirm the negative impacts of embedded ads and paid items. We believe our measurement results can provide useful insights and advice for users, developers, and app store operators.

Content distribution, especially the distribution of video content, unavoidably consumes bandwidth resources heavily. Internet content providers invest heavily in purchasing content distribution network (CDN) services. By deploying tens of thousands of edge servers close to end users, CDN companies are able to distribute content efficiently and effectively, but at considerable cost. Thus, it is of great importance to develop a new system that distributes content at a lower cost but comparable service quality. In lieu of expensive CDN systems, we implement a crowdsourcing-based content distribution system, Thunder Crystal, by renting bandwidth for content upload/download and storage for content cache from agents. This is a large-scale system with tens of thousands of agents, whose resources significantly amplify Thunder Crystal’s content distribution capacity. The involved agents are either from ordinary Internet users or enterprises. Monetary rewards are paid to agents based on their upload traffic so as to motivate them to keep contributing resources. As far as we know, this is a novel system that has not been studied or implemented before. This article introduces the design principles and implementation details before presenting the measurement study. In summary, with the help of agent devices, Thunder Crystal is able to reduce the content distribution cost by one half and amplify the content distribution capacity by 11 to 15 times.

Unraveling quality of experience (QoE) of video streaming is very challenging in bandwidth shared wireless networks. It is unclear how QoE metrics such as buffering time and starvation behavior interact with dynamics of streaming traffic load. In this paper, we collect view records from one of the largest streaming providers in China over two weeks and perform an in-depth measurement study on flow arrival and viewing time that shed light on realistic streaming traffic pattern. Our most important observation is that the viewing time of streaming users fits a hyper-exponential distribution quite well. This implies that all the videos can be categorized into two classes, short and long viewing time with separated time scales. We then map the traffic pattern of large-scale measurement to bandwidth sharing cellular networks. We propose two models to compute the close-form starvation probability and mean sojourn time on the basis of ordinary differential equations (ODEs). Extensive trace-driven simulations validate their accuracy. The proposed models precisely capture how the QoE metrics of video streaming in each class are influenced by the scheduling algorithms at a base station.

Large-scale Internet VoD services often adopt a hybrid overlay approach using both P2P and CDN to stream video content. When the underlay ISP network interconnection is weak, as the situation is found in China, it is important to provision the CDN bandwidth carefully in different ISP networks. Through measurement studies using real world data (from Tencent Video), it is found that users in large ISP networks can leverage more on P2P service, compared to the case for smaller ISP networks, and this is especially true during heavy load periods. We study systematic strategies for allocating CDN bandwidth given fixed total bandwidth purchasing budget, and the influence of P2P factor. We show that a more optimal strategy would bias towards allocating proportionally more CDN bandwidth to smaller ISP networks. This result from our model and analysis is validated by simulation. This problem has not been considered by past works and has practical implications for improving the operation of real-world VoD service providers.

Large Internet video delivery systems serve millions of videos to tens of millions of users on daily basis, via Video-on-Demand (VoD) and live streaming. Video popularity (measured by view count) evolves over time. It represents the workload, as well as business value, of the video to the overall system. The ability to predict video popularity is very helpful for improving service quality and operating efficiency. Previous studies adopted simple (usually static) models for video popularity, or directly adopted patterns from measurement studies. In this paper, we develop a fluid model that tries to capture two hidden processes that give rise to different patterns of a given video's popularity evolution: (a) the information spreading process, and (b) the user reaction process. Specifically, these processes model how the video is recommended to the users, the video's inherent attractiveness, and users' reaction rate; and yield different popularity evolution patterns. We validate our model by fitting the data obtained from a large content provider in China. This model gives us the insight to explain the common and different video popularity evolution patterns and why.

Crowdsourcing-based video service systems, e.g., Thunder Crystal, are novel content distribution platforms composed by a large number of agent devices, acting like miniservers. Most agents are normal Internet users who would like to earn rewarded cash by uploading content through their devices. Compared with CDN, the bandwidth cost is much cheaper; while compared with Peer-to-Peer(P2P), the bandwidth supply is more stable. In this work, we create a stochastic model to analyze the live and VoD streaming performance in such crowdsourcingbased video service systems. Simulation is conducted to validate the accuracy of our analytical results.

Content distribution, especially the distribution of video content, unavoidably consumes bandwidth resource heavily. Internet content providers (ICP) spend lots of money to buy content distribution network (CDN) service. By deploying thousands of edge servers close to end users, CDN companies are able to distribute content efficiently. In lieu of traditional CDN systems, we implement a crowdsourcing-based content distribution system, Thunder Crystal, which utilizes agents' upload bandwidth to amplify the content distribution capacity. Agents are well motivated to contribute storage and upload bandwidth to the system by rebated cash. As far as we know, this is a novel system that has not been studied before. In this work, we will present its design principles first. Then, we study agent behavior and methods to evaluate system efficiency and user efficiency. We evaluate the system by simulations, and observe that agents are well motivated to keep online most of the time and amplify the content distribution capacity by 10~20 times.

There are two types of P2P systems satisfying two different user demands: 1) file downloading and 2) video-on-demand (VoD) streaming. An example of file downloading is the original BitTorrent, and examples for VoD streaming include various commercial P2P-based VoD streaming systems such as that offered by PPLive. We have a hypothesis - by combining a type: 1) system and 2) system as a single P2P system, both the file downloading users and the streaming users of the same video will benefit in performance. The reasoning is that at any moment, only a subset of the file downloading peers can provide good service to VoD streaming peers and the VoD streaming peers are only good at providing service to a different subset of the file downloading peers. The former subset is the set of peers close to completing the downloading of the video file; whereas the latter subset is the set of peers starting to download a video. In this paper, we propose a novel design for a mesh-based video distribution system without depending on video replication on streaming peers. We produce simple back-of-the-envelop analysis to show its effectiveness. Then, we further validate our design and compare it with other designs through simulation and experiments in practical networking environment by implementing a prototype.

Bandwidth cost is a significant concern for online video service providers. Today's video streaming systems mostly use HTTP streaming, with users accessing video segments as HTTP requests. A frequently used strategy is to serve all user requests as fast as possible, as if the user is downloading a file. The downloading rate can often far exceed the playback rate, when the system is below the peak load. This is known as progressive downloading. Since users may quit before viewing the complete video, however, much of the downloaded video can be “wasted.” By studying and exploiting the predictability of users' departure behavior , the authors developed a smart streaming strategy that can significantly improve overall streaming service quality under given server bandwidth. The improvement is achieved by avoiding the waste based on predicted user departure behavior. The proposed smart streaming technique is evaluated by modeling, analysis, and simulation, as well as experimentation using a prototype implementation.

Popular online video-on-demand (VoD) services all maintain a large catalog of videos for their users to access. The knowledge of video popularity is very important for system operation , such as video caching on content distribution network (CDN) servers. The video popularity distribution at a given time is quite well understood. We study how the video popularity changes with time, for different types of videos, and apply the results to design video caching strategies. Our study is based on analyzing the video access levels over time, based on data provided by a large video service provider. Our main finding is, while there are variations, the glory days of a video’s popularity typically pass by quickly and the probability of replaying a video by the same user is low. The reason appears to be due to fairly regular number of users and view time per day for each user, and continuous arrival of new videos. All these facts will affect how video popularity changes, hence also affect the optimal video caching strategy. Based on the observation from our measurement study, we propose a mixed replication strategy (of LFU and FIFO) that can handle different kinds of videos. Offline strategy assuming tomorrow’s video popularity is known in advance is used as a performance benchmark. Through trace-driven simulation, we show that the caching performance achieved by the mixed strategy is very close to the performance achieved by the offline strategy.

Online video-on-demand(VoD) services invariably maintain a view count for each video they serve, and it has become an important currency for various stakeholders, from viewers, to content owners, advertizers, and the online service providers themselves. There is often significant financial incentive to use a robot (or a botnet) to artificially create fake views. How can we detect fake views? Can we detect them (and stop them) efficiently? What is the extent of fake views with current VoD service providers? These are the questions we study in this article. We develop some algorithms and show that they are quite effective for this problem.

We consider a peer-to-peer (P2P)-assisted video-on-demand (VoD) system where each peer can store a relatively small number of movies to offload the server when these movies are requested. User requests are stochastic based on some movie popularity distribution. The problem is how to replicate (or place) content at peer storage to minimize the server load. Several variations of this replication problem have been studied recently with somewhat different conclusions. In this paper, we first point out and explain that the main difference between these studies is in how they model the scheduling of peers to serve user requests, and show that these different scheduling assumptions will lead to different “optimal” replication strategies. We then propose a unifying request scheduling model, parameterized by the maximum number of peers that can be used to serve a single request. This scheduling is called Fair Sharing with Bounded Degree (FSBD). Based on this unifying model, we can compare the different replication strategies for different degree bounds and see how and why different replication strategies are favored depending on the degree. We also propose a simple (primarily) distributed replication algorithm and show that this algorithm is able to adapt itself to work well for different degrees in scheduling.

Popular online Video-on-Demand (VoD) services all maintain a large catalog of videos for their users to access. The number of servers assigned to serve each video is directly related to the relative popularity of the video. The distribution of popularity at a given time is quite well understood. We study how the video popularity changes over its lifetime, for different types of videos. Our study is based on analyzing the video access levels over time, based on data provided by a large video service provider. Our main finding is, while there are variations, the glory days of a video typically pass by quickly and probability of replaying a video by the same user is low. The reason appears to be due to fairly regular number of users and view time per day for each user, and continuous arrival of new videos. We then discuss the implication of our findings for video replication and recommendation.

Online video-on-demand (VoD) services invariably maintain a view count for each video they serve, and it has become an important currency for various stakeholders, from viewers, to content owners, advertizers, and the online service providers themselves. There is often significant financial incentive to use a robot (or a botnet) to artificially create fake views. How can we detect the fake views? Can we detect them (and stop them) efficiently? What is the extent of fake views with current VoD service providers? These are the questions we study in this paper. We develop some algorithms and show their effectiveness for this problem.

Currently, video-on-demand (VoD) streaming over Internet is a popular application. Because of the rapidly growing video population and user population, how to maintain high user quality of experience (QoE) with low cost is a challenging problem for Internet video streaming providers. A promising technique to potentially benefit VoD streaming system is network coding. A number of recent works studied how to use network coding to simplify chunk scheduling strategy to enhance VoD streaming performance. Most of these works only covered extreme cases of pure coding or pure chunk scheduling, emphasizing implementation, and experimentation in peer-to-peer (P2P) scenario without analytically evaluating the realizable performance gains explicitly. In this paper we discuss the strength and weakness of a family of coding strategies for CDN-based VoD streaming systems. The coding schemes are characterized by block sizes while the chunk scheduling strategy is characterized by the order to download chunks. Both pure coding strategy and pure chunk scheduling strategy are special cases of this family of strategies. We then propose a model to evaluate the benefits brought by each strategy. Basically, the coding scheme with larger block size gives more streaming and scheduling benefits with the cost of heavier overheads (e.g., encoding and decoding). System designers can take advantage of our model to balance the tradeoff between coding gain and coding overheads.

Video streaming service in wireless networks is increasingly using dynamic selection of video bit-rates to provide a high quality of user experience (QoE). The bit-rate switching mechanism, performed at client side, plays a key role in determining QoE metrics. In this paper, we present the first analytical framework to compute starvation probability of playout buffer, continuous playback time and mean video quality, given the bit-rate switching logics. Wireless channel is modeled as a continuous time Markov process, and playout buffer is modeled as a fluid queue with Markov modulated fluid arrival. We construct a set of ordinary differential equations (ODEs) to characterize the dynamics of starvation probability and expected continuous playback time with regard to buffer length, and simple models to analyze mean bit-rate for different bit-rate switching algorithms. Our framework is very general in that by adding appropriate parameters, it can be utilized to predict the QoE metrics of dynamic adaptive streaming with a variety of features: i) buffer-aware bit-rate switching ii) (im)patience of the user, and iii) receiver-side flow control.

A big portion of Internet traffic nowadays is video. A good understanding of user behavior in online VoD systems can help us design, configure and manage video content distribution. With the help of a major video on demand (VoD) service provider, we conduct a detailed study of user behavior watching streamed videos over the Internet. We engineered the video player at the client side to collect user behavior reports for over 540 million sessions.In order to isolate the possible effect of session quality of experience (QoE) on user behavior, we focus on the sessions with perfect QoE, and leave out those sessions with QoE impairments (such as freezes).Our main finding is that users spend a lot of time browsing: viewing part of one video after another, and only occasionally (around 20% of the time) watching a video to its completion. We consider seek (jump to a new position of the video) as a special form of browsing - repeating partial viewing of the same video. Our analysis leads towards a user behavior model in which a user transitions through a random number of short views before a longer view, and repeats the process a random number of times. A purely abstract version of such user behavior model was proposed by Wu et al [1] as a closed queueing network formulation. Our study uncovers the parameters and distributions of such a stochastic behavior model based on observations in practice.

Video content downloading using the P2P approach is scalable, but does not always give good performance. Recently, subscription-based premium services have emerged, referred to as cloud downloading. In this service, the cloud storage and server caches user-interested content and updates the cache based on user downloading requests. If a requested video is not in the cache, the request is held in a waiting state until the cache is updated. We call this design server mode. An alternative design is to let the cloud server serve all downloading requests as soon as they arrive, behaving as a helper peer. We call this design helper mode. Our model and analysis show that both these designs are useful for certain operating regimes. The helper mode is good at handling a high request rate, while the server mode is good at scaling with video population size. We design an adaptive algorithm (AMS) to select the service mode automatically. Intuitively, AMS switches service mode from server mode to helper mode when too many peers request blocked movies, and vice versa. The ability of AMS to achieve good performance in different operating regimes is validated by simulation .

Traditional video-on-demand (VoD) systems rely purely on servers to stream video content to clients, which does not scale. In recent years, peer-to-peer assisted VoD (P2P VoD) has proven to be practical and effective. In P2P VoD, each peer contributes some storage to store videos (or segments of videos) to help the video server. Assuming peers have sufficient bandwidth for the given video playback rate, a fundamental question is what is the relationship between the storage capacity (at each peer), the number of videos, the number of peers, and the resultant off-loading of video server bandwidth. In this paper, we use a simple statistical model to derive this relationship. We propose and analyze a generic replication algorithm Random with Load Balancing (RLB) that balances the service to all movies for both deterministic and random (but stationary) demand models and both homogeneous and heterogeneous peers (in upload bandwidth). We use simulation to validate our results for sensitivity analysis and for comparisons to other popular replication algorithms. This study leads to several fundamental insights for P2P VoD system design in practice.

We consider a P2P-assisted Video-on-Demand (VoD) system where each peer can store a relatively small number of movies to offload the server when these movies are requested. User requests are stochastic based on some movie popularity distribution. The problem is how to replicate (or place) content at peer storage to minimize the server load. Several variation of this replication problem have been studied recently with somewhat different conclusions. In this paper, we first point out that the main difference between these studies is in how they model the scheduling of peers to serve user requests, and show that these different scheduling assumptions will lead to different “optimal” replication strategies. We then propose a unifying request scheduling model, parameterized by the maximum number of peers that can be used to serve a single request. This scheduling is called Fair Sharing with Bounded Out-Degree (FSBD). Based on this unifying model, we can compare the different replication strategies for different out-degree bounds and see how and why different replication strategies are favored depending on the out-degree. We also propose a new simple, adaptive, and essentially distributed replication algorithm, and show that this algorithm is able to adapt itself to work well for different out-degree in scheduling.

We consider a P2P-assisted content storage and delivery system to support a streaming Video-on-Demand (VoD) service. In this system, the peers are part of the service provider (e.g. set-top boxes) with limited storage space. Servers with ample storage and bandwidth are deployed to guarantee the availability and quality, but it is desirable to minimize the server utilization to reduce costs. Based on experience of implementing a deployed P2P VoD system, it was suggested in [1] that a movie's availability should be proportional to the movie's popularity. Based on further refinement, it is observed [2] that performance can be further improved by more (than proportional) availability for cold movies in P2P system. In this paper, we show that as the number of movies becomes large and there is some skewness in movie popularity, then one cannot expect the P2P part of the system to reduce server load as well as provide availability to all movies at the same time. It is a trade-off between coverage of movies and streaming throughput provided by the P2P system. If the goal is to minimize server load, under some reasonable conditions, we show that it is best to store and replicate only the hottest K* movies in the P2P part of the system. We also study the relationship between the skewness of the movie popularity distribution, P2P resources and the value of K*. Finally, we use simulation to validate our results.

Traditional Video-on-Demand (VoD) systems reply purely on servers to stream video content to clients, which does not scale. In recent years, Peer-to-peer assisted VoD (P2P VoD) has proven to be practical and effective. In P2P VoD, each peer contributes some storage to store videos (or segments of videos) to help the video server. Assuming peers have sufficient bandwidth for the given video playback rate, a fundamental question is what is the relationship between the storage capacity (at each peer), the number of videos, the number of peers and the resultant off-loading of video server bandwidth. In this paper, we use a simple statistical model to derive this relationship. We propose and analyze a generic replication algorithm RLB which balances the service to all movies, for both deterministic and random demand models, and both homogeneous and heterogeneous peers (in upload bandwidth). We use simulation to validate our results, for sensitivity analysis and for comparisons with other popular replication algorithms. This study leads to several fundamental insights for design P2P VoD systems in practice.

Peer-to-peer (P2P) streaming tries to achieve scalability (like P2P file distribution) and at the same time meet real-time playback requirements. It is a challenging problem still not well understood. In this paper, we describe a simple stochastic model that can be used to compare different downloading strategies to random peer selection. Based on this model, we study the tradeoffs between supported peer population, buffer size, and playback continuity. We first study two simple strategies: Rarest First (RF) and Greedy. The former is a well-known strategy for P2P file sharing that gives good scalability by trying to propagate the chunks of a file to as many peers as quickly as possible. The latter is an intuitively reasonable strategy to get urgent chunks first to maximize playback continuity from a peer's local perspective. Yet in reality, both scalability and urgency should be taken care of. With this insight, we propose a Mixed strategy that achieves the best of both worlds. Furthermore, the Mixed strategy comes with an adaptive algorithm that can adapt its buffer setting to dynamic peer population. We validate our analytical model with simulation. Finally, we also discuss the modeling assumptions and the model's sensitivity to different parameters and show that our model is robust.

P2P streaming tries to achieve scalability (like P2P file distribution) and at the same time meet real-time playback requirements. It is a challenging problem still not well understood. In this paper, we describe a simple stochastic model that can be used to compare different data-driven downloading strategies based on two performance metrics: continuity (probability of continuous playback), and startup latency (expected time to start playback). We first study two simple strategies: rarest first and greedy. The former is a well-known strategy for P2P file sharing that gives good scalability, whereas the latter an intuitively reasonable strategy to optimize continuity and startup latency from a single peer's viewpoint. Greedy, while achieving low startup latency, fares poorly in continuity by failing to maximize P2P sharing; whereas rarest first is the opposite. This highlights the trade-off between startup latency and continuity, and how system scalability improves continuity. Based on this insight, we propose a mixed strategy that can be used to achieve the best of both worlds. Our algorithm dynamically adapts to the peer population size to ensure scalability; at the same time, it reserves part of a peer's effort to the immediate playback requirements to ensure low startup latency.