Conventional fairness in multi-tenant Large Language Model (LLM) inference services is typically defined by system-centric metrics such as equitable resource allocation. We argue that this is unilateral and it creates a gap between measured system performance and actual user-perceived quality. We challenge this notion by introducing and formalizing Experiential Fairness, a user-centric paradigm that shifts the objective from equality of opportunity (resource access) to equity of outcome (user experience). With this motivation we propose ExFairS, a lightweight scheduling framework that perceives each user's satisfaction as a composite measure of Service Level Objective (SLO) compliance and resource consumption, and dynamically re-orders the serving queue guided by a credit-based priority mechanism. Extensive experiments on an 8-GPU NVIDIA V100 node show that ExFairS reduces the SLO violation rate by up to 100% and improves system throughput by 14-21.9%, outperforming state-of-the-art schedulers and delivering a demonstrably higher degree of Experiential Fairness.

With the proliferation of Internet of Things (IoT) applications, advanced time series forecasting techniques have become increasingly critical for managing and responding to complex temporal dynamics. However, traditional RNN-based methods have faced challenges in the Long-term Time Series Forecasting (LTSF) domain when dealing with excessively long look-back windows and forecast horizons. Consequently, the dominance in this domain has shifted towards Transformer, MLP, and CNN approaches. The substantial number of recurrent iterations are the fundamental reasons behind the limitations of RNNs in LTSF. To address these issues, we propose two novel strategies to reduce the number of iterations in RNNs for LTSF tasks: Segment-wise Iterations and Parallel Multi-step Forecasting (PMF). RNNs that combine these strategies, called SegRNN, significantly reduce the required recurrent iterations for LTSF, resulting in notable improvements in forecast accuracy and inference speed. Extensive experiments demonstrate that SegRNN not only outperforms state-of-the-art Transformer-based models but also reduces runtime and memory usage by more than 78%, making it highly suitable for resource-constrained IoT scenarios. These achievements provide strong evidence that RNNs continue to excel in LTSF tasks and encourage further exploration of this domain with more RNN-based approaches. The code is available at: https://github.com/lss-1138/SegRNN.

This paper introduces SparseTSF, a novel and extremely lightweight method for Long-term Time Series Forecasting (LTSF), designed to address the challenges of modeling complex temporal dependencies over extended horizons with minimal computational resources. At the heart of SparseTSF lies the Cross-Period Sparse Forecasting technique, which simplifies the forecasting task by downsampling the original sequences to focus on cross-period trend prediction. This technique not only significantly reduces model complexity and the number of parameters but also serves as an implicit regularization mechanism that enhances the model's robustness, achieving an optimal balance between performance and efficiency. Based on this technique, SparseTSF uses fewer than 1,000 parameters to achieve competitive performance compared to state-of-the-art methods, with evident advantages under longer look-back windows (e.g., 720) that allow the model to better exploit inherent periodicity and trend information. Furthermore, SparseTSF showcases remarkable generalization capabilities, making it well-suited for scenarios with limited computational resources, small samples, or low-quality data.

The rapid expansion of cloud computing has made precise performance evaluation a critical necessity. However, conventional cloud benchmarks often face significant limitations in simulation environments—necessary for scalable and cost-effective testing—due to the complexity of technology stacks and substantial runtime overheads. Proxy benchmarking has thus emerged as a practical alternative. Existing methods primarily focus on the global similarity of micro-architectural metrics between proxy benchmarks and real workloads but neglect their temporal similarity, leading to inaccurate performance evaluations, flawed cache behavior simulations, and misguided architectural optimization decisions. To address this, we present PaTGen , a phase-aware method for generating proxy benchmarks that accurately reflect both global and temporal similarity. By partitioning workloads into phases and formulating proxy generation as nonlinear optimization problems, PaTGen further refines intra-phase execution patterns via the delay-based temporal similarity optimization (DTSO) technique. Evaluations on 15 real-world workloads show PaTGen achieves over 97% global similarity in key metrics while significantly outperforming state-of-the-art methods in temporal similarity. Ablation studies confirm the efficacy of phase division and DTSO. Further experiments confirm its scalability and generalizability across architectures. Moreover, the effectiveness observed in downstream tasks provides empirical evidence that preserving temporal similarity is a fundamental requirement for proxy benchmarks to faithfully capture real workload behavior.

Data centers often deploy servers that exceed the capacity of the power infrastructure to increase power utilization, i.e., power over-subscription. To address potential power overloads, power capping mechanisms are implemented for protection. However, the power efficiency disparities among heterogeneous servers, along with the adjustment intervals required for power capping at the cluster level, pose challenges to capping decisions, especially for latency-sensitive applications with higher quality of service requirements. To tackle this challenge, we propose a prediction-based, power efficiency-aware dynamic power capping framework (PPE-DPC), comprising two stages. First, we design a lightweight online prediction to capture requests generated within the power adjustment time intervals. Then, leveraging prior knowledge of the power efficiency of heterogeneous servers, we design a greedy strategy to fine-tune the capping power for better capping decisions. Extensive simulations and conducted in Testbed using Alibaba traces demonstrate that PPE-DPC outperforms existing solutions, optimizing request latency, power utilization, and mitigating the negative impact of power adjustment intervals. Finally, we also explore the effect of prediction error on PPE-DPC and find that only 3x the true prediction error is weaker than the existing optimal capping algorithm.

Cloud-edge-device (CED) cooperative computing is an emerging paradigm that extends the reach of cloud services, providing higher flexibility and scalability to modern AI-driven computing services. However, traditional “one-size-fits-all” AI model construction at the edge struggles to accommodate the strong heterogeneity of target devices. This leads to an increasing demand for specialized models tailored for local resources in AI applications. To this end, we introduce ComHA, a cooperative model building framework based on hierarchical Automated Machine Learning (AutoML) to bridge the gap between hyperparameter optimization on the cloud and local model customization at the edge. In this framework, the cloud performs high-level AutoML to reduce the search space of learning algorithms, model architectures, and relevant hyperparameters to a specific set based on target device specifications. Subsequently, edge devices execute low-level AutoML to identify and train the optimal model, customized for their local data and resources. This approach aims to strike a balance between the benefit and cost of customized model building. Through extensive experiments conducted on a real-world testbed with public datasets, our results demonstrate that ComHA outperforms traditional methods in producing tailored models of high accuracy and low inference latency in various environments.

Accurately forecasting power consumption in data center servers requires addressing the temporal distribution shift caused by dynamic resource demands. However, existing methods rely on global normalization, which cannot capture short-term localized shift, leading to unsatisfactory performance when forecasting non-stationary time series. To address this challenge, we propose a novel bi-level optimization framework for forecasting non-stationary long-term power consumption, namedNoSPF. The framework employs hierarchical optimization to separately model the stationary and local non-stationary components of power consumption time series, offering a flexible, model-agnostic paradigm for time-series forecasting. Using Discrete Wavelet Transform (DWT) for multi-scale time–frequency analysis,NoSPFdecomposes the series into non-stationary components driven by short-term fluctuations and stationary components that capture long-term trends. Furthermore,NoSPFintegrates a lightweight Multi-Layer Perceptron (MLP) to predict the local non-stationary components, enhancing the framework’s forecasting accuracy by providing more precise approximations of the future power distribution. Extensive experiments on real-world server datasets demonstrate the superior performance and effectiveness ofNoSPF.

In resource-constrained edge computing, the execution efficiency of workflow applications is significantly affected by bandwidth contention, especially during data transmissions between dependent tasks. However, existing workflow scheduling studies often struggle to optimize transmission delay effectively, whereas bandwidth slicing offers promising potential by leveraging the dynamic nature of bandwidth resources. To address this issue, we propose Workflow Scheduling with Dynamic Bandwidth Slicing (WSDBS), a novel scheduling algorithm that integrates bandwidth slicing into the workflow execution process. By introducing a dual-prediction strategy, WSDBS estimates the availability of both computational and bandwidth resources on servers, facilitating efficient task scheduling decisions under transmission uncertainty. Moreover, a novel transmission urgency metric is developed, which is derived from both link load and transmission criticality. This metric guides bandwidth slicing for the dynamic allocation of server-side bandwidth resources, ultimately alleviating contention among concurrent transmissions. Extensive experiments based on real-world Alibaba cluster traces show that WSDBS consistently improves scheduling efficiency, reducing the average makespan by 10.87%-14.44% over state-of-the-art baselines. These results validate its effectiveness in alleviating bandwidth contention and improving scheduling performance in edge computing environments.

In End-Edge-Cloud (EEC) computing environments, the diversity of devices often requires cloud-trained models to be adapted for end/edge devices, complicating decentralized project management. To address this, end/edge devices are increasingly using local model sharing instead of traditional cloud solutions. Popular platforms like GitHub and DockerHub lack the necessary data authenticity and security for high-stakes applications. While blockchain can ensure secure data sharing, permissioned blockchains struggle with the dynamic nature of EEC devices. To solve this, we propose BlockEdge, a hybrid blockchain architecture combining a permissioned blockchain with Practical Byzantine Fault Tolerance (PBFT) for cloud-based data management and a permissionless blockchain with Proof of Work (PoW) for decentralized model sharing at the end/edge. We enhance the PoW process with a dynamic mining algorithm and a lazy-loading Merkle tree structure, improving energy efficiency and computational performance. Experimental results show that BlockEdge reduces energy consumption by over 50% and cuts data update time by 91.73%, effectively addressing the energy and time inefficiencies of mainstream consensus mechanisms.

Serverless computing, or Function-as-a-Service, continues to gain popularity due to its pay-as-you-go billing model, flexibility, and cost efficiency. However, these same features introduce significant security risks, such as the Denial-of-Wallet (DoW) attack. In this paper, we conduct real-world DoW attacks on commercial serverless platforms to evaluate their severity. To detect such attacks, we design, implement, and evaluate Clover, an accurate and user-friendly DoW detection system with negligible performance overhead. Clover addresses information ambiguity in serverless environments by deploying a request-oriented metric collection agent. At its core, Clover proposes a workload verification approach to bridge performance metrics and execution duration. Specifically, Clover uses a multivariate linear model to learn the benign relationship between metrics and execution duration, effectively characterizing normal workload behavior. It then continuously monitors runtime workloads by calculating their Mahalanobis distance from this learned benign model. Deviations identified through this distance indicate potential DoW attacks. Implemented as a practical system, Clover introduces performance overhead of less than 3.2%, maintains an average model execution time of only 0.84 microseconds, and achieves an accuracy of 92.7% under the most challenging scenario.

In edge computing, workflow applications are optimally scheduled onto edge servers that are pre-equipped with the necessary services to satisfy stringent low-latency demands. However, prior research has not fully addressed the joint optimization of service placement and workflow scheduling, particularly the exploitation of task parallelism to reduce overall makespan. To address this shortcoming, we explore the combined workflow scheduling and service placement (WSP-SP) problem with the goal of minimizing the average makespan of applications. Recognizing that WSP-SP is NP-hard, we propose a two-stage, Parallelism Aware Workflow Scheduling and Service Placement strategy (PAWSSP) that minimizes the makespan with low complexity. In the first stage, a Parallelism Aware Service Placement module (PASP) is designed to adjust the service layout by allocating services with high parallelism onto distinct servers to fully leverage task-level concurrency. In the subsequent workflow scheduling stage, PAWSSP determines task priority by resolving inter-task competition and further reduces waiting times by assigning tasks to servers experiencing lower resource contention. We further extend PAWSSP to make it applicable to both offline and online scenarios. Extensive evaluations demonstrate that PAWSSP performs robustly across diverse scenarios, reducing the average makespan by 3.9%-14.7% compared to existing baselines, while maintaining modest runtime overhead.

As a privacy-preserving solution, federated learning (FL) demonstrates great potential in distributed model training, but limited bandwidth, particularly in near-field communication (NFC)-based systems, emerges as a key bottleneck by restricting the number of participating clients. To address this challenge, over-the-air FL leverages the superposition property of wireless multiple-access channels, enabling faster model training and accommodating more clients, even in bandwidth-constrained scenarios like NFC. However, due to its analog-integrated nature, the FL performance is also affected by other factors, such as channel noise. These motivate us to consider how the selected client set and channel noise affect FL performance. To explore this concern, in this article, we consider an over-the-air FL system with analog gradient aggregation and analyze the impact of the selected client set and channel noise on FL training performance. The theoretical analysis effectively shows the importance of the clients’ number and the power scaling factor to the FL training performance. Based on the theoretical analysis, we transform the global optimization problem into the client selection problem and propose a dynamic client selection scheme to optimize the training performance under the aggregation error constraint. Experimental results demonstrate that our proposed scheme can boost FL by speeding up the convergence of the global model (at least 35%) and saving energy consumption.

ABSTRACT Background In the data era, big data systems have emerged as pivotal tools, underscoring the importance of performance prediction in enhancing the efficiency of big data clusters. Numerous performance models have been proposed, often grounded in artificial intelligence or simulation methodologies. While the bulk of research focuses on refining prediction precision and minimizing overhead, limited attention has been given to the consignation and standardization of these models. Objectives To bridge this gap between model developers and end‐users, this paper introduces AMORA—a novel versatile framework tailored for predicting the performance of big data systems. Methods Leveraging the identified behavior descriptions‐computation submodels (BD‐CS) pattern that is prevalent among various big data job performance models, AMORA allows access to different plugins accommodating different performance models' implementations. This framework also integrates a novel mutable computation graph technique to facilitate backtracking computation. Furthermore, AMORA's functionality extends to comprehensive end‐to‐end usability by enabling the acceptance of origin configuration files from diverse big data systems and presenting easily interpretable prediction reports. Results This work demonstrates AMORA's efficacy in producing an accurate trace of Hadoop job through the selection of appropriate performance model plugins and parameter adjustments and showcasing the application of the proposed mutable computation graph technique in calculating the starting moment of an early‐start reducer. Additionally, two validation experiments are conducted, involving the implementation of various Hadoop and Spark performance models, respectively. The experiment results manifest the prediction precision and overheads of these performance models. Conclusion These experiments exhibit AMORA's role as a benchmark platform for implementing various types of big data job performance models catered to diverse big data systems.

With the development of edge computing, DNN services have been widely deployed on edge devices. The deployment efficiency of deep learning models relies on the optimization of inference and scheduling policy. However, traditional optimization methods on edge devices still suffer from prohibitively long tuning time due to devices’ low computational power. Meanwhile, the widely used scheduling algorithm, the dominant resource fairness algorithm(DRF algorithm), struggles to maximize the efficiency of model execution on edge devices and inevitably increases average waiting time as it is not applicable in the real-time distributed computing environment. In this paper, we propose Cacomp, a distributed cloud-assisted deep learning compiler framework that features accelerating the optimization on edge devices with assistance from the cloud and a novel inference task scheduling algorithm. Our framework utilizes the tuning records from the cloud devices and proposes a two-step distillation strategy to obtain the best tuning record set for the edge device. For the scheduling process, we propose an RD-DRF algorithm to allocate inference tasks to edge devices based on dominant resource matching in real time. Extensive results show that our framework can achieve up to 2.19× improvement in the optimization time compared with other methods on edge devices. Our proposed scheduling algorithm significantly shortens the average waiting time of inference tasks by 30% and improves resource utilization by 20% on edge devices.

Nowadays, machine learning (ML)-based power prediction models for servers have shown remarkable performance, leveraging large volumes of labeled data for training. However, collecting extensive labeled power data from servers in cloud data centers incurs substantial costs. Additionally, varying resource demands across different workloads (e.g., CPU-intensive, memory-intensive, and I/O-intensive) lead to significant differences in power consumption behaviors, known as domain shift. Consequently, power data collected from one type of workload cannot effectively train power prediction models for other workloads, limiting the exploration of the collected power data. To tackle these challenges, we proposeTGCP, a cross-workload power prediction method based on multi-source transfer Gaussian process regression.TGCPtransfers knowledge from abundant power data across multiple source workloads to a target workload with limited power data. Furthermore, Continuous normalizing flows adjust the posterior prediction distribution of Gaussian process, making it locally non-Gaussian, enhancingTGCP's ability to handle real-world power data distribution. This method enhances prediction accuracy for the target workload while reducing the expense of acquiring power data for real cloud data centers. Experimental results on a realistic power consumption dataset demonstrate thatTGCPsurpasses four traditional ML methods and three transfer learning methods in cross-workload power prediction.

The fuzzy broad learning system (FBLS) has received increasing attention due to its ability to quickly train from broad learning systems (BLS) and interpretability with fuzzy inference. However, the randomness of BLS brings instability to the training performance of the model, so the hyperparameters of the model are crucial for its performance. Currently, many FBLS use grid search to determine hyperparameters. However, grid search brings longer search time and the parameters obtained have randomness, which may not necessarily be the optimal hyperparameters. In response to these challenges, this paper proposes a fuzzy broad learning system with automatic determination of hyperparameters (ADHFBLS). We construct a novel FBLS based on the interval type-2 fuzzy set and design an incremental learning algorithm for rules and enhancement nodes to support rapid model expansion. Meanwhile, a heuristic hyperparameter automatic optimization algorithm is designed to overcome the randomness and long optimization time of grid search. Experiments have shown that ADHFBLS has higher accuracy and shorter model tuning time compared to some state-of-the-art models based on FBLS.

Harmful fine-tuning attack poses a serious threat to the online fine-tuning service. Vaccine, a recent alignment-stage defense, applies uniform perturbation to all layers of embedding to make the model robust to the simulated embedding drift. However, applying layer-wise uniform perturbation may lead to excess perturbations for some particular non-safety-critical layers, resulting in defense performance degradation and unnecessary memory consumption. To address this limitation, we propose a Targeted Vaccine (T-Vaccine), a memory-efficient safety alignment method that applies perturbation to only selected layers of the model. T-Vaccine follows two core steps: First, it uses the harmful gradient norm as a statistical metric to identify the safety-critical layers. Second, instead of applying uniform perturbation across all layers, T-Vaccine only applies perturbation to the safety-critical layers while keeping other layers frozen during training. Results show that T-Vaccine outperforms Vaccine in terms of both defense effectiveness and resource efficiency. Comparison with other defense baselines, e.g., RepNoise and TAR also demonstrate the superiority of T-Vaccine. Notably, T-Vaccine is the first defense that enables a fine-tuning-based alignment method for 7B pre-trained models trained on consumer GPUs with limited memory (e.g., RTX 4090).

Federated Learning (FL) mitigates data leakage by sharing only local machine learning models instead of raw data. However, it remains vulnerable to differential attacks. Differential Privacy (DP) addresses this concern by introducing noise to make it challenging for adversaries to reconstruct training samples. Nonetheless, clients often have varying attitudes toward data privacy, quantified by their privacy budgets. Low privacy budgets indicate the stringent privacy requirements of clients, requiring high compensations to incentivize their participation. Focusing solely on privacy budgets, however, can introduce selection bias, potentially compromising model generalization. Therefore, it is essential to emphasizes the fairness of client participation, ensuring that clients with lower privacy budgets also have opportunities to contribute to the training process. To tackle the above challenges, this paper formulates a novel DP-based incentive problem in FL, aiming to optimize the utilities of both the server and the clients. Specifically, we propose an auction mechanism that jointly selects participants based on their heterogeneous privacy budgets and determines appropriate payments. The proposed auction mechanism is proven to achieve several desirable properties, including computational efficiency, individual rationality, budget balance, truthfulness, and guaranteed optimization performance. Finally, simulation results validate the effectiveness of the proposed mechanism.

In Mobile Edge Computing (MEC) scenarios, computational tasks are popularly deployed using containerization to isolate the runtime environment. To complete the execution of the task, the edge server first pulls the image, then instantiates and runs the container. Since it takes a lot of time for the edge server to download the image from the cloud, image reuse reduces the pulling latency significantly. However, the limited storage capacity of edge servers hinders image reuse. Recent works have enhanced reuse efficiency by leveraging the hierarchical structure of images and caching high-value layers. However, their efficiency remains limited due to the lack of multi-container collaboration. This paper proposes a novel container scheduling strategy based on image layer reuse and sequence arrangement (ILR-SA) for MEC scenarios, which achieves efficient scheduling by collaborating multiple containers. First, containers are greedily deployed into the edge cluster. Then, the execution sequence of containers is modeled as an optimal Hamiltonian path problem, efficiently solved by our proposed decomposition algorithm. Finally, an efficient image layer update strategy is used to achieve layer reuse. We conduct rigorous experiments to demonstrate that our proposed container scheduling strategy reduces the computational task completion time by up to 91.3% compared to existing approaches.

Federated learning (FL) is a distributed machine learning paradigm that enables heterogeneous devices to train a model collaboratively. Recognizing communication as a bottleneck in FL, existing communication-efficient solutions, e.g., HeteroFL and LotteryFL, etc., utilize gradient sparsification to reduce communication costs. However, existing solutions fail to address the dynamic bandwidth issue in which the bandwidth of each client is constantly changing throughout the training process. In this article, we propose AdaptiveFL, a communication-adaptive FL framework, considering the dynamic constraints of bandwidth. The design of AdaptiveFL follows two key steps: 1) in each round, each device selects a best-fit sub-model for communication per currently available bandwidth; and 2) to guarantee the performance of each sub-model sent under dynamic bandwidth constraints, AdaptiveFL employs a local training method that enables each device to train a "tailorable" local model, which can be tailored to any sparsity with competitive accuracy. We compare AdaptiveFL with several communication-efficient SOTA methods and demonstrate that AdaptiveFL outperforms other baselines by a large margin.

Task scheduling in cloud computing is a challenging but crucial task for ensuring service quality and load balance. Mainstream scheduling algorithms, such as heuristic algorithms and reinforcement learning, have made progress in this area. However, online task scheduling algorithms, such as reinforcement learning, can pose computational challenges in scenarios with limited computational power and heterogeneous resources. Heuristic algorithms, which are more suitable for offline scheduling where the types and quantities of tasks are known in advance, also require substantial computational resources for online scheduling. In this work, we propose the end-edge-cloud (EEC) heterogeneous resources scheduling method (EHRSM) based on a recurrent neural network (RNN) model and particle swarm optimization (PSO). EHRSM uses an RNN model trained on a dataset generated by dynamic programming to recognize and cache online tasks, efficiently transforming online task scheduling into offline scheduling. Additionally, a PSO algorithm with Cantor expansion (CE) for coding optimization is used to complete the offline scheduling. Experimental results show that the method is effective in converting online scheduling to offline scheduling, reducing the average task completion time and waiting time. Compared with existing online scheduling methods, EHRSM reduces task completion time by up to 48.24%.

As deep learning (DL) workloads scale in complexity and volume, ensuring predictable job queuing times has become a critical challenge for data centers. Existing scheduling solutions primarily focus on minimizing tardiness or job completion times (JCT), often neglecting the need for deterministic queuing, particularly in dynamic and preemptive environments. This paper introducesKairos, a preemption-based scheduling framework enhanced by user collaboration to address these gaps.Kairoscombines adivide-and-conquerstrategy—segmenting jobs into sequential units with adaptive priorities—and a user-collaborative mechanism for better duration estimation. By leveraging real-time feedback from resource contention and queuing delays,Kairosminimizes a novel metric, theQueue inStability Index(QSI), achieving significant improvements in queuing predictability while maintaining competitive JCT. Experimental results demonstrate thatKairosreduces QSI by over 99.8% compared to state-of-the-art deadline-aware baselines, offering robust performance for diverse DL workloads.

Since the rapid development of deep learning (DL) technology, large-scale GPU clusters receive a large number of DL workloads daily. To speed up the completion time, the workloads usually occupy several GPUs on a server. However, workload scheduling inevitably generates resource fragmentation, which results in many scattered GPU resources being unavailable. Existing works address improving resource utilization by reducing GPU resource fragmentation, while they focus on resource scheduling for a single cluster and ignore multiple clusters. Multi-cluster scenarios, such as virtual clusters and geo-distributed clusters, require load balancing to avoid some clusters exhausting resources while some clusters are idle while improving resource utilization, which is not well addressed by existing works. In this paper, we propose MCG-Sched, a scheduling strategy to reduce resource fragmentation in multiple GPU clusters while maintaining load balancing among clusters. MCG-Sched measures the fragmented resources with the distribution of workload demands and uses a scheme that minimizes fragmentation in workload scheduling. Meanwhile, MCG-Sched achieves balanced load scheduling across clusters through the load balancing index. MCG-Sched senses the workload requests in the waiting queue, and prioritizes the workloads by combining fragmentation measurement and load balancing index to maximize resource utilization and load balancing during load peak. Our experiments show that MCG-Sched reduces unallocated GPUs up to 1.45× and workload waiting time by more than 40% compared to existing fragmentation-aware methods and achieves effective load balancing.

Power consumption prediction is one of the keys to optimize the energy consumption of servers. Existing traditional regression-based methods are too simple and poorly generalized, while popular deep learning methods require too much data. Therefore, they are difficult to be widely generalized. In this study, we propose a framework of linear interpolation multi-head sparse temporal pattern attention (Li-MSA) based on few-shot learning for power consumption prediction of servers with small-scale datasets in environments such as cloud data centers or edge computing. First, the interpolation reconstruction module extends and smooths the data. Then, the embedding learning module is used to narrow the scope of the hypothesis space. Finally, the multi-head sparse temporal pattern attention module emphasizes features and predicts power consumption. The results of the experiments show that Li-MSA outperforms the best results among the other methods for two datasets with different time steps in the RMSE metric by 15.34%, 17.35%, 18.18%, 6.28%, 4.05%, 7.73%.

Predicting the runtime of tasks is of great significance as it can help users better understand the future runtime consumption of the tasks and make decisions for their heterogeneous devices, or be applied to task scheduling. Learning features from user task history data for predicting task runtime is a mainstream method. However, this method faces many challenges when applied to edge intelligence. In the Big Data era, user devices and data features are constantly evolving, necessitating frequent model retrains. Meanwhile, the noisy data from these devices requires robust methods for valuable insight extraction. In this paper, we propose an edge server-oriented deep neuro-fuzzy system (ESODNFS) that can be trained and inferred on edge servers, for providing users with task runtime prediction services. We divided the dataset and trained it on multiple improved adaptive-network-based fuzzy inference system units (ANFISU), and finally conducted joint training on a deep neural network (DNN). By partitioning the dataset, we reduced the number of parameters for each ANFISU, and at the same time, multiple units can be trained in parallel, supporting fast training and iteration. Additionally, the application of fuzzy inference can effectively learn the features in noisy data and make accurate predictions. The experimental results show that ESODNFS can accurately predict the runtime of real tasks. Compared with other DNN and DNFS, it can achieve good prediction results while reducing training time by over 35%.

Accurate workload prediction is crucial for resource allocation and management in large-scale cloud data centers. While many approaches have been proposed, most existing methods are based on Recurrent Neural Networks (RNNs) or their variants, focusing on short-term cloud workload prediction without considering or identifying the long-term changes and different periodic patterns of cloud workloads. Due to variations in user demands or workload dynamics, cloud workloads that appear stable in the short term often exhibit distinct patterns in the long term. This can lead to a significant decline in prediction accuracy for existing methods when applied to long-term cloud workload forecasting. To address these challenges and overcome the limitations of current approaches, we propose a Multi-Scale Network with Convolutions (MSCNet) for accurate long-term cloud workload prediction. MSCNet employs multi-scale modeling of the original cloud workload to effectively extract multi-scale features and different periodic patterns, learning the long-term dependencies among the cloud workload. Our core component, the Multi-Scale Block, combines the Multi-Scale Patch Block, Transformer Encoder, and Multi-Scale Convolutions Block for comprehensive multi-scale learning. This enables MSCNet to adaptively learn both short-term and long-term features and patterns of cloud workloads, resulting in accurate long-term cloud workload predictions. Extensive experiments are conducted using real-world cloud workload data from Alibaba, Google, and Azure to validate the effectiveness of MSCNet. The experimental results demonstrate that MSCNet achieves accurate long-term cloud workload prediction with a computational complexity of$O(L^{2}d)$, outperforming existing state-of-the-art methods.

Accurate prediction of cloud workload is crucial for effective resource allocation in cloud computing. However, due to the complexity and high dimensionality of workloads in the cloud environment, achieving precise workload prediction is a complex and challenging problem. Current approaches to cloud workload prediction mainly rely on deep learning methods based on the Recurrent Neural Network (RNN), which struggle to capture the long-term dependencies inherent in workloads effectively. To tackle these challenges and overcome the limitations of existing methods, we propose an effective approach Time-Frequency Enhanced Gated Recurrent Unit with Attention (TFEGRU) for cloud workload prediction. First, we design a Time-Frequency Enhanced Block (TFEB) to capture complex workload patterns and extract features from both the frequency and temporal domains. Next, we integrate channel independent strategy and channel embedding into the model to adapt to high-dimensional workloads and enhance predictive performance. Finally, we apply a Gated Recurrent Unit (GRU) in conjunction with a multi-head self-attention mechanism to achieve accurate workload prediction. To validate the effectiveness of TFEGRU, comprehensive experiments are conducted using real-world traces from Google and Alibaba cloud data centers. The experimental results demonstrate that TFEGRU achieves accurate and efficient predictions across diverse cloud workloads, outperforming existing state-of-the-art methods.

With the escalating demand for cloud computing services, data centers (DCs) encounter formidable challenges extending beyond capital investment needs to accommodate increasing computational demands and routine infrastructure maintenance. These challenges include substantial electricity costs due to high energy consumption and the environmental issues caused by high carbon emissions. To reduce costs and mitigate environmental impacts, modern DCs not only use energy-efficient technologies to improve the efficiency of common IT and cooling systems, but also actively optimize the direct, indirect and environmental costs of the power supply side, posing significant challenges for DCs power management. Therefore, this paper presents a comprehensive survey of cost-aware optimization from the power supply perspective. First, it reviews the structures and key metrics of the power supply system, modeling methods and supporting techniques for main power and IT system components, establishing a foundation for optimization. Second, traditional (brown) and green energy sources are categorized to survey and compare existing critical works, analyzing the application of power management methods to tackle cost-related challenges. Finally, future research trends in the power supply perspective for DCs are discussed. This survey aims to provide recommendations for power supply side cost optimization to further advance the sustainable development of DCs.

This paper introduces SparseTSF, a novel, extremely lightweight model for Long-term Time Series Forecasting (LTSF), designed to address the challenges of modeling complex temporal dependencies over extended horizons with minimal computational resources. At the heart of SparseTSF lies the Cross-Period Sparse Forecasting technique, which simplifies the forecasting task by decoupling the periodicity and trend in time series data. This technique involves downsampling the original sequences to focus on cross-period trend prediction, effectively extracting periodic features while minimizing the model’s complexity and parameter count. Based on this technique, the SparseTSF model uses fewer than 1k parameters to achieve competitive or superior performance compared to state-of-the-art models. Furthermore, SparseTSF showcases remarkable generalization capabilities, making it well-suited for scenarios with limited computational resources, small samples, or low-quality data. The code is publicly available at this repository: https://github.com/lss-1138/SparseTSF.

The stable periodic patterns present in time series data serve as the foundation for conducting long-horizon forecasts. In this paper, we pioneer the exploration of explicitly modeling this periodicity to enhance the performance of models in long-term time series forecasting (LTSF) tasks. Specifically, we introduce the Residual Cycle Forecasting (RCF) technique, which utilizes learnable recurrent cycles to model the inherent periodic patterns within sequences, and then performs predictions on the residual components of the modeled cycles. Combining RCF with a Linear layer or a shallow MLP forms the simple yet powerful method proposed in this paper, called CycleNet. CycleNet achieves state-of-the-art prediction accuracy in multiple domains including electricity, weather, and energy, while offering significant efficiency advantages by reducing over 90% of the required parameter quantity. Furthermore, as a novel plug-and-play technique, the RCF can also significantly improve the prediction accuracy of existing models, including PatchTST and iTransformer. The source code is available at: https://github.com/ACAT-SCUT/CycleNet.

Abstract Emerging computing paradigms provide field‐level service responses for users, for example, edge computing, fog computing, and MEC. Edge virtualization technologies represented by Docker can provide a platform‐independent, low‐resource‐consumption operating environment for edge service. The image‐pulling time of Docker is a crucial factor affecting the start‐up speed of edge services. The layer reuse mechanism of native Docker cannot fully utilize the duplicate data of node local images. In this paper, we propose a chunk reuse mechanism (CRM), which effectively targets node‐local duplicate data during container updates and reduces the volume of data transmission required for image building. We orchestrate the CRM process for cloud and remote‐cloud nodes to ensure that the resource overhead from container update data preparation and image reconstruction is within an acceptable range. The experimental results show that the CRM proposed in this paper can effectively utilize the node local duplicate data in the synchronous update of containers in multiple nodes, reduce the volume of data transmission, and significantly improve container update efficiency.

This paper investigates the implementation of Federated Learning (FL) in an over-the-air computation system with volatile clients, where each client operates under a limited energy budget and may unexpectedly drop out during local training sessions. The dropout of clients not only wastes energy but also diminishes their participation frequency, necessitating careful client selection by the server in each communication round. However, the diversity of training tasks and the random nature of client dropout present challenges such as the absence of an explicit objective function and the unavailability of client performance metrics. To address these challenges, we first analyze the convergence of the over-the-air federated learning system with volatile clients to identify the key factor influencing the model's convergence speed. Building upon this analysis, we propose an approximation of the objective function as the optimization goal for client selection. To mitigate energy waste, we introduce a dynamic client selection strategy termed DCSE, based on Exp3 with multiple plays and energy constraints, aiming to reconcile the dilemma of unknown local training states and limited resource constraints. Theoretical analysis demonstrates that our proposed solution maintains a constant bound on the difference from the optimal solution, affirming its theoretical feasibility. Furthermore, experimental results validate the effectiveness of the proposed strategy in enhancing FL by accelerating convergence speed, improving test accuracy, and reducing wasted energy.

As a promising paradigm, edge computing enhances service provisioning by offloading tasks to powerful servers at the network edge. Meanwhile, Non-Orthogonal Multiple Access (NOMA) and renewable energy sources are increasingly adopted for spectral efficiency and carbon footprint reduction. However, these new techniques inevitably introduce reliability risks to the edge system generally because of i) imperfect Channel State Information (CSI), which can misguide offloading decisions and cause transmission outages, and ii) unstable renewable energy supply, which complicates device availability. To tackle these issues, we first establish a system model that measures service reliability based on probabilistic principles for the NOMA-based edge system. As a solution, a Reliable Offloading method with Multi-Agent deep reinforcement learning (ROMA) is proposed. In ROMA, we first reformulate the reliability-critical constraint into an long-term optimization problem via Lyapunov optimization. We discretize the hybrid action space and convert the resource allocation on edge servers into a 0-1 knapsack problem. The optimization problem is then formulated as a Partially Observable Markov Decision Process (POMDP) and addressed by multi-agent proximal policy optimization (PPO). Experimental evaluations demonstrate the superiority of ROMA over existing methods in reducing grid energy costs and enhancing system reliability, achieving Pareto-optimal performance under various settings.

Constructing energy-efficient cloud data centers (CDCs) is an essential path for the further expansion of cloud computing. As one of the core subsystems of a data center, the cooling system provides a reliable thermal environment for the safe operation of IT equipment while posing a huge energy consumption and carbon emission problem. Thus, it is evident that optimizing energy management of cooling systems with considerable energy-saving potential will be essential to realize the green and low-carbon development of CDCs. Therefore, to track the research progress of data center thermal management technologies, this review focuses on two research efforts: thermal modeling and thermal-aware energy saving methods. First, various thermal modeling approaches are reviewed for air-cooled and liquid-cooled data centers. Secondly, a comprehensive review of existing advanced thermal management approaches is conducted from three perspectives: thermal-aware IT load scheduling, cooling system control optimization, and joint optimization of the IT and cooling systems. Finally, we put forward some open issues and future research directions for thermal management that have not been completely solved. This review aims to provide reasonable suggestions to enhance cooling energy efficiency and further promote the transformation of CDCs to lower energy consumption and sustainable direction.

Emerging cloud computing applications place a growing demand on resources, leading to increasingly large data centers with significant energy consumption and carbon emissions. Various research conduct optimization methods to improve the energy efficiency of the server in the cloud data center. However, most existing optimization methods are designed for specific applications, thus making it difficult to handle complex cloud environments. In this paper, we propose a general parameter optimization method called MPOD to improve the energy efficiency of cloud servers in real time. MPOD considers issues in the cloud environment, such as SLA guarantee, user privacy, and dynamic workloads. We introduce energy efficiency curves to DVFS, implementing a low-overhead, fast response, and general frequency optimization strategy. Moreover, we design a workload classification framework and three prediction models based on machine learning algorithms to achieve accurate and adaptive Linux kernel parameters optimization. According to the experiment, MPOD can improve the energy efficiency of the server by an average of 30.5%, 20.1%, 10.8% in BenchSEE, SERT and TPC-H, respectively.

Today's Industrial Internet of Things (IIoT) have achieved excellent manufacturing efficiency and automation results by leveraging machine learning (ML) and deep learning (DL). However, trustworthiness of ML/DL brings significant challenges to IIoT. This article proposes an evolving deep multiple kernel learning network through genetic algorithm (KNGA). Our KNGA method uses genetic algorithm (GA) to find the best deep multiple kernel learning structure, including the weights and the topology of the model. Compared with the current well-known models, KNGA has advantages in three aspects: 1) It can achieve good results without using many samples during model training; 2) the model can evolve in the process of training, including self-growth, and self-pruning; and 3) its trustworthiness and reliability can be guaranteed. Moreover, the whole model ensures excellent performance and requires manual adjustment of only a few parameters. Extensive experiments on the UCI, KEEL, Caltech256, and MNIST datasets demonstrate the effectiveness and trustworthiness of the proposed method.

Federated Learning (FL) has shown great potential as a privacy-preserving solution to learning from decentralized data that are only accessible to end devices (i.e., clients). The data locality constraint offers strong privacy protection but also makes FL sensitive to the condition of local data. Apart from statistical heterogeneity, a large proportion of the clients, in many scenarios, are probably in possession of low-quality data that are biased, noisy or even irrelevant. As a result, they could significantly slow down the convergence of the global model we aim to build and also compromise its quality. In light of this, we first present a new view of local data by looking into the representation space and observing that they converge in distribution to Normal distributions before activation. We provide theoretical analysis to support our finding. Further, we proposeFedProf, a novel algorithm for optimizing FL over non-IID data of mixed quality. The key of our approach is a distributional representation profiling and matching scheme that uses the global model to dynamically profile data representations and allows for low-cost, lightweight representation matching. Using the scheme we sample clients adaptively in FL to mitigate the impact of low-quality data on the training process. We evaluated our solution with extensive experiments on different tasks and data conditions under various FL settings. The results demonstrate that the selective behavior of our algorithm leads to a significant reduction in the number of communication rounds and the amount of time (up to 2.4× speedup) for the global model to converge and also provides accuracy gain.

To overcome the challenge of limited bandwidth, client selection has been considered an effective method for optimizing Federated Learning (FL). However, since the volatility of the learning environment, the available clients exhibit some volatility over the training process in terms of client population, client data, training status, and transmitting status, which greatly increases the difficulty of client selection. To find a practical solution, we explore a client selection problem in volatile federated learning (Volatile FL). Specifically, we first derive the convergence analysis for non-convex and strongly convex cases to illustrate the main factors affecting the convergence speed. Then, we introduce the client utility to quantify the client’s contribution to model training and discuss the key problems of client selection in Volatile FL. For an efficient settlement, we propose CU-CS, a Combinatorial Multi-Arm Bandit (C2MAB) based decision scheme for the proposed selection problem. Theoretically, we prove that the regret of CU-CS is strictly bounded by a finite constant, justifying its theoretical feasibility. The experimental results demonstrate that our method significantly boosts FL by speeding up model convergence, promoting model accuracy, and reducing energy consumption.

Federated Learning (FL), as a privacy-preserving machine learning paradigm, has been thrusted into the limelight. As a result of the physical bandwidth constraint, only a small number of clients are selected for each round of FL training. However, existing client selection solutions (e.g., the vanilla random selection) typically ignore the heterogeneous data value of the clients. In this paper, we propose the contribution-based selection algorithm (Contribution-Based Exponential-weight algorithm for Exploration and Exploitation, CBE3), which dynamically updates the selection weights according to the impact of clients' data. As a novel component of CBE3, a scaling factor, which helps maintain a good balance between global model accuracy and convergence speed, is proposed to improve the algorithm's adaptability. Theoretically, we proved the regret bound of the proposed CBE3 algorithm, which demonstrates performance gaps between the CBE3 and the optimal choice. Empirically, extensive experiments conducted on Non-Independent Identically Distributed data demonstrate the superior performance of CBE3—with up to 10% accuracy improvement compared with K-Center and Greedy and up to 100% faster convergence compared with the Random algorithm.

The redundancy in the benchmark suite will increase the time for computer system performance evaluation and simulation. The most typical method to solve this problem is to select subsets based on clustering. However, it is a challenge to validate benchmark subsetting results for unlabeled benchmark suites when using the clustering method, and existing research has not considered this problem. Also, there is no quantitative evaluation method for subsetting which can reflect the universal and the diversity characteristics of the benchmark suite at the same time. To solve the above problems, we propose BenchSubset, a framework for selecting benchmark subsets based on consensus clustering, which includes Group Principal Components Analysis, consensus clustering, and a new evaluation method considering the universal and the diversity characteristics of the benchmark suite. We conducted SPEC CPU2017 subsetting experiments on Huawei's Taishan 200, then verified the effectiveness of BenchSubset in selecting a benchmark subset. Compared with the mainstream principal components analysis with hierarchical clustering (PCA-H) method, the benchmark subset selected by BenchSubset performs better in representing the universal and the diversity characteristics of SPEC CPU2017.

With astonishing speed, bandwidth, and scale, mobile-edge computing (MEC) has played an increasingly important role in the next generation of connectivity and service delivery. Yet, along with the massive deployment of MEC servers, the ensuing energy issue is now on an increasingly urgent agenda. In the current context, the large-scale deployment of renewable-energy-supplied MEC servers is perhaps the most promising solution for the incoming energy issue. Nonetheless, as a result of the intermittent nature of their power sources, these special design MEC servers must be more cautious about their energy usage, in a bid to maintain their service sustainability as well as service standard. Targeting optimization on a single-server MEC scenario, we, in this article, propose neural network-based adaptive frequency adjustment (NAFA), an adaptive processor frequency adjustment solution, to enable an effective plan of the server’s energy usage. By learning from the historical data revealing request arrival and energy harvest pattern, the deep reinforcement learning-based solution is capable of making intelligent schedules on the server’s processor frequency, so as to strike a good balance between service sustainability and service quality. The superior performance of NAFA is substantiated by real-data-based experiments, wherein NAFA demonstrates up to 20% increase in the average request acceptance ratio and up to 50% reduction in average request processing time.

Federated learning (FL), arising as a privacy-preserving machine learning paradigm, has received notable attention from the public. In each round of synchronous FL training, only a fraction of available clients are chosen to participate, and the selection decision might have a significant effect on the training efficiency, as well as the final model performance. In this article, we investigate the client selection problem under a volatile context, in which the local training of heterogeneous clients is likely to fail due to various kinds of reasons and in different levels of frequency. Intuitively, too much training failure might potentially reduce the training efficiency, while too much selection on clients with greater stability might introduce bias, thereby resulting in degradation of the training effectiveness. To tackle this tradeoff, we, in this article, formulate the client selection problem under joint consideration of effective participation and fairness. Furthermore, we propose E3CS, a stochastic client selection scheme as a solution. According to our experimental results over a public data set, the proposed selection scheme is able to achieve up to$2\times $faster convergence to a fixed model accuracy while maintaining the same level of final model accuracy, compared with the state-of-the-art selection schemes.

Federated learning (FL) has shown great potential as a privacy-preserving solution to training a centralized model based on local data from available clients. However, we argue that, over the course of training, the available clients may exhibit some volatility in terms of the client population, client data, and training status. Considering these volatilities, we propose a new learning scenario termed volatile federated learning (volatile FL) featuring set volatility, statistical volatility, and training volatility. The volatile client set along with the dynamic of clients’ data and the unreliable nature of clients (e.g., unintentional shutdown and network instability) greatly increase the difficulty of client selection. In this article, we formulate and decompose the global problem into two subproblems based on alternating minimization. For an efficient settlement for the proposed selection problem, we quantify the impact of clients’ data and resource heterogeneity for volatile FL and introduce the cumulative effective participation data (CEPD) as an optimization objective. Based on this, we propose upper confidence bound-based greedy selection, dubbed UCB-GS, to address the client selection problem in volatile FL. Theoretically, we prove that the regret of UCB-GS is strictly bounded by a finite constant, justifying its theoretical feasibility. Furthermore, experimental results show that our method significantly reduces the number of training rounds (by up to 62%) while increasing the global model’s accuracy by 7.51%.

Abstract As a key technology in the 5G era, mobile edge computing (MEC) has developed rapidly in recent years. MEC aims to reduce the service delay of mobile users, while alleviating the processing pressure on the core network. MEC can be regarded as an extension of cloud computing on the user side, which can deploy edge servers and bring computing resources closer to mobile users, and provide more efficient interactions. However, due to the user's dynamic mobility, the distance between the user and the edge server will change dynamically, which may cause fluctuations in Quality of Service. Therefore, when a mobile user moves in the MEC environment, certain approaches are needed to schedule services deployed on the edge server to ensure the user experience. In this article, we model service scheduling in MEC scenarios and propose a delay‐aware and mobility‐aware service management approach based on concise probabilistic methods. This approach has low computational complexity and can effectively reduce service delay and migration costs. Furthermore, we conduct experiments by utilizing multiple realistic datasets and use iFogSim to evaluate the performance of the algorithm. The results show that our proposed approach can optimize the performance on service delay, with 8%–20% improvement and reduce the migration cost by more than 75% compared with baselines during the rush hours.

On-line detection of anomalies in time series is a key technique used in various event-sensitive scenarios such as robotic system monitoring, smart sensor networks and data center security. However, the increasing diversity of data sources and the variety of demands make this task more challenging than ever. First, the rapid increase in unlabeled data means supervised learning is becoming less suitable in many cases. Second, a large portion of time series data have complex seasonality features. Third, on-line anomaly detection needs to be fast and reliable. In light of this, we have developed a prediction-driven, unsupervised anomaly detection scheme, which adopts a backbone model combining the decomposition and the inference of time series data. Further, we propose a novel metric, Local Trend Inconsistency (LTI), and an efficient detection algorithm that computes LTI in a real-time manner and scores each data point robustly in terms of its probability of being anomalous. We have conducted extensive experimentation to evaluate our algorithm with several datasets from both public repositories and production environments. The experimental results show that our scheme outperforms existing representative anomaly detection algorithms in terms of the commonly used metric, Area Under Curve (AUC), while achieving the desired efficiency.

In addition to the stationary mobile edge computing (MEC) servers, a few MEC surrogates that possess a certain mobility and computation capacity, e.g., flying unmanned aerial vehicles (UAVs) and private vehicles, have risen as powerful counterparts for service provision. In this article, we design a two-stage online scheduling scheme, targeting computation offloading in a UAV-assisted MEC system. On our stage-one formulation, an online scheduling framework is proposed for dynamic adjustment of mobile users' CPU frequency and their transmission power, aiming at producing a socially beneficial solution to users. But the major impediment during our investigation lies in that users might not unconditionally follow the scheduling decision released by servers as a result of their individual rationality. In this regard, we formulate each step of online scheduling on stage one into a non-cooperative game with potential competition over the limited radio resource. As a solution, a centralized online scheduling algorithm, called ONCCO, is proposed, which significantly promotes social benefit on the basis of the users' individual rationality. On our stage-two formulation, we are working towards the optimization of UAV computation resource provision, aiming at minimizing the energy consumption of UAVs during such a process, and correspondingly, another algorithm, called WS-UAV, is given as a solution. Finally, extensive experiments via numerical simulation are conducted for an evaluation purpose, by which we show that our proposed algorithms achieve satisfying performance enhancement in terms of energy conservation and sustainable service provision.

As data centers are consuming massive amount of energy, improving the energy efficiency of cloud computing has emerged as a focus of research. However, it is challenging to reduce energy consumption while maintaining system performance without increasing the risk of Service Level Agreement violations. Most of the existing consolidation approaches for virtual machines (VMs) consider system performance and Quality of Service (QoS) metrics as constraints, which usually results in large scheduling overhead and impossibility to achieve effective improvement in energy efficiency without sacrificing some system performance and cloud service quality. In this article, we first define the metrics of peak power efficiency and optimal utilization for heterogeneous physical machines (PMs). Then we propose Peak Efficiency Aware Scheduling (PEAS), a novel strategy of VM placement and reallocation for achieving dual improvement in performance and energy conservation from the perspective of server clusters. PEAS allocates and reallocates VMs in an on-line manner and always attempts to maintain PMs working in their peak power efficiency via VM consolidation. Extensive experiments on Cloudsim show that PEAS outperforms several energy-aware consolidation algorithms with regard to energy consumption, system performance as well as multiple QoS metrics.

Federated learning (FL) has attracted increasing attention as a promising approach to driving a vast number of end devices with artificial intelligence. However, it is very challenging to guarantee the efficiency of FL considering the unreliable nature of end devices while the cost of device-server communication cannot be neglected. In this article, we propose SAFA, a semi-asynchronous FL protocol, to address the problems in federated learning such as low round efficiency and poor convergence rate in extreme conditions (e.g., clients dropping offline frequently). We introduce novel designs in the steps of model distribution, client selection and global aggregation to mitigate the impacts of stragglers, crashes and model staleness in order to boost efficiency and improve the quality of the global model. We have conducted extensive experiments with typical machine learning tasks. The results demonstrate that the proposed protocol is effective in terms of shortening federated round duration, reducing local resource wastage, and improving the accuracy of the global model at an acceptable communication cost.

Leveraging power consumption models in software systems can achieve easy deployment of low-cost, high-availability power monitoring in cloud datacenters that are usually large-scale, heterogeneous and frequently scaling up. However, traditional regression-based power consumption models generally have two drawbacks. First, their mathematical forms are usually fixed and determined a priori. This may cause unacceptable increase of error or over-fitting as the power signatures of cloud servers are usually uncertain. Second, the characteristic of workload dispatched to cloud servers is constantly changing while regression-based models can hardly generalize to a wide range of servers and workload types. As a novel solution, we in this paper propose a server power consumption model based on Elman Neural Network (PCM-ENN), aiming to allow accurate and flexible power estimation. PCM-ENN is an end-to-end black box model capable of learning the temporal relation between samples in a time series of power consumption. We trained and evaluated PCM-ENN on two power sequence datasets collected from heterogeneous hardware and operating systems running quasi-production benchmarks like CloudSuite. Experimental result shows that PCM-ENN generated accurate estimates on server power consumption with only small errors, outperforming widely-used linear regression model and NARX model in terms of accuracy.

In recent years, fog computing has emerged as a new paradigm for the future Internet-of-Things (IoT) applications, but at the same time, ensuing new challenges. The geographically vast-distributed architecture in fog computing renders us almost infinite choices in terms of service orchestration. How to properly arrange the service replicas (or service instances) among the nodes remains a critical problem. To be specific, in this article, we investigate a generalized service replicas placement problem that has the potential to be applied to various industrial scenarios. We formulate the problem into a multiobjective model with two scheduling objectives, involving deployment cost and service latency. For problem solving, we propose an ant colony optimization-based solution, called multireplicas Pareto ant colony optimization (MRPACO). We have conducted extensive experiments on MRPACO. The experimental results show that the solutions obtained by our strategy are qualified in terms of both diversity and accuracy, which are the main evaluation metrics of a multiobjective algorithm.

In mobile crowdsensing, it is a challenge to assign tasks to appropriate smartphones. Existing task allocation mechanisms mainly aim at optimizing the global system performance, while ignoring the personal preferences of individual crowdsensing tasks and smartphone users. Nevertheless, in an open crowdsensing system, a task assignment is prone to be unstable if smartphone users or tasks have incentives to deviate from the global assignment, and seek for alternative choices to improve their own utilities. Besides that, during task competition, the rational smartphone users might choose to adjust their payments after the first few failures, which however, brings new challenges in achieving the stability. To address these issues, this paper constructs a distributed many-to-many matching model to capture the interaction between crowdsensing tasks and smartphone users, taking into account the budget constraints of tasks. Then, we design a stable matching algorithm to allocate the tasks to the users, and determine their payments. We prove that the proposed algorithm achieves several desirable properties including individual rationality, stability, and convergency. It is also proved that the proposed scheme achieves at least half of the optimal system efficiency when each smartphone provides homogeneous service quality. Finally, simulation results confirm the effectiveness of the proposed scheme.

The issue of potential privacy leakage during centralized AI's model training has drawn intensive concern from the public. A Parallel and Distributed Computing (or PDC) scheme, termed Federated Learning (FL), has emerged as a new paradigm to cope with the privacy issue by allowing clients to perform model training locally, without the necessity to upload their personal sensitive data. In FL, the number of clients could be sufficiently large, but the bandwidth available for model distribution and re-upload is quite limited, making it sensible to only involve part of the volunteers to participate in the training process. The client selection policy is critical to an FL process in terms of training efficiency, the final model's quality as well as fairness. In this article, we will model the fairness guaranteed client selection as a Lyapunov optimization problem and then a C2MAB-based method is proposed for estimation of the model exchange time between each client and the server, based on which we design a fairness guaranteed algorithm termed RBCS-F for problem-solving. The regret of RBCS-F is strictly bounded by a finite constant, justifying its theoretical feasibility. Barring the theoretical results, more empirical data can be derived from our real training experiments on public datasets.

Mobile Edge Computing (MEC), which incorporates the Cloud, edge nodes, and end devices, has shown great potential in bringing data processing closer to the data sources. Meanwhile, Federated learning (FL) has emerged as a promising privacy-preserving approach to facilitating AI applications. However, it remains a big challenge to optimize the efficiency and effectiveness of FL when it is integrated with the MEC architecture. Moreover, the unreliable nature (e.g., stragglers and intermittent drop-out) of end devices significantly slows down the FL process and affects the global model's quality in such circumstances. In this article, a multi-layer federated learning protocol called HybridFL is designed for the MEC architecture. HybridFL adopts two levels (the edge level and the cloud level) of model aggregation enacting different aggregation strategies. Moreover, in order to mitigate stragglers and end device drop-out, we introduce regional slack factors into the stage of client selection performed at the edge nodes using a probabilistic approach without identifying or probing the state of end devices (whose reliability is agnostic). We demonstrate the effectiveness of our method in modulating the proportion of clients selected and present the convergence analysis for our protocol. We have conducted extensive experiments with machine learning tasks in different scales of MEC system. The results show that HybridFL improves the FL training process significantly in terms of shortening the federated round length, speeding up the global model's convergence (by up to 12×) and reducing end device energy consumption (by up to 58 percent).

Virtual machines consolidation is known as a powerful means to reduce the number of activated physical machines (PMs), so as to achieve energy-saving for the data centers. Although the consolidation technique is widely studied in non-NUMA systems, we could only trace a few studies targeting NUMA systems. But the virtual machines (VMs) deployment of NUMA systems is quite different from that of non-NUMA systems. More specifically, consolidating VMs in NUMA systems need to decide both target physical machines and NUMA architectures to host the VMs, and more complicated constraints originated from the real usage of NUMA systems that need to be considered. Being motivated by these challenges, we in this paper formally derive the system model according to the real business model of NUMA systems and based on which, we propose a hybrid heuristics swarm intelligence optimization algorithm HHGWA for an efficient solution. To do the evaluation, extensive simulations that integrate real VM and PM information are conducted, the result of which indicates a superior performance of our proposed algorithm.

Nowadays, cloud co-location is often used for data centers to improve the utilization of computing resources. However, batch jobs in a Co-location Datacenter (CLD) are vulnerable to failures due to the competition for limited resources with online service jobs. Such failed batch jobs would be rescheduled and failed repeatedly, resulting in the waste of computing resources and instability of the computing clusters. Therefore, we propose a method to accurately predict the potential failures of batch jobs for CLD. The core of the proposed method is STLF (SMOTE Tomek and LightGBM [5] Framework), which is divided into three parts. First, we use the co-feature extraction method to generate Co-located Feature Dataset (CLFD). Then SMOTE Tomek is used to oversampling the CLFD to ensure that the classifier can learn more minority features. Finally, we use LightGBM classifier to predict batch jobs' failure. The performance experiments conducted on the Ali Trace 2018 dataset show that our proposed STLF significantly outperforms the existing popular classifiers in terms of the ROC curve, the area under the ROC curve (AUC), precision, and recall.

Job scheduling is a necessary prerequisite for performance optimization and resource management in the cloud computing system. Focusing on accurate scaled cloud computing environment and efficient job scheduling under Virtual Machine (VM) resource and Server Level Agreement (SLA) constraints, we introduce the architecture of cloud computing platform and optimization job scheduling scheme in this study. The system model is comprised of clearly defined separate constituent parts, including portal, job scheduler, and resources pool. By analyzing the execution process of user jobs, we designed a novel job scheduling scheme based on reinforcement learning to minimize the makespan and Average Waiting Time (AWT) under the VM resource and deadline constraints, and employ parallel multi-age parallel technologies to balance the exploration and exploitation in learning process and accelerate the convergence of Q-learning algorithm. Both simulation and real cloud platform experiment results demonstrate the efficiency of the proposed job scheduling scheme.

The power consumption estimation or prediction of cloud servers is the basis of energy-aware scheduling to realize energy saving in cloud datacenters. The existing works are mainly based on the static mathematical formulas which establish the relationship between the server power consumption and the system performance. However, these models are weak in adaptability and generalization ability, not adaptable to the changes and fluctuation of different workload, and demanding on the clear and profound understanding of the inner relationship among related power consumption parameters. Therefore, we propose the ANN (Artificial Neural Network) method to model the power consumption of the servers in datacenters, a kind of end-to-end black box model. We performed a fine-grained and in-depth analysis about the system performance and power consumption characteristics of the CPU, memory, and disk of the server running different types of task loads, and selected a set of performance counters that can fully reflect the status of system power consumption as the input of the model. Then, we establish power consumption models based on BP neural network, Elman neural network, and LSTM neural network, respectively. In order to get a better result, we use data collected from four different types of task loads (i.e., CPU-intensive, memory-intensive, I/O-intensive, and mixed load) to train, validate, and test our target models. The experimental results show that, compared with multiple linear regression and support vector regression, the proposed three power models have better performance in predicting the server's real-time power consumption.

To study the influence of market characteristics on stock prices, traditional neural network algorithm may also fail to predict the stock market precisely, since the initial weight of the random selection problem can be easily prone to incorrect predictions. Based on the idea of word vector in deep learning, we demonstrate the concept of stock vector. The input is no longer a single index or single stock index, but multi-stock high-dimensional historical data. We propose the deep long-short term memory neural network (LSMN) with embedded layer to predict the stock market. In this model, we use the embedded layer to vectorize the data, in a bid to forecast the stock via long-short term memory neural network. The experimental results show that the deep long short term memory neural network with embedded layer is state-of-the-art in developing countries. Specifically, the accuracy of this model is 57.2% for the Shanghai A-shares composite index. Furthermore, this is 52.4% for individual stocks.

Summary Data structure description, conceptual modeling, and logic reasoning for knowledge discovery are three critical factors for the integration of information with heterogeneity. In particular, technologies of NoSQL databases and Internet of Things raise an urgent requirement for a uniform expression of heterogeneous data, and little attention has been paid to researches on the integration of NoSQL databases with traditional data models, as well as the semantic description of big data. To tackle these problems, in this paper, a concept‐and‐relation‐oriented grid data model called GODM model is first proposed based on the definitions of Monad, Compounder, Relation, etc. Then, the GODM model is utilized to uniformly describe traditional data models and NoSQL data models, which eliminates structure differences of heterogeneous data. Next, based on the GODM relation mechanism, an extendable semantic system is built up by choosing SHOIQ(D) description logic as the example to establish the correspondence with GODM grammar subset, providing a fundamental support for semantic integration and knowledge discovery of heterogeneous data. After that, comprehensive comparisons with GODM and other models are made, especially the distinctions between GODM and OWL on the aspects of relation mechanism, hybrid schema, description logic, grammatical constructors, etc. Besides, experimental evaluations and analyses on time and space efficiencies of some primary common data models are conducted after the proposal of a general evaluation model, with the results showing that the GODM model has great advantage on properties of expressiveness, flexibility, etc, particularly time and space efficiency. In summary, the GODM model describes heterogeneous data from both aspects of data structure and semantic relationship and realizes a hybrid schema reconciling the schemaful and schemaless data models, making it especially suitable for dynamic data integration and knowledge discovery from big data models.

The present study constructs a multi-agent organizational framework with virtual machine clusters as the unit of granularity based on the results of that previous study. In addition, the present study also investigates the requirements for an orderly and efficient workflow of this multi-agent organizational framework and constructs an abstract system structure based on collaborative agents (including vms-agents and apps-agents) and a collaborative mechanism. Finally, the present study reports a comparative experiment conducted on the CloudSim platform. The experimental results show that the proposed multi-agent organization method with virtual machine clusters as the unit of granularity improves the efficiency of resource utilization and accounts for the interests of both cloud users and providers while ensuring the maintenance of the application service level agreement. Compared with the mechanism of the multi-agent organizational method proposed in the present study, existing studies have generally considered the configuration of a given party only locally and in isolation, severing the intrinsic connection between the resources of virtual machines and the application systems/components that are operating on those virtual machines. As a result, the interests of the cloud users and providers cannot be simultaneously ensured.

Summary This paper presents a novel list‐based scheduling algorithm called Improved Predict Earliest Finish Time for static task scheduling in a heterogeneous computing environment. The algorithm calculates the task priority with a pessimistic cost table, implements the feature prediction with a critical node cost table, and assigns the best processor for the node that has at least 1 immediate successor as the critical node, thereby effectively reducing the schedule makespan without increasing the algorithm time complexity. Experiments regarding aspects of randomly generated graphs and real‐world application graphs are performed, and comparisons are made based on the scheduling length ratio, robustness, and frequency of the best result. The results demonstrate that the Improved Predict Earliest Finish Time algorithm outperforms the Predict Earliest Finish Time and Heterogeneous Earliest Finish Time algorithms in terms of the schedule length ratio, frequency of the best result, and robustness while maintaining the same time complexity.

Job scheduling is a necessary prerequisite for performance optimization and resource management in the cloud computing system. Focusing on accurate scaled cloud computing environment and efficient job scheduling under Virtual Machine (VM) resource and Server Level Agreement (SLA) constraints, we introduce the architecture of cloud computing platform and optimization job scheduling scheme in this study. The system model is comprised of clearly defined separate constituent parts, including portal, job scheduler, and resources pool. By analyzing the execution process of user jobs, we designed a novel job scheduling scheme based on reinforcement learning to minimize the makespan and Average Waiting Time (AWT) under the VM resource and deadline constraints, and employ parallel multi-age parallel technologies to balance the exploration and exploitation in learning process and accelerate the convergence of Q-learning algorithm. Both statistical and numerical analysis results demonstrate the efficiency of the proposed job scheduling scheme.

Reinforcement learning is an area of machine learning inspired by behaviorist psychology, concerned with how software agents ought to take actions in an environment so as to maximize some notion of cumulative reward. Recently reinforcement learning has been given abroad attention, but when it is applied to solve problems with large-scale discrete or contiguous state space environments, the results are likely to be unsatisfactory and even fail to find optimal policies. In order to solve this problem, we establish a new generative model about the value function and use Gaussian Process Regression to approximate the state-action pairs which were never or seldom visited. We testify to the performance of the proposed algorithm by an access-control queuing job in a cloud computing environment. The computational results demonstrate the scheme can balance the exploration and exploitation in the learning process and accelerate the convergence to a certain extent.

Summary The topology of interconnection networks plays an important role in the performance of parallel and distributed computing systems. In this paper, we propose a new interconnection network called twisted crossed cube (TCQn) and investigate its basic network properties in terms of the regularity, connectivity, fault tolerance, recursiveness, hamiltonicity and ability to simulate other architectures, and so on. Then, we develop an effective routing algorithm Route (u, v) for TCQn that takes no more than d(u, v) + 1 steps for any two nodes (u, v) to communicate with each other, and the routing process shows that the diameter, wide diameter, and fault‐tolerant diameter of TCQn are about half of the corresponding diameters of the equivalent hypercube with the same dimension. In the end, by combining TCQn with crossed cube (CQn), we propose a preferable dynamic network structure, that is, the dynamic crossed cube, which has the same network diameter as TCQn/CQn and better properties in other respects, for example, its connection complexity is half of that of TCQn/CQn when the network scale is large enough, and the number of its average routing steps is also much smaller than that in TCQn/CQn. Copyright © 2015 John Wiley & Sons, Ltd.

SUMMARY Task scheduling is a fundamental issue in achieving high efficiency in cloud computing. However, it is a big challenge for efficient scheduling algorithm design and implementation (as general scheduling problem is NP‐complete). Most existing task‐scheduling methods of cloud computing only consider task resource requirements for CPU and memory, without considering bandwidth requirements. In order to obtain better performance, in this paper, we propose a bandwidth‐aware algorithm for divisible task scheduling in cloud‐computing environments. A nonlinear programming model for the divisible task‐scheduling problem under the bounded multi‐port model is presented. By solving this model, the optimized allocation scheme that determines proper number of tasks assigned to each virtual resource node is obtained. On the basis of the optimized allocation scheme, a heuristic algorithm for divisible load scheduling, called bandwidth‐aware task‐scheduling (BATS) algorithm, is proposed. The performance of algorithm is evaluated using CloudSim toolkit. Experimental result shows that, compared with the fair‐based task‐scheduling algorithm, the bandwidth‐only task‐scheduling algorithm, and the computation‐only task‐scheduling algorithm, the proposed algorithm (BATS) has better performance. Copyright © 2012 John Wiley & Sons, Ltd.