基于多维度特征融合的云工作流任务执行时间预测方法

任务执行时间估计是云数据中心环境下工作流调度的前提.针对现有工作流任务执行时间预测方法缺乏类别型和数值型数据特征的有效提取问题,提出了基于多维度特征融合的预测方法.首先,通过构建具有注意力机制的堆叠残差循环网络,将类别型数据从高维稀疏的特征空间映射到低维稠密的特征空间,以增强类别型数据的解析能力,有效提取类别型特征;其次,采用极限梯度提升算法对数值型数据进行离散化编码,通过对稠密空间的输入向量进行稀疏化处理,提高了数值型特征的非线性表达能力;在此基础上,设计多维异质特征融合策略,将所提取的类别型、数值型特征与样本的原始输入特征进行融合,建立基于多维融合特征的预测模型,实现了云工作流任务执行时间的精准预测;最后,在真实云数据中心集群数据集上进行了仿真实验.实验结果表明,相对于已有的基准算法,该方法具有较高的预测精度,可用于大数据驱动的云工作流任务执行时间预测.     

基于动态系数的三次指数平滑算法负载预测

数据中心是企业信息化的重要组成部分,云计算的核心思想就是把数据中心整成一个资源池,对资源池进行统一的调度与管理。随着虚拟化技术的发展,目前对数据中心的资源利用率越来越高,但是还是存在大量资源浪费的情况,其原因在于当前对数据中心未来负载预测的算法还存在一定的局限性,如果对未来负载预测值远远大于实际负载情况,则导致大量的虚拟机资源利用率不高,反之则会导致虚拟机的资源使用率消耗增大,云平台中不同物理服务器之间的负载情况不平衡,一部分物理服务器负载过大,导致云计算平台响应时间过长。因此云计算平台选取一个合适的负载预测算法显得越发重要,如何权衡以上问题,是云计算里面的一个重点研究方向。负载预测选取时间序列预测算法中的三次指数平滑法,在该算法原有的静态系数基础之上,设计了一种动态系数提取方法。通过等距法把静态系数分成若干份进行训练,然后在预测过程中提取该时段误差最小值所对应的系数。在预测结束后,重新计算其误差,并通过均值法覆盖旧误差。实验结果表明,基于自适应三次指数平滑算法其预测误差明显小于静态系数所预测的误差,计算复杂度低,具有一定的应用价值。     

Fuzzy Ants and Clustering

A swarm-intelligence-inspired approach to clustering data is described. The algorithm consists of two stages. In the first stage of the algorithm, ants move the cluster centers in feature space. The cluster centers found by the ants are evaluated using a reformulated fuzzy C-means (FCM) criterion. In the second stage, the best cluster centers found are used as the initial cluster centers for the FCM algorithm. Results on 18 data sets show that the partitions found using the ant initialization are better optimized than those obtained from random initializations. The use of a reformulated fuzzy partition validity metric as the optimization criterion is shown to enable determination of the number of cluster centers in the data for several data sets. Hard C-means (HCM) was also used after reformulation, and the partitions obtained from the ant-based algorithm were better optimized than those from randomly initialized HCM.     

PEACE-Home: Probabilistic estimation of abnormal clinical events using vital sign correlations for reliable home-based monitoring

The objective of this study is to develop a probabilistic model for predicting the future clinical episodes of a patient using observed vital sign values prior to the clinical event. Vital signs (e.g. heart rate, blood pressure) are used to monitor a patient’s physiological functions of health and their simultaneous changes indicate a transition of a patient’s health status. If such changes are abnormal then it may lead to serious physiological deterioration. Chronic patients living alone at home die of various diseases due to the lack of an efficient automated system having prior prediction ability. Our developed system can make probabilistic predictions of future clinical events of an unknown patient in real-time using the learned temporal correlations of multiple vital signs from many similar patients. In this paper, Principal Component Analysis (PCA) is used to separate patients with known medical conditions into multiple categories and then Hidden Markov Model (HMM) is adopted for probabilistic classification and prediction of future clinical states. The advantage of using dynamic probabilistic model over static predictor model for solving our problem is analysed by comparing the results obtained from HMM with a neural network based learning model. Both the learning models are trained and evaluated using six vital signs data of 1023 patient records collected from the MIMIC-II database of MIT physiobank archive. The best HMM models are selected using maximum likelihood probabilities and further used in personalized remote monitoring system to forecast the most probable forthcoming clinical states of a continuously monitored patient. The scalable power of cloud computing is utilized for fast learning of various clinical events from large samples. Our results suggest that the developed technique using multiple physiological parameter trends can significantly enhance the traditional home-based monitoring systems in terms of clinical abnormality detections and predictions.     

基于最大偏差相似性准则的BP神经网络短期电力负荷预测算法

针对企业电力负荷随机性强、稳定性低、预测精度不理想等问题,提出了一种基于最大偏差相似性准则的BP神经网络短期电力负荷预测算法。首先对最大偏差相似性准则算法进行修改,并提出使用预测日的负荷特征向量与最大偏差相似性准则算法聚类之后的类中心负荷特征的距离来确定预测日的相似日类别;然后将聚类后的相似日类别负荷数据作为BP网络的训练数据,输出预测日起始的连续三天96整点负荷值。实验表明,该方法提出的短期电力负荷预测方法在精度和网络训练时间上都有较大的提升,具有较高的有效性和实用性。     

基于负载预测的虚拟机动态调度算法研究与实现

在云计算系统中为了实现负载均衡和资源的高效利用,需要在虚拟机粒度上对云计算系统进行调度,通过热迁移技术将虚拟机从高负载物理节点迁移到低负载物理节点。把负载预测技术和虚拟机动态调度技术相结合,提出了LFS算法,通过虚拟机历史负载数据对虚拟机未来的负载变化情况进行预测,然后根据预测结果对虚拟机进行调度,能够有效地避免云计算系统中高负载物理节点出现,实现负载均衡,提高资源使用率。     

A combination of hidden Markov model and fuzzy model for stock market forecasting

This paper presents a novel combination of the hidden Markov model (HMM) and the fuzzy models for forecasting stock market data. In a previous study we used an HMM to identify similar data patterns from the historical data and then used a weighted average to generate a ‘one-day-ahead’ forecast. This paper uses a similar approach to identify data patterns by using the HMM and then uses fuzzy logic to obtain a forecast value. The HMM's log-likelihood for each of the input data vectors is used to partition the dataspace. Each of the divided dataspaces is then used to generate a fuzzy rule. The fuzzy model developed from this approach is tested on stock market data drawn from different sectors. Experimental results clearly show an improved forecasting accuracy compared to other forecasting models such as, ARIMA, artificial neural network (ANN) and another HMM-based forecasting model.     

云环境下的高效K-Medoids并行算法

传统聚类算法K-Medoids对初始点的选择具有随机性,容易产生局部最优解;替换聚类中心时采用的全局顺序替换策略降低了算法的执行效率;同时难以适应海量数据的运算。针对上述问题,提出了一种云环境下的改进K-Medoids算法,该改进算法结合密度法和最大最小原则得到优化的聚类中心,并在Canopy区域内对中心点进行替换,再采用优化的准则函数,最后利用顺序组合MapReduce编程模型的思想实现了算法的并行化扩展。实验结果表明,该改进算法与传统算法相比对初始中心的依赖降低,提高了聚类的准确性,减少了聚类的迭代次数,降低了聚类的时间。     

A load prediction model for cloud computing using PSO-based weighted wavelet support vector machine

In order to reduce the energy consumption in the cloud data center, it is necessary to make reasonable scheduling of resources in the cloud. The accurate prediction for cloud computing load can be very helpful for resource scheduling to minimize the energy consumption. In this paper, a cloud load prediction model based on weighted wavelet support vector machine(WWSVM) is proposed to predict the host load sequence in the cloud data center. The model combines the wavelet transform and support vector machine to combine the advantages of them, and assigns weight to the sample, which reflects the importance of different sample points and improves the accuracy of load prediction. In order to find the optimal combination of the parameters, we proposed a parameter optimization algorithm based on particle swarm optimization(PSO). Finally, based on the WWSVM model, a load prediction algorithm is proposed for cloud computing using PSO-based weighted support vector machine. The Google cloud computing data set is used to verify the algorithm proposed in this paper by experiments. The experiment results indicate that comparing with the wavelet support vector machine, autoregressive integrated moving average, adaptive network-based fuzzy inference system and tuned support vector regression, the proposed algorithm is superior to the other four prediction algorithms in prediction accuracy and efficiency.     

电网供电系统短期电力负荷预测优化仿真

对电网供电系统短期电力负荷预测模型进行优化,能提升预测结果的准确性和鲁棒性.虽然现有预测模型可以满足预测速度的要求,但预测结果的精确性和稳定性却无法保证.为了得到更加准确和稳定的预测结果,提出了细菌觅食算法优化极限学习机预测模型.首先在电力负荷样本数据中形成训练样本和预测样本集,利用细菌觅食优化算法对极限学习机预测模型中的不确定参数进行优化,然后利用改进后的模型进行电力负荷预测.新模型的优化仿真结果显示,利用细菌觅食算法优化极限学习机预测模型的预测精度和稳定性均优于传统预测模型的预测结果,该算法具有很好地实用性.     

基于LSTM网络的移动云计算多元负载预测模型

针对移动云主机负载变化大、难以精准预测的问题,提出一种联合特征选择下基于长短期记忆网络的AR-LSTM-ED负载预测模型,能够对云主机负载进行单步和长时间多步预测.首先采用联合特征选择的方法得到与目标预测负载序列相关的其他负载序列,并且利用适用于在线预测的无抽取的小波变换方法将目标预测特征分解成更加易于预测的子序列.最...     

基于隐马尔可夫模型的非侵入式负荷监测泛化性能改进

隐马尔可夫模型(HMM)是非侵入式负荷监测常用的算法.由于电压波动与负荷自身电气特性变化等原因,负荷的测量状态如功率可能持续变化,运行过程中出现新的状态转移,但当前基于HMM的非侵入式负荷监测方法并未考虑如何处理该情况,缺乏状态辨识与功率分解的泛化能力.针对这一问题,本文提出并构建二元参数隐马尔科夫模型(BPHMM),结合DBSCAN聚类算法,基于有功功率和稳态电流对负荷状态进行聚类,降低了因电压波动和噪声数据对负荷状态聚类结果造成干扰的可能性;改进维特比算法使其考虑到HMM模型参数更新以实现对负荷状态预测泛化性能的改进;考虑到功率的随机波动性,基于极大似然估计原理构建功率计算优化模型并实现负荷的功率分解.本文采用公共数据集AMPds2对所述方法进行验证,测试算例验证了所述方法的有效性.     

面向应用性能管理系统的运行负载预测

在应用性能管理系统中,系统未来的负载情况对运维调度有重要的指导意义。在云计算环境下,弹性伸缩计算能力为调整系统规模提供了可能,根据系统将来的负载情况可以提前做出相应的调整：可以在负载加重前扩展好集群,保证服务质量;在负载降低之后若预测一定时间内没有负载加重的情况,则可以及时缩减集群规模,降低企业运营成本。在金融领域,ARIMA模型是常用的时序预测模型,但其应用需要人工介入分析时序的平稳性,调参过程过于复杂。近年来神经网络技术的发展带动了人工更智能技术的发展,本论文设计并测试了ANN、RNN、GRU、LSTM等神经网络的负载预测的效果。实验结果表明LSTM网络预测精准且表现稳定,是系统负载预测的理想模型。     

基于密度聚类的HMM协作频谱预测算法

针对传统隐马尔科夫频谱预测中的时延长、预测准确度低的问题,提出了一种基于密度聚类(Density-Based Spatial Clustering of Applications with Noise,DBSCAN)的HMM协作频谱预测算法。该算法采用DBSCAN算法将具有强相关性的频域信道聚为一簇,并以簇为单位对信道状态进行预测,通过减少预测次数来降低频谱预测时延;同时在时域利用多个次级用户协作预测的方法,通过融合各次级用户的初始预测结果来降低预测的不确定度。仿真实验表明,相比于传统的隐马尔科夫频谱预测算法,所提算法的频谱预测时延更短,准确度更高。     

Resource scheduling algorithm with load balancing for cloud service provisioning

Cloud computing uses scheduling and load balancing for virtualized file sharing in cloud infrastructure. These two have to be performed in an optimized manner in cloud computing environment to achieve optimal file sharing. Recently, Scalable traffic management has been developed in cloud data centers for traffic load balancing and quality of service provisioning. However, latency reducing during multidimensional resource allocation still remains a challenge. Hence, there necessitates efficient resource scheduling for ensuring load optimization in cloud. The objective of this work is to introduce an integrated resource scheduling and load balancing algorithm for efficient cloud service provisioning. The method constructs a Fuzzy-based Multidimensional Resource Scheduling model to obtain resource scheduling efficiency in cloud infrastructure. Increasing utilization of Virtual Machines through effective and fair load balancing is then achieved by dynamically selecting a request from a class using Multidimensional Queuing Load Optimization algorithm. A load balancing algorithm is then implemented to avoid underutilization and overutilization of resources, improving latency time for each class of request. Simulations were conducted to evaluate the effectiveness using Cloudsim simulator in cloud data centers and results shows that the proposed method achieves better performance in terms of average success rate, resource scheduling efficiency and response time. Simulation analysis shows that the method improves the resource scheduling efficiency by 7% and also reduces the response time by 35.5 % when compared to the state-of-the-art works.     

一种中间件数据库集群的负载均衡算法

为满足用户对数据库集群系统高输入高输出应用的需求,设计一种采用中间件技术的数据库集群系统,并针对该系统提出一种基于Markov模型的数据库集群负载均衡算法。该算法在执行节点负载信息采样周期内,利用Markov模型预测集群系统各执行节点的负载信息状态,根据预测的执行节点负载信息对集群系统进行负载均衡。实验结果表明,该算法能够有效提高数据库集群的性能。     

面向智慧医疗云的SDN动态负载均衡方法

文中 引入软件定义网络(Software Defined Network,SDN)对智慧医疗云进行网络管理,并且针对传统SDN控制器存在单点失效和负载均衡的问题,设计了智慧医疗分布式SDN控制器系统。SDN控制系统分为SDN控制器集群、数据转发平面和智慧医疗云服务系统3层。在此基础上,提出一种实时负载动态自调节的快速负载均衡算法DAF(Dynamic Adaptive and Fast Load Balancing)。在该算法中,负载信息感知组件周期性地采集自己的负载信息,自动地进行控制器间的负载信息交互;控制器的负载值超过阈值时,会触发交换机迁移动作,以动态配置交换机与控制器之间的映射关系。实验结果表明,面向智慧医疗云的分布式SDN控制系统的性能良好,且DAF算法能够快速地实现SDN控制器间的负载均衡,提升了智慧医疗云的网络吞吐量。     

基于GRU-LSTM组合模型的云计算资源负载预测研究

日益增多的应用部署在云端使得云数据中心的功耗波动剧烈,从而导致云数据中心资源利用率不平衡,高效的负载预测是解决该问题的关键技术。针对目前负载预测模型预测精度低、预测时间长的问题,建立一种基于门控循环单元（GRU）与长短期记忆（LSTM）网络的组合预测模型GRU-LSTM。该模型的网络结构包括3层,第一层采用GRU,利用GRU参数少、易收敛的特点减少模型训练时间,第二、第三层采用LSTM,结合LSTM参数多的优势提高模型的预测精度。在此基础上,对数据集作缺失值处理和标准化处理,使用随机森林算法对原始序列进行特征选择后得到一组新的序列值,将该序列值作为GRU-LSTM组合预测模型的输入,以对云计算资源进行高效预测。在集群公开数据集Cluster-trace-v2018上进行实验,结果表明,与传统的单一预测模型ARIMA、LSTM、GRU以及现有的组合预测模型ARIMA-LSTM、Refined LSTM等相比,GRU-LSTM模型预测结果的均方误差减少6~9,预测时间平均缩短约10%。     

Priori information and sliding window based prediction algorithm for energy-efficient storage systems in cloud

One of the major challenges in cloud computing and data centers is the energy conservation and emission reduction. Accurate prediction algorithms are essential for building energy efficient storage systems in cloud computing. In this paper, we first propose a Three-State Disk Model (3SDM), which can describe the service quality and energy consumption states of a storage system accurately. Based on this model, we develop a method for achieving energy conservation without losing quality by skewing the workload among the disks to transmit the disk states of a storage system. The efficiency of this method is highly dependent on the accuracy of the information predicting the blocks to be accessed and the blocks not be accessed in the near future. We develop a priori information and sliding window based prediction (PISWP) algorithm by taking advantage of the priori information about human behavior and selecting suitable size of sliding window. The PISWP method targets at streaming media applications, but we also check its efficiency on other two applications, news in webpage and new tool released. Disksim, an established storage system simulator, is applied in our experiments to verify the effect of our method for various users’ traces. The results show that this prediction method can bring a high degree energy saving for storage systems in cloud computing environment.     

云计算环境中基于访问量和依赖性评价的数据分配算法

大量的大规模密集型数据需要存储在多个数据存储中心,而应用越来越广泛的云计算环境很好地解决了大规模密集型数据在分配中遇到的规模性问题。但是,云计算环境中多数据存储中心的数据分配会带来数据存储中心之间数据量的传输,从而导致数据访问效率低下。同时,单位时间上数据访问量的不平衡性会引起数据存储中心的访问瓶颈。以大规模密集型数据中的数据流为建模对象,提出了一种数据分配算法,它在保证数据存储中心负载平衡的基础上兼顾了密集型数据之间的依赖性。实验表明,相比于同类的数据分配算法,所提算法具有更好的综合表现,特别是在保证数据存储中心的负载平衡方面,效果突出。