A Hybrid Method for Short-Term Host Utilization Prediction in Cloud Computing

,


Introduction
Cloud computing assembles a large number of computing, storage, and network resources into a data center.ese resources are cut and allocated efficiently to satisfy users' resource demands through virtualization technology.In addition to rich resources, cloud computing also provides a pay-as-you-go model.Users can rent various resources as they demand, which reduces their costs.ese characteristics of rich resources, on-demand resource provision, and low costs prompt cloud computing to be widely applied in various domains.However, it is still a challenge to allocate and schedule resources effectively to improve resource utilization and guarantee QoS. e general process of resource allocation and scheduling in cloud computing is shown in Figure 1.When a new virtual machine (VM) request is initiated, the cloud data center selects a suitable physical host to allocate resources for this VM according to a specified resource allocation policy.
is policy can maximize resource utilization per host to minimize the number of active hosts or balancing resource utilization of all active hosts.Whichever policy you use, it is important to know the future host utilization for the selection of a suitable host.Additionally, VM migration is also an effective method for resource scheduling.When the host utilization exceeds a predefined threshold, the performance of VMs running on this host will decrease.It will not guarantee the QoS of applications running on these VMs.erefore, it is necessary to migrate some hotspot VMs from one overloaded host to other nonoverloaded hosts.Similarly, if the host utilization is below a predefined threshold, all VMs on this host will be migrated to other hosts.us, this host can be closed to reduce energy consumption.
VM migration is a reactive method that cannot be initiated until the host is overloaded or underloaded.erefore, it is very important to detect when the host is overloaded or underloaded.Most existing approaches monitor host utilization to determine its state.If its resource utilization exceeds a predefined threshold during an observation period, this host is overloaded.If its resource utilization is always below a predefined threshold during an observation period, it is declared underloaded.Basically, it usually takes some time to migrate VMs from an overloaded host to other hosts.If the host utilization changes faster than the provision time of the resources, users will suffer poor QoS until resources are available.In addition, host underload detection based on a single host utilization value also leads to unnecessary VM migration and stability problems.
ese problems can be addressed via proactive methods that actively predict short-term host utilization to allocate resources in advance.For example, if the host utilization within the future 15 minutes is always over 80%, this host will be overloaded.erefore, the VMs should be migrated in advance from this host to other hosts to ensure QoS.If the host utilization within the current and future 1 hour is always below 15%, this host is underloaded and should be closed to save energy after VM migration.However, a large number of random resource demands and concurrent access to applications cause stochastic volatility of host utilization.
ey change very fast and exhibit strong instability with many bursts.It is difficult to predict short-term host utilization in a timely and accurate manner based on such data.
Although some machine learning methods, such as a neural network (NN) [1,2], support vector regression (SVR) [3], and backpropagation neural networks (BPNN) [4], achieve good prediction accuracy in cloud computing, they require too much time to train a model to allocate resources rapidly.Line regression (LR) can implement prediction more quickly than ARIMA, but it demands that the training data have simpler behaviors.ARIMA is a prediction model for nonstationary time series, but it is not suitable if a large amount of random variation exists in the data.In our previous work [5], we proposed a resource demand prediction method EEMD-ARIMA that combines the EEMD method and ARIMA model to predict future resource demands. is method first uses the EEMD method to decompose the original resource demand sequence into multiple IMF components and the residual (R) component.Next, we forecast the future values of each component by the ARIMA model.Finally, the overall forecasting results are obtained by superposing the forecasting results of each component.Although this method alleviates random variation of resource demands and improves prediction accuracy by combining EEMD and ARIMA methods, two problems arise.One is the prediction error accumulation caused by the superposition of ARIMA prediction of all components.
e ARIMA prediction of each component decomposed by the EEMD method generates a certain prediction error.e superposition of the prediction results of all components leads to the prediction error accumulation.Another is the high time cost due to EEMD decomposition and ARIMA prediction of too many decomposition components.e ARIMA prediction of each component takes some time.
us, the total time of the ARIMA prediction of multiple components greatly increases compared with a single ARIMA prediction of the original sequence.
To solve these problems of the EEMD-ARIMA method, this paper further proposes a hybrid method, EEMD-RT-ARIMA, for short-term host utilization prediction that not only further improves prediction accuracy by combining the EEMD method with the ARIMA model but also reduces time cost by selecting and reconstructing efficient IMF components.
e comparison and evaluation are made among our EEMD-RT-ARIMA method, ARIMA model, and EEMD-ARIMA method in terms of error, effectiveness, and time-cost analysis.

Related Works
Many studies have been conducted on various predictions in cloud computing.From the perspective of research objectives, some researchers have studied server load prediction 2 Journal of Electrical and Computer Engineering [6][7][8][9][10], VM load prediction [11,12], VM utilization prediction [13,14], host utilization prediction [15], web application workload prediction [16], cloud service workload prediction [17][18][19], workflow workload prediction [20], service quality prediction [21], and workload characterization [22][23][24].Toumi et al. [6] described a server load according to the submitted task types and the submission rate and applied a stream mining technique to predict server loads.Jheng et al. [11] proposed a VM workload prediction method based on the gray forecasting model, which determines the migrated VMs according to power savings and workload balance.Dabbagh et al. [13] proposed a prediction approach that uses Wiener filters to predict the future resource utilization of VMs.Mason et al. [15] predicted host CPU utilization for a short time using evolutionary neural networks, which showed a high prediction accuracy and a high degree of generality.In this paper, we focus on host utilization prediction using EEMD and ARIMA methods to not only improve prediction accuracy but also reduce prediction time as much as possible.
From the perspective of approaches, prediction methods are usually divided into two categories.One is based on machine learning methods.Tseng et al. [25] proposed a prediction method for CPU and memory utilization of VMs and physical machines based on a genetic algorithm (GA), which precedes the gray model under stable tendency and unstable tendency in terms of prediction accuracy.Shyam and Manvi [26] proposed a shortand long-term prediction model of virtual resource requirements for CPU/memory-intensive applications based on Bayesian networks, where the relationships and dependencies between variables are identified to facilitate resource prediction.Lu et al. [27] proposed a workload prediction model RVLBPNN (Rand Variable Learning Rate Backpropagation Neural Network) based on BPNN algorithm, which achieves higher prediction accuracy than the hidden Markov model and the naive Bayes classifier. is method not only predicts CPU-intensive and memoryintensive workloads but also improves prediction accuracy by using the intrinsic relations among the arriving cloud workloads.Rajaram and Malarvizhi [28] compared the prediction accuracies of a few machine learning methods, such as LR, SVR, and multiplayer perceptron.Li and Zhang [29] proposed an optimal combination prediction method for resource demands, which combines the induced ordered weighted geometry averaging operator and the generalized dice coefficient with the improved Elman neural network and gray model to enhance the prediction accuracy.Minarolli and Freisleben [30] presented a cross-correlation prediction approach based on support vector machine (SVM), which considers the cross relation of VMs running the same application to improve prediction accuracy.Zhang et al. [31] proposed a deep belief network-(DBN-) based prediction approach of cloud resource requests in which orthogonal experimental design and analysis of variance are used to enhance the prediction accuracy.Compared with the ARIMA model, this method greatly reduces mean square error (MSE) by over 60% for CPU and RAM request predictions.Although machine learning methods are effective in improving prediction accuracy, they are complex and usually demand a large number of data to extract features and train a model.It requires too much time for the prediction to guarantee QoS of the running applications.Cloud computing requires a simple and rapid host utilization prediction method to support resource allocation and scheduling.
Another method is based on statistical methods, such as Brown's quadratic exponential smoothing method [32], autoregressive integrated moving average (ARIMA) model [33][34][35], and the kernel canonical correlation analysis [36].Tran et al. [37] applied the ARIMA model in the long-term prediction of server workload, while our method aims to predict short-term host utilization.It is more difficult because host utilization can be extremely random and nonstationary in a short time.Calheiros et al. [33] proposed a short-term prediction model of cloud workload using the ARIMA model and evaluated the prediction accuracy and its impact on user applications' QoS.ey suggested that users' behaviors must be considered to reflect real conditions in workload simulation.Our method combines the ARIMA model with EEMD and RT methods to improve prediction accuracy and reduce prediction time as much as possible.It is compared with EEMD-ARIMA and ARIMA methods in terms of error, effectiveness, and time-cost analysis.
Moreover, some studies combine the ARIMA model with other techniques to improve prediction accuracy.Xu et al. [38] constructed a model GFSS-ANFIS/SARIMA combining the seasonal ARIMA model with the generalized fuzzy soft sets and adaptive neuro-fuzzy inference system. is model improves the prediction accuracy of resource demands.Li et al. [39] proposed a workload predictor combined with ARIMA and dynamic error compensation to reduce the service-level agreement (SLA) default rate.Fu and Zhou [40] proposed a predicted affinity model to implement VM placement, which uses the resource demands predicted by the ARIMA model to calculate a VM-host affinity value.Jiang et al. [41] presented a self-adaptive ensemble prediction method for cloud resource demands, which uses a two-level ensemble method to predict VM demands based on a historic time series.is method not only combines multiple prediction methods: moving average (MA), autoregressive (AR), artificial neural network (ANN), gene expression programming (GEP), and SVM but also adjusts the weight of each method adaptively to obtain the best average performance according to the relative errors.In contrast, our method uses the EEMD method to deal with the nonstationary host utilization and then selects and reconstructs efficient components to improve prediction accuracy and reduce the time cost.e EEMD proposed by Wu and Huang [42] is an effective noise-aided method that can handle nonlinear and nonstationary time series.It has been widely used in wind speed forecasting [43,44], aircraft auxiliary power unit (APU) degradation prediction [45], turbine fault trend prediction [46], and rolling bearing fault diagnosis [47].It has shown a good effect on enhancing the prediction accuracy.Our method also uses EEMD to decompose the nonstationary host Journal of Electrical and Computer Engineering utilization for improving the prediction accuracy and further uses correlation coefficients, RT values, and average periods to select and reconstruct efficient components for reducing prediction error accumulation and prediction time.

Empirical Mode Decomposition (EMD).
EMD is a method of signal processing that can decompose a signal into multiple IMFs and an R trend item [48].Two conditions must be satisfied for an IMF: (1) e number of extrema and zero-crossings must either be identical or differ by at most one (2) e mean value of the envelopes of the local maxima and the local minima must be zero EMD includes the following steps: Step 1. Make f(t) � x(t), where x(t) is given as the original data.
Step 2. Find all the local maxima and minima of f i (t), where i is the loop times and its initial value is 1.
Interpolate between the local maxima and minima to obtain an upper envelope and a lower envelope and then compute the mean value m i (t) of these envelopes.
Step 3. Compute the new component Step 4. Verify whether h i (t) satisfies the abovementioned two conditions for an IMF.If it does not, make f i+1 (t) � h i (t) and repeat steps 2 and 3.If it satisfies the condition, h i (t) is regarded as the first IMF component p 1 (t), where p 1 (t) � h i (t).en, compute the R component by the formula r 1 (t) � x(t) − p 1 (t).
Step 5. Repeat step 1-4 with r 1 (t) as the new data until the R is a monotonic function.us, x(t) is decomposed into n IMFs and an R as follows: (1)

Ensemble Empirical Mode Decomposition (EEMD).
e EMD method has a noticeable drawback of mode mixing that can cause signal intermittency.Wu and Huang proposed a new method named ensemble empirical mode decomposition (EEMD) to solve this problem.Compared with the EMD method, the EEMD method first executes the decomposition process k times.Each time, it adds a different white noise to the signal and then decomposes the new signal.Generally, the k iterations are set as an integer in the range [50, 100], and the standard deviation d of the white noise is set as a value in the range [0.1, 0.2].Next, k groups of decomposition results are obtained.Each group includes n IMFs p mi (t)(i � 1, . . ., n) and an R r m (t), where m denotes the group number.Finally, the mean values of these groups of IMFs and Rs are calculated as the final IMFs p i (t)(i � 1, . . ., n) and the R r(t): (2) e IMF components have three main characteristics.
(1) Completeness: the total of all IMFs and the R have the same feature as the original data.(2) Orthogonality: each IMF with a certain physical meaning is independent and has no effect on other IMFs.e product of any two IMFs equals 0 in mathematics.
(3) Adaptability: an IMF with a higher frequency is decomposed from the original data faster than those with low frequencies.e frequencies of IMFs reflect the features of the original data.

Runs Test (RT).
RT is a nonparametric test method that checks the randomness of a sequence with only two symbols or two values, such as + and − and 0 and 1.An RT is defined as a sequence with successive symbols (0 or 1).For example, a data sequence "11110000011111000110010" includes 8 runs, 4 of which involve successive "1" and the others involve successive "0." RT can also be used to test a time series.Assume that M � < m 1 (t), . . ., m i (t), . . ., m n (t) > denotes a time series, where m i (t) is an element of this time series and n is the total number of elements.e mean value of these elements is calculated by the following formula: en, the element of this time series can be denoted as follows: us, this time series is transformed into a sequence with a series of 0 and 1, in which the elements are independent and identically distributed.e total number of RT reflects the fluctuation of the sequence.

A Hybrid Method for Short-Term Host Utilization Prediction
To improve prediction accuracy and reduce prediction time of the EEMD-ARIMA method, we propose a hybrid method, EEMD-RT-ARIMA, for short-term host utilization prediction as shown in Figure 2. First, the host utilization sequence is decomposed into multiple IMF components and the R component using the EEMD method.Next, we calculate the correlation coefficients between IMF components and the original data sequence to select the efficient IMF components.en, we use RT values and average periods to reconstruct these efficient IMF and R components into three new components: high-frequency and strong-volatility component, medium-frequency and weakvolatility component, and low-frequency trend component.en, we use the ARIMA method to predict the results of three new components.Finally, the overall prediction results are achieved by summing the prediction results of the three new components.
e key to our EEMD-RT-ARIMA method is to select and reconstruct efficient components.Compared with the EEMD-ARIMA method, the number of its components involving in ARIMA prediction is reduced.us, the EEMD-RT-ARIMA method can reduce the prediction error accumulation and the total prediction time by reducing the number of components.Obviously, both the EEMD-RT-ARIMA method and EEMD-ARIMA method have a higher prediction time than the ARIMA model from their implementation processes.However, our EEMD-RT-ARIMA method focuses on cost-effectiveness, which has a tradeoff among prediction accuracy, effectiveness, and time cost.

Use of EEMD to Decompose the Host Utilization Sequence.
A host utilization sequence is classified into different categories according to the CPU, memory, and disk, such as CPU utilization sequence e CPU utilization sequence is usually random and unstable owing to random and sudden resource demands in cloud computing.It is necessary to transform such data into relatively stationary data to improve prediction accuracy.
e EEMD method appears to be more effective in processing nonlinear and nonstationary data sequences than other decomposition algorithms.erefore, we use the EEMD method to decompose the host utilization sequence and obtain a series of the IMF i components and the R component.
A running example shows the nonstationary CPU utilization trace of a physical host from our cloud platform.We divide it into the training set (673 data points) and the testing set (24 data points) in Figure 3. en, we use the EEMD method to decompose the training set and obtain IMF1-IMF8 components and the R component.ey are shown from the high frequency to low frequency in Figure 4.

Calculation of the Correlation Coefficients to Select Efficient IMF Components.
A correlation coefficient measures the correlation between two sequences.We calculate the correlation coefficient P j (X, Y) between the IMF j component and the original training set based on the following formula, where cov(X, Y) is a covariance between the sequences X and Y and Var(X) and Var(Y) are the variances of the sequence X and the sequence Y: en, the correlation coefficient P j (X, Y) is checked to determine whether it is negative.If it is negative, the IMF j component is inefficient and dropped.If it is not negative, the IMF j component is efficient and reserved.
We calculate the correlation coefficient between each IMF component and the original training set.Only IMF6 and IMF7 have negative correlation coefficients of −0.08 and −0.15.Hence, they are dropped.IMF1-IMF5 and IMF8 are selected as efficient IMF components.Journal of Electrical and Computer Engineering features.us, the prediction error accumulation and the prediction time of the EEMD-ARIMA method can be reduced by reducing the number of components.e average period reflects the frequency of host utilization variation.ere exists a reciprocal relation between them.e smaller the average period, the higher the frequency.If the average periods of IMF components are closer, they are closer in frequency.e average period is calculated by the following formula, in which n is the number of the training set and l j is the number of extrema:

Reconstruction of Efficient IMFs and R into New
Similarly, the RT value reflects the trend of amplitude fluctuation.If the RT value is larger, the amplitude volatility is stronger.If the RT values of the two IMFs are closer, the overall trend of the two IMFs is similar in amplitude volatility.
To enhance the prediction accuracy and reduce the prediction time of the EEMD-ARIMA method, we reconstruct the IMF components and the R component into three new components according to their average periods and RT values in the EEMD-RT-ARIMA method.Because the average period and RT value have different units, we normalize the average period T j as follows: where T nj denotes the normalized average period of the IMF j component.T max and T min represent the maximum and minimum of the average periods of all IMF components.Similarly, the RT value R j can be normalized as follows: where R nj is the normalized value of R j .R max and R min are the maximum and minimum of all RT values.us, the reconstruction factor (RF) is defined as follows: An IMF component is higher in frequency and stronger in volatility, and its RF value is greater.If the RF values of the two IMF components are closer, their overall trends are more similar.us, they can be reconstructed into a new component.All efficient IMF and R components are reconstructed into three new components: high-frequency and strong-volatility component, medium-frequency and weak-volatility component, and low-frequency trend component.
e high-frequency and strong-volatility component reflects the strong volatility and randomness of the high-frequency part of the original host utilization sequence.
e medium-frequency and weak-volatility component shows the detailed features of the volatility of the original host utilization sequence.
e low-frequency trend component depicts the overall trend of the volatility of the original host utilization sequence.
Table 1 shows the RT values, average periods, and RF values of efficient IMF and R components.e RF values of IMF1 and IMF2 are large and relatively close, while the RF values of IMF8 and R are equal to 0.
e RF values of IMF3-IMF5 are close.erefore, we reconstructed IMF1-IMF2, IMF3-IMF5, and IMF8-R into three new components, as shown in Figures 5(a)-5(c).ey separately reflect the randomness, the fluctuation details, and the overall trend of the original host utilization sequence.

Use of the ARIMA Model to Predict the Future Host
Utilization.We use the ARIMA model to predict the future results for each new component.en, the overall prediction results are obtained by superposing the prediction results of each new component.e ARIMA prediction is described as follows (Algorithm 1).
For example, we assume that three new components C h , C m , and C l are obtained, which represent the high-frequency and strong-volatility component, medium-frequency and weak-volatility component, and low-frequency trend component, respectively.en, we use the ARIMA method to predict the future 24-point values for each new component.
e prediction results P h , P m , and P l of three new components can be described in the following formulas, each of which includes the values of the predicting 24-point data: Finally, we calculate the overall prediction result P by superposing the prediction results of each new component as follows: From this process of the EEMD-RT-ARIMA method, we find that the number of components decreases from 9 to 3, which can reduce the total prediction time and the error accumulation of the component prediction compared with the EEMD-ARIMA method.6 Journal of Electrical and Computer Engineering

Experimental Setup
We conduct an experiment to evaluate our method.e experimental dataset and measurement metrics are introduced as follows.

Experimental Dataset.
Host utilization mainly involves in CPU utilization, memory utilization, network utilization, and disk utilization.In this paper, we mainly focus on host CPU utilization.We randomly select CPU utilization traces of 7 physical hosts from the dataset released by Alibaba in August 2017 [49], each of which includes 144 points (5 minutes per point).ese traces are all time-dependent sequences as shown in Figures 6(a)-6(g).
Each sequence is divided into a training set and a testing set.We first use the training set to predict the future data, and then, these predicting data are compared with those actual data in the testing set to evaluate our method.In this paper, each training set is set as the first 120 points, and the testing set is defined as the subsequent points, such as 6 points, 12 points, and 24 points.We set the number of iterations k � 50 and the standard deviation d � 0.2 in EEMD decomposition.

Measurement Metrics.
We evaluate our method in terms of error, effectiveness, and time-cost analysis as follows.

Error Analysis.
To evaluate our method, we use the mean absolute percentage error (MAPE) to reflect the prediction accuracy.MAPE is defined as follows: where x f i denotes the value of the prediction point, x t i denotes the actual value in the testing set, and m denotes the  number of prediction points.It is obvious that the prediction accuracy is higher when the MAPE is lower.

Effectiveness Analysis.
Host utilization underprediction or overprediction can lead to resource underprovision or overprovision.Resource underprovision cannot guarantee applications' QoS, while resource overprovision can cause resource waste and low resource utilization.erefore, a good prediction method should avoid underprediction and overprediction.In particular, underprediction should be avoided as much as possible because it results in a lower QoS to users.We set up the positive and negative errors to reflect the overprediction and underprediction and then use them to evaluate the effectiveness of our method.A good prediction method should have a smaller negative error to avoid underprediction.e positive and negative prediction errors are calculated by the following formula, where p i is the predicting data, r i is the actual data and m is the number of underprediction data (i.e., negative deviation) or overprediction data (i.e., positive deviation):

Experimental Results and Analysis
To validate the prediction effectiveness of our EEMD-RT-ARIMA method, we conduct experiments on ARIMA, EEMD-ARIMA, and EEMD-RT-ARIMA methods and compare their predictive results.All experiments were performed on a PC with 2.5 GHz Intel (R) i7 CPU running MATLAB.To make three methods comparable, we use the same original dataset to execute it 5 times for each method.e mean values of the prediction results are shown in the following tables and figures. 2 shows the MAPE values of host utilization predictions for 7 physical hosts.We can see that EEMD-ARIMA and EEMD-RT-ARIMA methods have lower MAPE values than ARIMA models for 6-point and 12point predictions.For example, EEMD-ARIMA and EEMD-RT-ARIMA methods achieve MAPE values of 6.06% and 5.05% for the 6-point prediction of host 109, while the ARIMA model has a far higher MAPE value (up to 16.85%).

Error Analysis. Table
ey obtain MAPE values of 10.13% and 5.46% for the 12point prediction of host 109, while the ARIMA model obtains 11.08%.For host 22, both the EEMD-ARIMA and EEMD-RT-ARIMA methods obtain far lower MAPE values of 5.31% and 5.42% than the 10.66% of the ARIMA model for 6-point prediction.Similarly, they also obtain better effectiveness on 12-point prediction.e same situation also exists in 6-point and 12-point predictions of other hosts.
is indicates that both EEMD-ARIMA and EEMD-RT-ARIMA methods have higher prediction accuracy than ARIMA models in 6-point and 12-point predictions for host utilization.EEMD reduces the inherent volatility of the host utilization sequence, which improves the prediction accuracy of the EEMD-ARIMA and EEMD-RT-ARIMA methods.However, the situation changes in 24-point prediction.e MAPE values of hosts 1162, 424, 1060, and 237 are all over 30% using these three methods.Although the EEMD-RT-ARIMA method has lower MAPE values than the ARIMA and EEMD-ARIMA methods for hosts 839 and 109, it has far higher MAPE values in the 24-point prediction than those of 6-point and 12-point predictions.is shows that the EEMD-ARIMA and EEMD-RT-ARIMA methods are not suitable for long-term but suitable for short-term prediction.
For further analysis, we find that the EEMD-RT-ARIMA method achieves lower prediction error than the EEMD-ARIMA method for the 6-point and 12-point predictions of hosts 839, 109, and 1162, although the EEMD-RT-ARIMA method only selects efficient IMF components.However, it is the opposite for hosts 22, 424, 1060, and 237.e original CPU utilization sequences of all physical hosts are identical in frequency, so we calculate the RT value of each CPU utilization sequence shown in Table 3. Hosts 839, 109, and 1162 achieve lower RT values under 10, which shows that their CPU utilization is more stationary than other hosts.Smaller RT values indicate more stationary host utilization sequences.
is phenomenon can also be seen in Figures 6(a)-6(c).From Tables 2 and 3, it can be found that the EEMD-RT-ARIMA method achieves a lower MAPE value than the EEMD-ARIMA method if the RT value is smaller.Conversely, the EEMD-RT-ARIMA method has a higher MAPE value than the EEMD-ARIMA method if the (1) For each new component (2) Set the order of difference d � 0 (3) Execute the augmented Dickey-Fuller (ADF) test.If it is a stationary time series, go to step 5; else, go to step 4 until it is stationary (4) Difference the time series and set d � d + 1, go to step 3 (5) Determine the order of the ARIMA model using Bayesian information criterion (BIC) (6) Estimate the parameters of the ARIMA model using the maximum likelihood (7) Forecast the future n values of this new component using the ARIMA model ( 8) End (9) Obtain the overall prediction results by superposing the prediction results of each new component ALGORITHM 1: ARIMA prediction.RT value is larger.For instance, hosts 839, 109, and 1162 with smaller RT values obtain lower MAPE values using the EEMD-RT-ARIMA method than the EEMD-ARIMA method, while hosts 22, 424, 1060, and 237 with larger RT values obtain higher MAPE values using the EEMD-RT-ARIMA method than the EEMD-ARIMA method.Furthermore, the difference in the MAPE values between EEMD-RT-ARIMA and EEMD-ARIMA methods decreases with the increase in the RT values from host 839 to host 1162.
eir difference changes to negative from host 22, which indicates that the EEMD-ARIMA method has higher prediction accuracy than the EEMD-RT-ARIMA method.
en, their difference becomes larger as the RT values increase.e 12-point host CPU utilization prediction illustrates this situation.For example, host 839, with an RT value of 2, has a MAPE of 5.03% for 12-point prediction by using the EEMD-RT-ARIMA method, which is 5.45% lower than the 10.48% of the EEMD-ARIMA method.e MAPE value of the EEMD-RT-ARIMA method is only 4.19% lower than that of the EEMD-ARIMA method for host 1162, with an RT value of 10.For host 22 with an RT value of 16, the EEMD-RT-ARIMA method has a slightly higher MAPE of 5.37% than 5.33% of the EEMD-ARIMA method.With the increase of the RT value, the differences of MAPE values between EEMD-RT-ARIMA and EEMD-ARIMA methods further increase to 0.87%, 1.96%, and 4.63% for hosts 424, 1060, and 237, respectively.is indicates that the EEMD-RT-ARIMA method is less effective than the EEMD-ARIMA method in CPU utilization prediction for these hosts.Undoubtedly, the ARIMA prediction of each component decomposed by the EEMD method generates a certain error.e superposition of the prediction results of each component causes error accumulation.
e EEMD-RT-ARIMA method reduces the error accumulation by selecting and reconstructing the efficient IMF components into fewer components, so it achieves better prediction accuracy than the EEMD-ARIMA method for hosts 839, 109, and 1162.Certainly, the selection and reconstruction of efficient IMF components also cause a certain prediction error due to the absence of nonefficient components, especially for nonstationary host utilization sequences.When this kind of prediction error exceeds the error accumulation of ARIMA prediction of all components in the EEMD-ARIMA method, the EEMD-RT-ARIMA method is no more effective than the EEMD-ARIMA method for the nonstationary CPU utilization prediction of some hosts, such as hosts 22, 424, 1060, and 237.

Effectiveness Analysis.
To verify the effectiveness of our method in short-term prediction, we select the experimental results of hosts 839, 22, and 237 with the minimum, middle, and maximum RT values for further analysis.Figure 7 shows the prediction results of the EEMD-RT-ARIMA, ARIMA, and EEMD-ARIMA methods.We find that the future resource utilization of host 839 decreases below 11%.According to a predefined policy, host 839 is underloaded and can be closed to save energy.Figure 7 shows that our method is more accurate and effective than the ARIMA model.In particular, our method tends to change with the trend of data variation, while the ARIMA model cannot keep up with it.Our method is more suitable for handling nonstationary time series than the ARIMA model.Additionally, the predicting data using the EEMD-RT-ARIMA method are closer to the testing data than those of the EEMD-ARIMA method for host 839.ese 10 Journal of Electrical and Computer Engineering results show that the EEMD-RT-ARIMA method is more effective than the EEMD-ARIMA method for CPU utilization sequences with weak fluctuations.When the host utilization sequence shows stronger fluctuation, the absence of nonefficient IMF components will greatly influence the prediction results.e EEMD-RT-ARIMA method is no more effective than the EEMD-ARIMA method for CPU utilization prediction of host 237.
To further analyze the effectiveness of our method, we calculate the positive and negative errors for 6-point and 12point predictions of these hosts shown in Table 4.When the negative error is smaller, the prediction method is more suitable for cloud resource provision because of avoiding underprediction.It can be observed that most of the prediction results of the ARIMA model are underpredicted (the cells of positive error are all "null" for hosts 839 and 22).Furthermore, the negative errors of the ARIMA model are all far higher than those of other methods for host 237.For instance, the ARIMA model has a high negative error of up to 27.51% for the 12-point prediction of host 237, while the EEMD-ARIMA and EEMD-RT-ARIMA methods only have negative errors of 8.00% and 8.92%, respectively.If the ARIMA model is used to predict future host utilization, it can cause resource underprovision, which cannot ensure applications' QoS. e EEMD-RT-ARIMA method achieves smaller negative errors than the EEMD-ARIMA method for hosts 839 and 22, while it has a larger negative error for hosts 237.For instance, the EEMD-ARIMA method achieves the negative error of 10.71% for the 12-point prediction of host 839, while EEMD-RT-ARIMA only achieves the negative error of 4.74%.Similarly, the EEMD-ARIMA method obtains a negative error of 6.09% for the 12-point prediction of host 22, while the EEMD-RT-ARIMA method achieves a lower value of only 5.05%.However, the situation changes e EEMD-ARIMA method has a smaller negative error than the EEMD-RT-ARIMA method.For instance, the EEMD-RT-ARIMA method obtains negative errors of 11.37% and 8.92% for 6-point and 12-point predictions, while the EEMD-ARIMA method only has negative errors of 4.62% and 8.00%.

Time-Cost Analysis.
To verify the applicability of our method, we further compare the time cost of these methods in Figure 8.
e running time of the EEMD-ARIMA method is the largest by over 180 s while the ARIMA model takes the least time at less than 50 s.e EEMD-RT-ARIMA method time cost is between 70 s and 117 s, which decreases the time cost by 40%-80% compared with the EEMD-ARIMA method.For example, the running time of the EEMD-RT-ARIMA method is 69.37 s, far less than the 337.20 s of the EEMD-ARIMA method for the 6-point prediction of host 22.Our method saves up to 80% of the time cost.For the CPU utilization sequence of host 237 with strong variability, it requires 190.46 s to predict the future 6-point values using the EEMD-ARIMA method, while it only takes 113.64 s using the EEMD-RT-ARIMA method.
e running time is reduced by approximately 40%.Considering the prediction accuracy, effectiveness and time cost, our EEMD-RT-ARIMA method is more cost-effective for short-term host utilization prediction in cloud computing.

Conclusions
Host utilization is an indicator of host performance, whose prediction can promote effective resource scheduling in cloud computing.However, host utilization demonstrates strong randomness and instability caused by users' random and various resource demands.It is difficult to improve prediction accuracy.In this paper, we propose a hybrid and cost-effective method, EEMD-RT-ARIMA, for short-term host utilization prediction in cloud computing.e EEMD method is first used to decompose the nonstationary host utilization sequence into a few relatively stationary IMF components and an R component.
en, we calculate the correlation coefficient between each IMF component and the original data to select efficient IMF components and use RT values and average periods to reconstruct these components into three new components to reduce error accumulation and time cost.Finally, three new components are predicted by the ARIMA model, and their prediction results are superposed to form the overall prediction results.We the real host utilization traces from a cloud platform to conduct the experiments and compare our EEMD-RT-ARIMA method with the ARIMA model and EEMD-ARIMA method in terms of error, effectiveness, and time-cost analysis.
e results show that our method is cost-effective and is more suitable for short-term host utilization prediction in cloud computing.

Figure 1 :
Figure 1: Resource allocation and scheduling process.

Figure 3 :
Figure 3: CPU utilization trace of a physical host.

Figure 7 :
Figure 7: Prediction of the results of different methods.

Figure 8 :
Figure 8: Time cost of different prediction methods.

Table 1 :
RT, average periods, and RF of efficient and R components.

Table 2 :
MAPE values of host utilization prediction.

Table 3 :
RT values of each host utilization.

Table 4 :
Positive and negative error analysis.