• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Deep Learning Based Intrusion Detection in Cloud Services for Resilience Management

    2022-08-23 02:18:24SreenivasaChakravarthiJagadeeshKannanAnanthaNatarajanandXiaoZhiGao
    Computers Materials&Continua 2022年6期

    S.Sreenivasa Chakravarthi,R.Jagadeesh Kannan,V.Anantha Natarajan and Xiao-Zhi Gao

    1CSE,Center for Intelligent Computing,Sree Vidyanikethan Engineering College&School of Computer Science and Engineering,VIT,Chennai,Tamilnadu,600127,India

    2School of Computer Science and Engineering,VIT,Chennai,Tamilnadu,600127,India

    3CSE,Center for Intelligent Computing,Sree Vidyanikethan Engineering College,Tirupati,517102,India

    4School of Computing,University of Eastern Finland,Kuopio,02130,Espoo,Finland

    Abstract: In the global scenario one of the important goals for sustainable development in industrial field is innovate new technology, and invest in building infrastructure.All the developed and developing countries focus on building resilient infrastructure and promote sustainable developments by fostering innovation.At this juncture the cloud computing has become an important information and communication technologies model influencing sustainable development of the industries in the developing countries.As part of the innovations happening in the industrial sector, a new concept termed as ‘smart manufacturing’has emerged, which employs the benefits of emerging technologies like internet of things and cloud computing.Cloud services deliver an on-demand access to computing, storage, and infrastructural platforms for the industrial users through Internet.In the recent era of information technology the number of business and individual users of cloud services have been increased and larger volumes of data is being processed and stored in it.As a consequence,the data breaches in the cloud services are also increasing day by day.Due to various security vulnerabilities in the cloud architecture;as a result the cloud environment has become non-resilient.To restore the normal behavior of the cloud, detect the deviations, and achieve higher resilience, anomaly detection becomes essential.The deep learning architectures-based anomaly detection mechanisms uses various monitoring metrics characterize the normal behavior of cloud services and identify the abnormal events.This paper focuses on designing an intelligent deep learning based approach for detecting cloud anomalies in real time to make it more resilient.The deep learning models are trained using features extracted from the system level and network level performance metrics observed in the Transfer Control Protocol (TCP)traces of the simulation.The experimental results of the proposed approach demonstrate a superior performance in terms of higher detection rate and lower false alarm rate when compared to the Support Vector Machine(SVM).

    Keywords:Anomaly detection;network flow data;deep learning;migration;auto-encoder;support vector machine

    1 Introduction

    Cloud computing has been in the lime light for around two decades and it holds many features for improving business efficiencies, cost-benefiting, and advantages over conventional computing mechanisms.As per a recent survey by the international data group already 69%of the business firms are utilizing cloud services and 18%of the remaining has plans to implement cloud services at any of the point in their business operations in near future.At the same time a report from the Dell Inc.,says that the firms who have adapted to modern big data computing,cloud services and security are earning 53%faster revenue than their competing firms.These reports exhibits that the business firms and the leaders are reaping the benefits of the cloud services in their business operations.They use this modern state of art cutting edge technologies to efficiently implement their operations,provide better service to their customers,and in parallel achieve high profit margins.Gartner predicts an exponential growth of cloud services industry by 2022.The worldwide public cloud services market is projected to grow 17.5 percent in 2019 to total$214.3 billion,up from$182.4 billion in 2018,according to Gartner,Inc.The world economy greatly relies on the manufacturing industries for employment and wealth creation.The cloud computing has become the optimal solution for industry to implement their automation processes by adapting to machine-to-machine translation.Also for storing and managing the ever increasing production and other data, use of cloud services become essential.The various services offered to the end users of cloud in manufacturing sector are presented in Fig.1.

    Figure 1:Cloud services for end users of manufacturing sector

    The cloud computing environment faces a number of security challenges and most of them can be fixed up to a certain extent using current anomaly based Intrusion Detection Systems (IDS) [1].IDS in cloud networks have an important role to play in providing security against attacks from both insiders and outsiders.The IDS must be implemented as a part of the cloud services as it is scalable,and efficient in nature.The conventional IDS used to detect attack in internet environment don’t have ability to adapt to cloud environment and they are not scalable in nature[2].Also they are not deterministic in nature and found to be not suitable for cloud environment.Hence a novel and reliable anomaly based intrusion detection system has to be developed and evaluated[3].

    Most of the earlier approaches for anomaly detection in cloud environment utilized Machine learning techniques.These techniques have the ability to improve their performance over time by updating its knowledge on the pattern observed in the input data.Whenever a new pattern is observed in the input data the machine learning model parameters are updated to detect the similar anomalies in the future traffic flow [4].Based on the new information extracted from the previous results the performance of the techniques is improved by changing the execution strategy if required.The various types of machine learning algorithms used for anomaly detection are Bayesian network[5],Genetic algorithm,and neural network[6].Implementing an anomaly detection mechanism in real time with in the cloud environment involves multiple challenges with respect to performance and scalability of the cloud services.The cloud services are on-demand in nature and hence the anomaly detection must be performed in real-time monitoring.The real time implementation should be highly scalable and must provide support for multiple service providers.This paper focuses on developing an anomaly detection system which has the ability to detect the intrusion accurately,and maintain the resilience of the cloud network.The proposed IDS can be employed to secure the sensitive files,programs,and the ports of the virtual or physical machine in the cloud environment.The network based approach is more efficient when compared to the host based techniques which are unable to detect the attacks in the network and consumes majority of the host computational power and storage resources.

    Cloud faces different types of security issues and challenges which affects the growth of cloud services utilization rate.The overall aim is to develop a resilient cloud services for manufacturing sector by implementing an intelligent anomaly detection system as an integrated service within the cloud environment.The objectives of the proposed research work are to identify the service misuse at the client as anomalies and classify the network traffic behavior into two categories as either Normal or Abnormal by deep learning model using network flow data.The efficiency of the proposed method is analyzed based on a dataset created in the cloud simulation.The dataset is comprised of a vector of features extracted from the simulated cloud network with Virtual Machine (VM) migration.The simulation consists of a traffic generator which synthesis both normal and network-based attacks of different categories and intensities.The performance metrics used for studying the efficiency of the anomaly detection in identifying network level attacks are precision,recall,accuracy,and F1-score.For effective implementation of the anomaly detection in the cloud environment,it is designed as a service that can be offered along with infrastructure to the clients.The elasticity of the cloud under various simulated network level attacks and anomaly detection as a service can be evaluated by including VM migration.

    The proposed approach includes techniques for efficiently processing,analyzing,and evaluating real time data for detecting anomaly patterns.The challenge is to develop scalable, fault-tolerant anomaly detection method and embed it within a resilient framework for providing warnings promptly in case of adverse conditions during the VM migration.The proposed approach utilizes deep learning architectures to characterize the normal behavior in cloud environment and detect anomalies in the cloud network.The focus is extended to develop methods for capturing and analyzing real time network flow-data in the cloud environment.Further,optimal features are extracted from the captured data and,abnormal traffic patterns are discriminated from the normal traffic patterns using trained Auto Encoder.

    2 Literature Survey

    In general the architecture of IDS is complex in nature which includes a variety of concepts,and techniques that change with respect to the environment.The working principle of IDS relies on two main approaches for detecting anomalies and these approaches differ by the analysis method and processing techniques.The first approach utilizes the signature of the abnormal traffics and the second approach tracks for a deviation in the normal traffic.The signature based detection approach is considered to be better in terms of lower false alarm rate but they suffer due to inability in detecting newer type of attacks.But the anomaly based detection methods are able to detect an attack for which no signature is available.This paper focus on the anomaly based detection techniques.

    The efficiency of the detection process majorly depends on the quality of the features extracted or engineered from the session/flow in the network.Domain knowledge is essential for a better feature engineering and the deep the learning algorithms have the ability to learn features automatically.The algorithms work in end-to-end nature and at present gaining more importance in the IDS research.These algorithms can analyze the raw data, learns features, and classifies normal from abnormal traffic.Convolutional Neural Network (CNN) based detection algorithms have been proposed in literatures which uses Network Security Laboratory (NSL)-Knowledge Discovery in Databases (KDD) and University of South Wales (UNSW)-NB 15 datasets [7].Initially the feature vectors are converted in to images for further processing.Using one hot coding the nominal features and the dimension of the features were increased.Each 8 byte of the feature vector is considered as a pixel in image.Each feature vector is thus transformed in to an image of size 8*8 pixels.A three layer CNN was designed to discriminate normal from abnormal traffic.Performance of the designed network was compared with other deep learning architectures including ResNet50 and GoogleNet.The proposed three layer design yields an accuracy of 91.14% on the NSL-KDD data and 94.9%accuracy on the UNSW-NB 15 dataset.In another work the features were extracted using a sparse auto-encode model and the attacks were classified using an XGBoost model [8].For experimental analysis NSL-KDD dataset was used.The imbalanced dataset was balanced using synthetic minority over-sampling technique algorithm.The minority classes were oversampled and majority classes are further sub divided in to new classes and thus achieving perfect balancing in the dataset.

    The sparse Auto-Encoder (AE) introduces inserts a sparsity constraint on the auto-encoder to enhance the ability of detecting new patterns [9].At last the data was classified by a XGBoost regression model.The model produces an accuracy of 99.6%,99.17%,99.50%,97.13%,and 89.00%,respectively while detecting Normal, Denial of Service (DOS), Probe and Remote-to-Local (R2L)attacks.Deep learning models have influenced a remarkable growth of Big Data Analytics and they suffer when the data is imbalanced or small.These deep architectures have the ability to model highly complex non-linear data distributions and their performance is better in multiple areas of applications[10].An important sub-field of deep learning is the generative networks that have shown remarkable performance on capturing data distribution and synthesizing new data samples from the same distribution.As Generative Adversarial Network(GAN)has the capacity to better capture and represent the distribution of the data,many of the recent literatures have used the generative networks for detecting anomalies through a variety of approaches[11].

    Adversarial learning methodologies could increase the accuracy of detection in small or imbalanced dataset.One of the recent literatures has used GAN for data augmentation[12].The KDD99 is considered to be one of the oldest dataset which is both unbalanced and doesn’t include new attack patterns.When models are trained on the KDD99 dataset they lack the generalizing ability.When the dataset is augmented with GAN model these issues can be overcome.The experiments included 8 attack categories and adversarial based augmentation increased the accuracy of the 7 attack categories.Using deep learning approach a higher level of features is extracted from low-level ones to achieve a powerful representation.They used recurrent architectures for learning patterns from network traffic sequences and detect network attacks[13].A two phase approach and improved version of CNN was used for anomaly detection in cloud datacenter networks.They focused on managing an optimal tradeoff between minimized error-rate and reduced feature set by using Grey Wolf Optimization (GWO)techniques for feature selection.The anomaly in the network is classified using an improved CNN architecture[14].

    A deep learning framework was developed for anomaly detection in cloud workloads where the usage patterns are analyzed for identifying failures in the cloud because of contention for resources.The resource utilization and performance of the working system are reviewed at regular intervals to model the normal and abnormal behaviors in the cloud network.A hybrid deep neural architecture is used to forecast the near future resource utilization and performance measures of the cloud service in the first stage.In stage two of the analysis the hybrid model is used for classifying the cloud behavior as either normal or abnormal.The proposed anomalous detector is evaluated in the virtual environment using docker containers.The hybrid model is constructed by combining bidirectional long short term memory and long short term memory architectures[15].

    Using the auto-scaling characteristics of the cloud an online malware detection approach was proposed in [16].This approach has the ability to detect the abnormality when the cloud service is running.The performance measures in the process level are used to model a CNN for classifying the abnormal traffic.A 2D-CNN was trained on the samples of VMs running in an auto-scaled scenario.The samples from multiple VM have no correlations.The detection accuracy is enhanced by considering the similarity between different VM through a sample-paring technique.Resilience can be considered as the capacity of a system to offer services in an acceptance level under various challenges.In the case of cloud services the resilience is a fundamental characteristics or feature by which new VM’s can be created on an on-demand basis when the load becomes high.Majority of the critical infrastructure services offered in the cloud network are not resilient due to issues in the network infrastructure[17].

    Automatic detection and classification of traffic pattern as anomalous is a challenging task that was handled using different approaches and techniques proposed in various literatures.Conventional machine learning techniques and algorithms are sub optimal and are not capable of extracting or uncovering patterns from a high dimensional data.They cannot capture the complex patterns in a high dimensional and voluminous data.This is the reason for engaging deep learning techniques and algorithms to detect anomalies in the cloud network.Based on the taxonomy presented in Fig.2 it is clear that the deep learning techniques are used at present to classify the traffic pattern as either normal or anomalous.

    In a recent work [18] based on heterogeneous data prevailing in the cloud network a hybrid detection model was proposed.The proposed method followed two essential steps for detecting the abnormality; first an optimal set of features were selected from the traffic stream using GWO [19]which is a meta-heuristic algorithm based on evolutionary approach.Then they are classified using CNN either as benign or anomalous traffic.

    To train a deep learning model a large volume of training samples are required and these data are preprocessed to resolve the problem of high dimensionality using dimensionality reduction,clustering,and sampling techniques.Later these data are discriminated using a deep classifier.For constructing a deep classifier a training and test data set are essential and additionally a validation data is used to tune the model hyper parameters.Using the training dataset the model parameters are tuned and test data is used for evaluating the performance of the trained model[20].The process of training and evaluating the deep model performance is presented in Fig.3.Deep learning algorithms are subset of machine learning and the performance of the deep learning models are superior to the conventional machine learning or shallow algorithms in majority of application context.

    Figure 2:Taxonomy of deep learning based anomaly detection

    Figure 3:General process flow in classification of traffic patterns

    In the cloud network the traffic originates from different heterogeneous sources and it varies immediately due to the elasticity of the services offered to the clients.The large volume of normal traffic and in contrast a low volume of abnormal traffic in the network poses certain challenges.Some of conventional IDS follow signature based approach in detecting the abnormal attacks.In contrast to signature based techniques,anomaly based techniques have been used in cloud computing at various levels which are highly capable of finding new attacks.

    The inbound and outbound network traffics are continuously monitored by the IDS and upon further analysis it raises an alarm when there is anomaly being detected.Based on the detection approach used, the IDS can be classified as either signature based or anomaly based intrusion detection.The predefined rules are matched against the pattern extracted from the present network traffic data.Then they are classified as intrusion attack if they vary from the usual network traffic pattern.This approach yield high accuracy in detecting the known category of attacks and generates low false alarm rate.It does not have ability to detect the new category of attacks as the predefined rules don’t match the pattern observed in the new attacks.The anomaly based intrusion detection has the ability to detect even new category of attacks.The accuracy of the anomaly based detection is more when compared to the signature based approach as per the theoretical proofs specified in the literatures.The main drawback of the anomaly based detection is high false alarm rate[21].The main challenges faced while designing anomaly detection are selecting an optimal set of feature from the network traffic and less volume of supervised labeled dataset.The patterns observed from the intrusion attacks are changing from period to period and hence a common set of features chosen for differentiating attacks from normal network traffic flow cannot be suitable in all the cases.

    The analysis of network traffic provides more insights on the behavior on the performance of the cloud environment.Due to such revenue growth of the cloud service providers it becomes essential to develop cloud traffic monitoring and analysis methods to increase the availability,and security of the cloud environment.Monitoring and analyzing such a huge volume of network traffic data is a more challenging task.The traditional methods used for monitoring and analyzing network traffic will not suit for cloud environment.The cloud network pattern differs greatly from the patterns observed in a corporate distributed network.

    3 Proposed Methodology

    Different type of anomaly detection techniques have been implemented so far in various literatures,but majority of them has not focused on the analysis of impact of elasticity of the cloud during VM migration.The proposed methodology focuses on evaluating the performance of the anomaly detection under such challenging conditions.The support for service migration and migration of VM to other physical nodes in the cloud network exploits the elastic property of the cloud for dynamic movement of the cloud resources.For effective management of cloud resources in online and perfect balancing of computing load across physical nodes of the cloud; this real time migration becomes essential.The anomaly detection module might consider the live migration as an anomaly (false positive) or some time an anomalous event occurring during the migration the detection may be masked(false negative).

    The network packets are used as source of data for the detection of anomalies and hence the U2Land R2L attacks can be detected effectively.As the packet headers contain the IP addresses the attack source can be detected precisely.Analysis of information extracted from the packets can be done in real time.A single packet does not reveal much information about the context and hence detecting attacks like distributed DoS become difficult.The detection process includes parsing packets, and analysis of the packet payloads.Fig.4 presents the schematic view of packet based real time anomaly detection.

    Figure 4:Real time anomaly detection

    3.1 Training Data

    For efficient training of deep learning models a large volume of balanced anomaly dataset is essential and it is obtained from the traces collected during the simulation of the cloud environment.The traces of TCP streams obtained from the simulated cloud network include both genuine and attack traffic.The effects of VM migration will also be reflected in the traces.The incoming traffic to the physical node becomes different as the VM migrates to a different physical machine.Also the anomalies especially volume based attacks will be best characterized by the traces of the TCP streams as they consume more bandwidth from the normal traffic.The data for training the deep learning model is collected at various level of the cloud including the traffic level, hypervisor and physical machine level.A feature vector is extracted from each one second bin of the traffic traces and the system level performance metrics such as CPU utilization rate,memory utilization,network bandwidth consumption,number of I/O operations,and number of processes waiting for execution in the queue were considered.The network level performance metrics included in the feature vector are number of lost packets,volume of traffic in the network port,and overall load rate of the network.

    3.2 Data Augmentation Using GAN

    The acquired dataset is highly imbalanced and an efficient model cannot be build using this dataset.Hence the dataset must be balanced before training the model.The aim is to use GAN architectures to discover the pattern in the data that makes them more realistic as uncovering patterns in the data is not possible using other methods.The generative networks help to balance the dataset by synthesizing new samples for minority classes.It is proved that GAN have shown impressive results when compared to other generative architectures including variation AE, and restricted boltzman machines.Synthesizing new samples is a complex task when the dimensionality of the data is high[22].From the literature review it was obvious that variation AEs[23]and GANs[24]are most successfully used deep architectures for data augmentation.The objective functions for GANs can be chosen among Jensen-Shannon [24] and f-divergences [25].There are other several methods to define the distance or divergence between the distribution of model generated and real data.The discriminator network in the GAN output the probability that a given data is real or synthesized.The discriminator is given a set of input that includes both real and synthesized data.It generates a probability estimate for each input.The loss of the discriminator network can be measured by cross-entropy function.Cross entropy based loss function is similar to the Jensen-Shannon objective function and they tend to fail in certain cases [26].To overcome such issues Wasserstein distance based loss function is used.The cross entropy loss function measures the accuracy in the detection of real and synthesized data by the discriminator network.The distribution of each vector in the real and synthesized data is estimated and the distance between them is calculated.It gives the measure of how much mass times distance is required to make the distribution of synthesized data similar to the distribution of real data.

    The flow of input data through layers in the GAN network is presented in Fig.5.Mathematical representation of the min-max loss function used in the GAN network is given in Eq.(1) where the generator network attempts to minimize it while the discriminator network tries to maximize it.

    where D(x)is the probability that the generated data is real estimated by the discriminator;Exis the expected values for all the data samples;G(z)is the output of the generator network when the random noise is given as input;D(G(z))is the probability that the fake instance is real;Ezis the expected value for all the fake instancesG(z) synthesized by the generator.The mathematical expression is derived from the cross entropy between the real and generated data distributions.The discriminator network produces output for the generator to improve its performance.The generator learns nothing when the gradient of the generator network diminishes and approaches close to zero.

    Figure 5:Process flow in wasserstein GAN

    To overcome the vanishing gradient problem the following alternate cost function can be used:

    In[23]the authors have illustrated that the alternate cost function has large variance of gradients which causes the model unstable.Also they suggested that adding noise to the generator output to stabilize the model.The schematic view of the Wasserstein GAN(WGAN)is shown in Fig.6.

    Figure 6:Schematic view of training a WGAN with added noise to the generator output

    Rather than adding noise, a new cost function based on the Wasserstein distance was proposed in[26].Based on the Kantorovich-Rubinstein duality[27]the Wasserstein distance can be expressed as given below.

    wheresuprepresents the least upper bound andfdenotes a 1-lipschitz function which obeys the following constraint

    When the lipschitz function is derived fromK-lipschitz functions parameterized by w, {fw}w ∈W.The discriminator network is trained to learn the lipschitz function to compute the Wasserstein distance.When the value of loss function decreases in the training then the Wasserstein distance gets smaller and the model of the generator becomes closer to the distribution of real data.

    The loss function can be formulated as to find the Wasserstein distance between prand pg

    In order to maintain theK-lipschitz continuity of fwduring the training process the weights are clamped after every gradient update to a small window such as [-0.01, 0.01].Thus the lipschitz’s continuity can be preserved by obtaining the lower and upper bound offw.

    3.3 Anomaly Detection

    The anomalous traffic are detected based on the compact representation of the feature vectors obtained using a deep AE.The AE learns to map the given input data to a compact representation with two unsupervised training steps.The AE can be classified as a generative model and it has the ability to learn and extract the similarity and correlation in the input data[28].An efficient intrusion detection model must be sensitive to the input, yield less reconstruction error, and it should not get overfit for certain input data.The loss function is designed with two terms to achieve the above mentioned constraints.

    The first part of the above equation makes the model sensitive to the given input data and the second terms prevent the model from overfitting on the training data.The trade-off between these two different objectives can be achieved by tuning the alpha scaling parameter.Based on the nature and characteristics of the given output the AE model can be considered as non-linear generalization of principal component analysis and the AE model has the ability to learn non-linear relationship between given input and expected output.The AE model helps to separate the normal data from the anomalous data by transforming the given input data on to new axes.The AE consists of two neural networks namely encoder and decoder.The encoder compresses the data points into a lower dimensional representation and the decoder network attempts to reconstruct the original input points from the latent representation generated by the encoder network.The AE parameters are tuned by minimizing the reconstruction error which is the difference between the input data points and reconstructed data points.The AE is trained in an unsupervised mode with features extracted from normal traffic data both under normal circumstances and VM migration.The AE will be capable of reconstructing the normal traffic data points and fails to reconstruct the anomalous traffic data points.The reconstruction error is used as anomalous score.The performance of the AE trained in an unsupervised learning strategy is compared with a binary classifier,SVM.The SVM model is trained with a Radial Basis Function(RBF)kernel function.

    The non-linear kernel represents the similarity between two different vectors as a function of the squared norm of their distance.That is,if the two vectors are close together then,‖x-x′‖will be small.Then,so long asγ >0, it follows that-γ‖x-x′‖2will be larger.Thus,closer vectors have a larger RBF kernel value than farther vectors.

    4 Experiments and Results

    The simulation of the cloud environment is accomplished in CloudSim 5.0 and the metricslisted in Section 3.1 are extracted from the normal network traffic.For creating the VM migration within the simulated cloud environment,initially the list of over utilized host is collected and the backup of over utilized VM to be migrated is done.Then they are mapped to a new suitable host using a migration map.The VM migration is considered to essential to test the resilience of the cloud services under different attacks when they are exposed to a variety of attacks.The following attacks were introduced in the simulated cloud network;Net Scan(NS),and DoS.The network traces and the metrics of the host machine are collected under multiple time instances of the simulation.Then required features are extracted from the traces and labeled as normal or abnormal respectively.Two set of feature set were extracted;one under VM migration and another without migration.Both during migration and normal period anomalous traffic were introduced in the cloud simulating the above mentioned attacks.The model was tested to detect network level attacks as more end users will be accessing the services of the cloud thus increasing the attack surface.More computational power will be allocated in the form of VM in on-demand basis.The anomalous traffic is generated by injecting the attacks into the legitimate traffic at irregular intervals.The VMs are migrated live among the nodes during the simulation may be during normal traffic period or anomalous traffic period.The network traces collected for every 1-second been using a packet analyzer script embedded within the simulated network.The harmonic meanFscore and geometric mean provides accurate measure of the performance of a particular anomalous detector keeping all the outcomes to certain degree.The network traffic in the cloud simulation is fixed constant and the cloud environment is simulated with five VMs running web servers.There are two physical host nodes available in the simulation;one with three VMs and another host contains two VMs.The web traffic is generated from a VM to another VM running inside another host node using TRex,an open source traffic generator.

    As described earlier the data augmentation task is implemented using WGAN to balance the volume of normal and anomalous traffic.The threshold value c,considered as one of the important hyperparameter is fixed based on Bayesian optimization technique.TheRMSPropoptimization algorithm is utilized instead of momentum based optimization algorithm like Adam.The Adam algorithm causes instability of the model during the training process.The performance of the WGAN model is sensitive to the clipping hyper parameter.The graph shown in Fig.6 shows the explosion gradient when the value of c is varied from 0.001 to 0.1.

    The weight clipping method acts as a weight regulator and it reduces the performance of the model thereby limiting the capacity of the model to learn complex function.Hence instead of gradient clipping, Gradient penalty technique was adopted in the experiments.The plot in Fig.7 illustrates the advantage of using gradient penalty over gradient clipping.When the value of weight clipping thresholdcis fixed either low or high values then the gradient either explodes or vanishes.Batch normalization is included in the discriminator network as it creates correlation between samples in the same batch.The experimental evaluations showed that it impacts the effectiveness of the gradient penalty.

    Figure 7:Plot of gradient norm in log scale

    Gradient penalty was calculated based on the following procedure;

    1.Estimate the gradient values with respect to the input data.

    a) Create a combined data by weighing the synthetic and real data using epsilon and fusing them together.

    b) Compute the discriminator’s output for the fused data.

    2.Calculate the gradient penalty for the estimated gradient.

    a.Consider the magnitude the gradient

    b.Find the penalty

    The performance of the AE and SVM based classification was analyzed using Receiver operating Characteristics (RoC) curves shown in Figs.8 and 9.Based on the experimental results under VM migration the performance of the anomaly detection under both volume based attack(DoS)and nonvolume based attack (NS) using SVM is degraded.From the RoC curves presented in Figs.8a and 8b,it is evident that the SVM classifier is sensitive to the type and volume of the anomalous traffic.At few instances the performance of the SVM classifier degraded when VM migration occurs.In case of volume based attack, DOS the SVM performs for both low and high intensity attacks but under influence of VM migration its performance decrease.The True Positive Rate (TPR) is decreased by 32%during VM migration even for low density NS attack.Fig.8b presents the influence of migration for DoS attack detection;also it is observed that more than 85%of the TPR is achieved during with and without migration.The False Positive Rate(FPR)is increased by 5%when the migration starts.In parallel from Fig.9a it can be observed that almost more than 85%of the anomalous traffic was detected under high anomalous(NS)traffic scenario for both with and without migration.Even for the same attack types under low anomalous traffic the model was able to yield an average of 80%of TPR with less false positive rate.For DoS attack the AE based anomaly detection is capable of identifying anomalous traffic for both attack types and it is not affected under migration.The FPR rate is less than 15% and it is acceptable under migration of VM.From the analysis of the results it is inferred that the virtual migration has no impact on the performance of the AE in detecting the anomalies.More traces are samples from the same scenario in order make the detection more efficient.

    Figure 8:(a)RoC for NS using SVM(b)RoC for DoS using SVM

    Figure 9:(a)RoC for NS using AE(b)RoC for DoS using AE

    The results presented in Tab.1 show the efficiency of the anomaly detection for network based and volume based attacks.The virtual migration has affected the performance of the detector only low density anomalous traffic in the simulation as the False Positive Rate was observed to be high during this period.The detector classifies the migration traffic also as anomalous.The increase in the FPR is also only 5%to 10%which does not impact the overall performance of the classifier.The feature set used for detection can be further expanded by including few statistical features derived from them.These additional features can be useful while detecting certain other complex attacks.Fig.10 presents the plot of the training loss curve,for both discriminator and generator network over generator iterations for the following models;WGAN with weight clipping andRMSPropand WGAN with gradient penalty withADAM.

    Table 1: Performance of AE in anomaly detection

    Figure 10:Training loss analysis

    For analyzing the performance of the VMs deployed with certain workloads memory, CPU utilization,and cap values are tabulated in Tab.2.Each VM hosts a selected workload by allocating one or more CPU cores,RAM from 512MB(shown in Fig.11)–2048MB.Simultaneously the CAP value is varied between the range 25%and 200%of the CPU utilization and the estimated the corresponding CPU utilization rate.

    The proposed hybrid approach does include a deep architecture for synthetic generation of samples and a conventional machine learning algorithm,SVM for detecting anomalies.The selection of important hyper-parameters and architecture of the deep network introduces certain computational complexity as more layers are used to construct the deep model.In general the linear SVM detects the anomaly with a complexity ofO(d)wheredis the dimension of the input data.The experiments adopted an RBF kernel which has a complexity ofO(d2)whereas the polynomial kernel will have a complexity ofO(n×d)where n denotes the number of supper vectors.During experiments it was observed that the complexity of the deep network increases as the depth of the network is increased.It was also observed that when optimal values of hyper parameters are configured for the network the synthetic samples were more similar to the original input data which shows that the depth of the network decides the performance of the network.

    Table 2: CPU utilization in%

    Figure 11:Analysis of CPU utilization(allocated RAM=512 MB)

    5 Conclusions

    This paper explored the effect of VM migration on the performance of the anomaly detection and proposed features and robust classification approach to manage the resilience of the cloud service to overcome security issues.Experiments were conducted by simulating the VM migration and varying the density of the anomalous traffic in the network.The deep learning based detection mechanism and the features extracted from the network traces helped to retain the resilience of the cloud environment.The reconstruction error of the AE model is used as the anomalous score to detect deviation in the network traffic patterns.This anomaly detection is tested in a simulation environment wherein the anomaly detection was executed in parallel with other events of the simulation.This work focused on detecting two attacks namely NS and DoS.Future work will focus on detecting more number of attacks by enhancing the features set used in this work with more optimal features.The results are found to be satisfactory in resolving the security concerns in cloud services for its application in manufacturing sector.

    As a benchmarked dataset is not available to test the resilience of the cloud infrastructure, data samples from the simulated network have been generated,balanced using GAN network and classified as either anomalous or normal using an AE model.The trained model is able to detect anomalous traffic only in similar cloud environment simulated in the experiments.To overcome this limitation and develop a generic deep learning based anomaly detection system further data samples must be collected from a real time network, benchmarked and must be used for training the deep learning model.

    Funding Statement:The authors received no specific funding for this study.

    Conflicts of Interest:The authors declare that they have no conflicts of interest to report regarding the present study.

    寂寞人妻少妇视频99o| 视频中文字幕在线观看| 精品国产乱码久久久久久小说| 国产精品一区二区性色av| 五月开心婷婷网| 精品少妇黑人巨大在线播放| 免费人成在线观看视频色| 男女那种视频在线观看| 国产一级毛片在线| 免费在线观看成人毛片| 最近2019中文字幕mv第一页| 久久久久网色| 春色校园在线视频观看| 91久久精品国产一区二区成人| 一级毛片久久久久久久久女| 一边亲一边摸免费视频| 五月玫瑰六月丁香| 97超视频在线观看视频| 丰满人妻一区二区三区视频av| 26uuu在线亚洲综合色| av国产免费在线观看| 哪个播放器可以免费观看大片| 亚洲成人精品中文字幕电影| 亚洲一区二区三区欧美精品 | 欧美bdsm另类| 精品久久久久久久久亚洲| av播播在线观看一区| 纵有疾风起免费观看全集完整版| 国产白丝娇喘喷水9色精品| 九草在线视频观看| 国产欧美亚洲国产| 国产精品一区二区三区四区免费观看| 内射极品少妇av片p| 日日摸夜夜添夜夜添av毛片| 人人妻人人看人人澡| 小蜜桃在线观看免费完整版高清| 小蜜桃在线观看免费完整版高清| 80岁老熟妇乱子伦牲交| a级一级毛片免费在线观看| 日韩欧美一区视频在线观看 | 免费电影在线观看免费观看| 色吧在线观看| 中国三级夫妇交换| 我的女老师完整版在线观看| 七月丁香在线播放| 免费av毛片视频| 日日撸夜夜添| 亚洲精品乱久久久久久| 国产欧美另类精品又又久久亚洲欧美| 亚洲av免费在线观看| 国产精品人妻久久久影院| 啦啦啦在线观看免费高清www| 又粗又硬又长又爽又黄的视频| 又粗又硬又长又爽又黄的视频| 啦啦啦在线观看免费高清www| 少妇的逼好多水| 夫妻午夜视频| 亚洲最大成人av| 精品少妇久久久久久888优播| 亚洲,一卡二卡三卡| 一个人观看的视频www高清免费观看| 亚洲aⅴ乱码一区二区在线播放| 欧美xxxx黑人xx丫x性爽| 色播亚洲综合网| 国模一区二区三区四区视频| 春色校园在线视频观看| 亚洲欧美清纯卡通| 亚洲精品国产色婷婷电影| 26uuu在线亚洲综合色| 91精品一卡2卡3卡4卡| 在现免费观看毛片| 亚州av有码| 亚洲色图av天堂| 久久精品久久久久久久性| 偷拍熟女少妇极品色| 亚洲精品,欧美精品| 色5月婷婷丁香| 亚洲图色成人| 男插女下体视频免费在线播放| 欧美少妇被猛烈插入视频| 精品一区二区免费观看| 久久久午夜欧美精品| 国产免费一级a男人的天堂| 欧美日韩综合久久久久久| 夫妻性生交免费视频一级片| 身体一侧抽搐| 大码成人一级视频| 亚洲,一卡二卡三卡| 超碰97精品在线观看| 免费av观看视频| 亚洲av欧美aⅴ国产| 免费av毛片视频| av在线播放精品| 午夜福利视频1000在线观看| 在线亚洲精品国产二区图片欧美 | 大陆偷拍与自拍| 交换朋友夫妻互换小说| 国产黄频视频在线观看| 亚洲精品色激情综合| 一级毛片电影观看| 丝袜美腿在线中文| 大香蕉97超碰在线| 18禁动态无遮挡网站| 80岁老熟妇乱子伦牲交| 欧美区成人在线视频| 日韩欧美精品v在线| 国产成人精品婷婷| 一个人观看的视频www高清免费观看| 亚洲欧美一区二区三区国产| 久久精品国产鲁丝片午夜精品| 色5月婷婷丁香| 国内精品美女久久久久久| 国产一区二区在线观看日韩| 国产一区二区亚洲精品在线观看| 熟女人妻精品中文字幕| 国产毛片在线视频| 久久久久久久久大av| 亚洲四区av| a级一级毛片免费在线观看| 亚洲精品久久久久久婷婷小说| 色视频在线一区二区三区| 亚洲国产成人一精品久久久| 1000部很黄的大片| 欧美一区二区亚洲| 久久久久久久大尺度免费视频| 天天一区二区日本电影三级| 纵有疾风起免费观看全集完整版| 久久久久久伊人网av| 青春草亚洲视频在线观看| 成人无遮挡网站| 少妇人妻精品综合一区二区| 天堂网av新在线| 免费看光身美女| 三级国产精品欧美在线观看| 我的女老师完整版在线观看| 精品久久久久久久末码| 国产精品伦人一区二区| 国产淫语在线视频| 日本黄大片高清| 97人妻精品一区二区三区麻豆| 久久人人爽人人爽人人片va| 热99国产精品久久久久久7| 久热久热在线精品观看| 国产精品伦人一区二区| 一级av片app| 狠狠精品人妻久久久久久综合| 91精品国产九色| 欧美成人精品欧美一级黄| 干丝袜人妻中文字幕| 国产v大片淫在线免费观看| 免费在线观看成人毛片| 大香蕉97超碰在线| 日韩欧美精品免费久久| 国产高清有码在线观看视频| 91精品伊人久久大香线蕉| 在现免费观看毛片| 国产91av在线免费观看| 国产大屁股一区二区在线视频| 日韩成人伦理影院| 国产精品一区二区性色av| 国产一区二区在线观看日韩| 国产男女内射视频| 亚洲国产精品成人综合色| 一本一本综合久久| 女人被狂操c到高潮| 国产高清三级在线| 亚洲国产精品999| 精品熟女少妇av免费看| 欧美日韩视频精品一区| 在线观看三级黄色| 69av精品久久久久久| 在线免费十八禁| 国产欧美日韩一区二区三区在线 | 涩涩av久久男人的天堂| 国产成人午夜福利电影在线观看| 黄色日韩在线| 日产精品乱码卡一卡2卡三| 我的女老师完整版在线观看| 视频中文字幕在线观看| 男人爽女人下面视频在线观看| 国产69精品久久久久777片| tube8黄色片| 亚洲在久久综合| 国产高潮美女av| 少妇人妻 视频| 国产精品爽爽va在线观看网站| 91精品伊人久久大香线蕉| 久久久久性生活片| 国产精品麻豆人妻色哟哟久久| 国产成人freesex在线| 永久网站在线| 哪个播放器可以免费观看大片| 国产精品人妻久久久影院| 欧美zozozo另类| 最近的中文字幕免费完整| 亚洲天堂国产精品一区在线| 中国美白少妇内射xxxbb| 精品熟女少妇av免费看| 夫妻性生交免费视频一级片| 小蜜桃在线观看免费完整版高清| 亚洲经典国产精华液单| 亚洲av国产av综合av卡| 高清日韩中文字幕在线| 亚洲精品久久午夜乱码| 亚洲国产精品成人久久小说| 高清毛片免费看| 国产伦理片在线播放av一区| 永久免费av网站大全| 精品国产三级普通话版| 欧美一区二区亚洲| 国产精品一区www在线观看| 久久久a久久爽久久v久久| 天天一区二区日本电影三级| 亚洲激情五月婷婷啪啪| 大码成人一级视频| 亚洲婷婷狠狠爱综合网| 欧美丝袜亚洲另类| av黄色大香蕉| 国产成人精品福利久久| 街头女战士在线观看网站| 亚洲av福利一区| 欧美日韩国产mv在线观看视频 | 在线观看免费高清a一片| 99热全是精品| 国产欧美日韩精品一区二区| 看黄色毛片网站| 色综合色国产| 国产精品成人在线| 国产亚洲午夜精品一区二区久久 | 久久女婷五月综合色啪小说 | 91在线精品国自产拍蜜月| 麻豆精品久久久久久蜜桃| kizo精华| 国产黄频视频在线观看| 欧美一区二区亚洲| 国产精品久久久久久精品古装| 国产成人一区二区在线| 国产真实伦视频高清在线观看| 欧美性感艳星| 国产精品久久久久久精品古装| 国产探花极品一区二区| 男女下面进入的视频免费午夜| 丝袜美腿在线中文| 免费看av在线观看网站| 欧美高清成人免费视频www| 两个人的视频大全免费| 丰满人妻一区二区三区视频av| 卡戴珊不雅视频在线播放| 天堂俺去俺来也www色官网| 亚洲av中文字字幕乱码综合| 在线看a的网站| 国产精品久久久久久精品电影| 69人妻影院| 蜜桃亚洲精品一区二区三区| 另类亚洲欧美激情| 日韩人妻高清精品专区| 日本猛色少妇xxxxx猛交久久| 一本久久精品| 久久久久久久久久久丰满| 黄色一级大片看看| 久久国产乱子免费精品| 国产高清国产精品国产三级 | 精品久久久久久久末码| 直男gayav资源| 国产精品一区二区在线观看99| 最新中文字幕久久久久| 日本av手机在线免费观看| 国产成人aa在线观看| av专区在线播放| 三级国产精品欧美在线观看| 免费电影在线观看免费观看| 一本久久精品| 亚洲精品一区蜜桃| 大陆偷拍与自拍| 亚洲av男天堂| 久久97久久精品| 久久99热6这里只有精品| 日韩视频在线欧美| 午夜福利在线观看免费完整高清在| 99久久精品热视频| 嫩草影院新地址| 亚洲精品影视一区二区三区av| av在线播放精品| av国产久精品久网站免费入址| 热99国产精品久久久久久7| 下体分泌物呈黄色| 久久人人爽人人片av| 特大巨黑吊av在线直播| 国产精品蜜桃在线观看| 日日啪夜夜爽| 国产欧美日韩精品一区二区| 另类亚洲欧美激情| 色5月婷婷丁香| 亚洲欧美日韩东京热| 亚洲经典国产精华液单| 大话2 男鬼变身卡| 亚洲精品久久久久久婷婷小说| 97人妻精品一区二区三区麻豆| 亚洲欧洲国产日韩| 日韩三级伦理在线观看| 亚洲最大成人av| 亚洲av一区综合| 亚洲精品国产av成人精品| 国产日韩欧美在线精品| 97在线人人人人妻| 精品少妇黑人巨大在线播放| 欧美日韩在线观看h| 黄色怎么调成土黄色| 亚洲精品456在线播放app| 国产成人精品福利久久| 亚洲美女视频黄频| 精品久久久精品久久久| 国产成人精品一,二区| 午夜福利网站1000一区二区三区| 久久人人爽人人爽人人片va| 少妇 在线观看| 国产伦在线观看视频一区| 婷婷色综合大香蕉| 欧美成人a在线观看| tube8黄色片| 国产爱豆传媒在线观看| 在线观看国产h片| 成人二区视频| 国产黄片美女视频| 一本色道久久久久久精品综合| 久久久久网色| 国产伦理片在线播放av一区| 国产真实伦视频高清在线观看| 国产免费视频播放在线视频| 真实男女啪啪啪动态图| av在线蜜桃| 在线观看国产h片| 极品教师在线视频| 精品国产一区二区三区久久久樱花 | 黄片无遮挡物在线观看| 在线观看国产h片| 国产一级毛片在线| 色5月婷婷丁香| 久久99精品国语久久久| 国产成人福利小说| 97精品久久久久久久久久精品| 日韩欧美一区视频在线观看 | 国产精品人妻久久久久久| 亚洲一级一片aⅴ在线观看| 亚洲av不卡在线观看| av播播在线观看一区| a级毛色黄片| 亚洲精品乱码久久久久久按摩| 大片免费播放器 马上看| 乱码一卡2卡4卡精品| 亚洲av欧美aⅴ国产| 国产一区二区亚洲精品在线观看| 青春草国产在线视频| 亚洲欧美日韩东京热| 国产欧美另类精品又又久久亚洲欧美| 51国产日韩欧美| 亚洲国产日韩一区二区| 亚洲一级一片aⅴ在线观看| 韩国av在线不卡| 国产老妇女一区| 亚洲av电影在线观看一区二区三区 | 免费大片黄手机在线观看| 又粗又硬又长又爽又黄的视频| 国产精品久久久久久久久免| 我的老师免费观看完整版| 久久99热这里只频精品6学生| 成人美女网站在线观看视频| 精品熟女少妇av免费看| av又黄又爽大尺度在线免费看| 午夜激情福利司机影院| 亚洲色图av天堂| 欧美精品国产亚洲| 国产精品蜜桃在线观看| 国产高潮美女av| 国产片特级美女逼逼视频| 我的老师免费观看完整版| 日韩亚洲欧美综合| 日韩欧美 国产精品| 在线免费十八禁| 男的添女的下面高潮视频| 男人爽女人下面视频在线观看| 两个人的视频大全免费| 少妇 在线观看| 日日啪夜夜撸| 久久精品国产亚洲网站| 亚洲熟女精品中文字幕| 国产爽快片一区二区三区| 亚洲欧洲国产日韩| 国产综合懂色| 日韩av免费高清视频| 亚洲三级黄色毛片| 亚洲精品国产成人久久av| 国产成人午夜福利电影在线观看| 在线观看美女被高潮喷水网站| 麻豆成人av视频| 精品少妇久久久久久888优播| 日本熟妇午夜| 久久午夜福利片| 精品国产一区二区三区久久久樱花 | 亚洲精品国产色婷婷电影| 97超碰精品成人国产| 亚洲av电影在线观看一区二区三区 | 亚洲国产欧美在线一区| 精品久久久久久久人妻蜜臀av| 综合色av麻豆| 日韩国内少妇激情av| 欧美zozozo另类| 一个人看的www免费观看视频| 18禁在线无遮挡免费观看视频| 精品国产一区二区三区久久久樱花 | 午夜亚洲福利在线播放| www.av在线官网国产| 亚洲国产精品成人综合色| 波野结衣二区三区在线| 一本久久精品| 国产爱豆传媒在线观看| 日本黄色片子视频| 在线观看一区二区三区| videos熟女内射| 一本色道久久久久久精品综合| 欧美一级a爱片免费观看看| 欧美精品国产亚洲| 免费av观看视频| 午夜免费鲁丝| 国产男女超爽视频在线观看| 毛片女人毛片| 国产黄色免费在线视频| 国产精品久久久久久久久免| 亚洲综合色惰| tube8黄色片| 99久久精品热视频| 蜜桃亚洲精品一区二区三区| 免费观看a级毛片全部| 欧美极品一区二区三区四区| 日韩大片免费观看网站| 亚洲丝袜综合中文字幕| 欧美性猛交╳xxx乱大交人| 国产一区二区三区综合在线观看 | 午夜精品国产一区二区电影 | 超碰97精品在线观看| 少妇被粗大猛烈的视频| 国产色爽女视频免费观看| 精品久久久久久久久av| 亚洲无线观看免费| 六月丁香七月| 两个人的视频大全免费| 美女高潮的动态| 少妇丰满av| 国产亚洲av片在线观看秒播厂| 黑人高潮一二区| 人妻 亚洲 视频| 综合色丁香网| 亚洲成人av在线免费| 成年人午夜在线观看视频| 亚洲精品自拍成人| 国产欧美日韩一区二区三区在线 | av国产精品久久久久影院| 热99国产精品久久久久久7| 国产淫语在线视频| 久久久久久久久大av| 午夜精品国产一区二区电影 | 精品久久久久久电影网| 日韩电影二区| 黄色怎么调成土黄色| 亚洲成色77777| 在线 av 中文字幕| 欧美亚洲 丝袜 人妻 在线| 嘟嘟电影网在线观看| 国产一区二区在线观看日韩| 真实男女啪啪啪动态图| 激情 狠狠 欧美| 少妇高潮的动态图| 五月天丁香电影| 亚洲精品,欧美精品| 亚洲精品国产av蜜桃| 我的女老师完整版在线观看| 80岁老熟妇乱子伦牲交| 啦啦啦中文免费视频观看日本| 美女脱内裤让男人舔精品视频| 一本一本综合久久| 黄片无遮挡物在线观看| videos熟女内射| 亚洲欧洲日产国产| 国产伦理片在线播放av一区| 男男h啪啪无遮挡| 国产成人精品久久久久久| 韩国av在线不卡| 日韩精品有码人妻一区| 欧美97在线视频| 亚洲av电影在线观看一区二区三区 | 自拍欧美九色日韩亚洲蝌蚪91 | 国产探花在线观看一区二区| 观看美女的网站| 夫妻性生交免费视频一级片| 日韩 亚洲 欧美在线| 精品久久久久久久久亚洲| 亚洲天堂国产精品一区在线| 国产视频首页在线观看| 人人妻人人澡人人爽人人夜夜| 国产淫片久久久久久久久| 日本与韩国留学比较| 欧美成人午夜免费资源| 国产亚洲午夜精品一区二区久久 | 亚洲四区av| 免费看a级黄色片| 欧美成人午夜免费资源| 男女边吃奶边做爰视频| 久久鲁丝午夜福利片| 精品久久久精品久久久| 国产精品国产三级专区第一集| 欧美高清性xxxxhd video| 久久久国产一区二区| 十八禁网站网址无遮挡 | 国精品久久久久久国模美| 欧美日韩一区二区视频在线观看视频在线 | 成年女人在线观看亚洲视频 | 亚洲欧美精品自产自拍| 国产69精品久久久久777片| 亚洲av男天堂| 欧美老熟妇乱子伦牲交| 国产在视频线精品| 日本与韩国留学比较| 乱码一卡2卡4卡精品| 久久影院123| 最近中文字幕2019免费版| 少妇 在线观看| 国产精品av视频在线免费观看| 国产欧美日韩精品一区二区| 一二三四中文在线观看免费高清| 亚洲av欧美aⅴ国产| 成人亚洲精品av一区二区| 大片电影免费在线观看免费| www.av在线官网国产| av女优亚洲男人天堂| 欧美日韩精品成人综合77777| 国产精品麻豆人妻色哟哟久久| 欧美一区二区亚洲| 欧美xxⅹ黑人| 极品教师在线视频| 亚洲内射少妇av| 成年版毛片免费区| 国产 精品1| 亚洲美女视频黄频| 国产精品国产av在线观看| 久久午夜福利片| 国产极品天堂在线| 乱码一卡2卡4卡精品| 美女xxoo啪啪120秒动态图| 你懂的网址亚洲精品在线观看| 成年免费大片在线观看| 国产一区亚洲一区在线观看| 亚洲,欧美,日韩| 禁无遮挡网站| 亚洲成人中文字幕在线播放| 日韩电影二区| 天天躁日日操中文字幕| 亚洲精品久久久久久婷婷小说| av在线app专区| 深爱激情五月婷婷| 肉色欧美久久久久久久蜜桃 | 免费电影在线观看免费观看| 伦精品一区二区三区| 男人和女人高潮做爰伦理| 久久99蜜桃精品久久| 免费黄频网站在线观看国产| 青青草视频在线视频观看| 最近中文字幕2019免费版| 日本黄色片子视频| 国产亚洲91精品色在线| 成年版毛片免费区| 日韩一区二区三区影片| 黑人高潮一二区| 免费大片黄手机在线观看| 国模一区二区三区四区视频| 大陆偷拍与自拍| 在线 av 中文字幕| 日韩人妻高清精品专区| 欧美3d第一页| av又黄又爽大尺度在线免费看| 国产欧美亚洲国产| 国产黄色视频一区二区在线观看| 免费av毛片视频| 欧美激情国产日韩精品一区| 老司机影院成人| 搡女人真爽免费视频火全软件| 日韩在线高清观看一区二区三区| 高清欧美精品videossex| 欧美日韩精品成人综合77777| 99久久精品一区二区三区| 亚洲国产最新在线播放| 少妇人妻一区二区三区视频| 可以在线观看毛片的网站| 国模一区二区三区四区视频| 欧美日韩在线观看h| 国产淫语在线视频| 一区二区三区免费毛片| 少妇人妻一区二区三区视频| av专区在线播放| 18禁裸乳无遮挡免费网站照片| 亚洲精品日韩av片在线观看| a级一级毛片免费在线观看| 18禁裸乳无遮挡免费网站照片| 日本色播在线视频| 国产在线男女| 大香蕉97超碰在线| 精品国产三级普通话版| 亚洲无线观看免费| 亚洲欧美中文字幕日韩二区| 少妇人妻久久综合中文| 伊人久久精品亚洲午夜| 日韩 亚洲 欧美在线| 一本色道久久久久久精品综合| 岛国毛片在线播放| 人妻系列 视频| 久久人人爽人人爽人人片va|