• <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 ?

    Intelligent Deep Learning Model for Privacy Preserving IIoT on 6G Environment

    2022-08-24 12:57:22AnwerMustafaHilalJaberAlzahraniIbrahimAbunadiNadhemNemriFahdAlWesabiAbdelwahedMotwakelIshfaqYaseenandAbuSarwarZamani
    Computers Materials&Continua 2022年7期

    Anwer Mustafa Hilal, Jaber S.Alzahrani, Ibrahim Abunadi, Nadhem Nemri,Fahd N.Al-Wesabi, Abdelwahed Motwakel, Ishfaq Yaseenand Abu Sarwar Zamani

    1Department of Computer and Self Development, Preparatory Year Deanship, Prince Sattam bin Abdulaziz University,Al-Kharj, 16278, Saudi Arabia

    2Department of Industrial Engineering, College of Engineering at Alqunfudah, Umm Al-Qura University, Saudi Arabia

    3Department of Information Systems, Prince Sultan University, Riyadh, 11586, Saudi Arabia

    4Department of Information Systems, King King Khalid University, Muhayel Aseer, 62529, Saudi Arabia

    5Department of Computer Science, King King Khalid University, Muhayel Aseer, 62529, Saudi Arabia

    6Faculty of Computer and IT, Sana’a University, Sana’a, 61101, Yemen

    Abstract: In recent times, Industrial Internet of Things (IIoT) experiences a high risk of cyber attacks which needs to be resolved.Blockchain technology can be incorporated into IIoT system to help the entrepreneurs realize Industry 4.0 by overcoming such cyber attacks.Although blockchain-based IIoT network renders a significant support and meet the service requirements of next generation network, the performance arrived at, in existing studies still needs improvement.In this scenario, the current research paper develops a new Privacy-Preserving Blockchain with Deep Learning model for Industrial IoT (PPBDL-IIoT) on 6G environment.The proposed PPBDLIIoT technique aims at identifying the existence of intrusions in network.Further, PPBDL-IIoT technique also involves the design of Chaos Game Optimization (CGO) with Bidirectional Gated Recurrent Neural Network(BiGRNN) technique for both detection and classification of intrusions in the network.Besides, CGO technique is applied to fine tune the hyperparameters in BiGRNN model.CGO algorithm is applied to optimally adjust the learning rate, epoch count, and weight decay so as to considerably improve the intrusion detection performance of BiGRNN model.Moreover,Blockchain enabled Integrity Check(BEIC)scheme is also introduced to avoid the misrouting attacks that tamper the OpenFlow rules of SDN-based IIoT system.The performance of the proposed PPBDL-IIoT methodology was validated using Industrial Control System Cyber-attack (ICSCA) dataset and the outcomes were analysed under various measures.The experimental results highlight the supremacy of the presented PPBDL-IIoT technique than the recent state-of-the-art techniques with the higher accuracy of 91.50%.

    Keywords: 6G networks; industrial iot; blockchain; security; intrusion detection; artificial intelligence

    1 Introduction

    The advancements made in industries, especially large-scale, in recent years have been rapid in the areas of surveillance, transportation, security, and factory automation.In this scenario, Industrial IoT (IIoT) [1] has drawn a considerable attention by integrating dense wireless devices like Radio Frequency Identification(RFID)tags[2]to identify the machines and sensor nodes for fault diagnosis,asset monitoring, large scale equipment monitoring, manufacturing, production, and other such applications for water supplies, power plants, gas, and oil refineries.Industrial Control System (ICS)is employed to describe distinct systems like Supervisory Control and Data Acquisition (SCADA)and Distributed Control Systems (DCS).SCADA analyses and collects the data from substations on a real-time basis.All substations have control devices such as Intelligent Electronic Device (IED),Remote Terminal Unit (RTU) and Programmable Logic Controller (PLC) that handle field devices,for instance, meters, actuators, sensors, etc.

    Having been considered as a fundamental enabler of Industry 4.0 and a developing offshoot of IoT,Industrial IoT(IIoT)systems are being rapidly deployed in numerous social and commercial fields like manufacturing,retail,pervasive monitoring, logistics, home automation, security surveillance, and healthcare[3].Furthermore,with astonishing advancements made in sensor network technologies and wireless communications, considerable number of devices are being presented to IIoT environment in which the raw data is processed and captured locally for data-driven decision-making process.This device has the potential to interact and communicate with one another, process and share the data and is independent of human interference [4].Hence, these devices should be safeguarded enough to preserve the data integrity and guarantee computing reliability and resource availability.Blockchain technologies introduced a novel model for the next-generation transaction-based applications, alongside a shared and distributed private or public ledger and a collective consensus method that enables accountability, trust, and transparency in the network.Fig.1 illustrates the evolution of 1G-6G.

    Figure 1: Evolution of 1G-6G

    Blockchain technique is transparent in nature owing to which it has been emphasized by the research community and industries as a possible solution to manage huge IIoT networks in an effective manner [5].Furthermore, trust-less network architectures and decentralized IIoT devices are anticipated to perform a significant part in the development of IIoT networks.These networks treat the data locally, from where it is generated and not in a centralized way.Also, it can enable device connectivity and create trustless data storage via sensors and devices which could function independent of central authorities.Also, a secured framework is generally provided by a blockchain i.e., strong against single point of failure [6].Since decentralized networks have several entry points, it adheres to fault tolerance and resilience of the network.Besides, IIoT infrastructure could be highly available with the help of distributed ledger technology.

    IIoT system is highly vulnerable and gets exposed to numerous security attacks, for instance,Distributed Denial-of-Service (DDoS) and Denial-of-Service (DoS).This attack could incur a significant loss for IIoT service and smart environmental application in the IIoT system.As a result, it becomes inevitable to protect the IIoT system from all types of cyber attacks [1].Intrusion Detection System (IDS) is a security system that mostly functions in the network layers of IIoT system.When IDS is installed in an IIoT network, it should be capable of analysing the data packets and produce the responses on a real-time basis.Furthermore, it should analyse the data packets present in distinct layers of IIoT system using distinct protocol stacks and adapt to different techniques used in IIoT framework [7].An IDS i.e., developed for IIoT-based smart environment, must function under stringent conditions in terms of fast response, high volume data processing, and low-processing capability.Hence, traditional IDS mayn’t be adequate to meet the needs of IIoT environment.IIoT security is a serious and continuous problem that needs to be addressed with priority; therefore, it becomes an inevitable need to have a clear and current understanding of security susceptibilities in IIoT systems and the growth of respective mitigation methods.Though blockchain and artificial intelligence are known to exhibit numerous advantages, they still have their own share of disadvantages too.Blockchain has a problem with scalability, energy consumption, privacy, efficiency, and security whereas AI overcomes a few problems like effectiveness and interpretability.Though both techniques are distinct areas of research, it could be associated with one another so as to have the advantage of natural incorporation.These two techniques have common requirements for security, trust, and data analysis due to which one could authorize the other.

    The current research paper develops a new Privacy Preserving Blockchain with Deep Learning model for Industrial IoT (PPBDL-IIoT) on 6G environment in order to identify the presence of intrusions in network.PPBDL-IIoT technique incorporates Chaos Game Optimization (CGO) with Bidirectional Gated Recurrent Neural Network (BiGRNN) technique for detection and classification of intrusions in the network.Additionally, CGO algorithm is applied in fine tuning the hyperparameters of BiGRNN model too.Furthermore, Blockchain Enabled Integrity Check (BEIC) scheme is introduced to avoid the misrouting attacks that tamper the OpenFlow rules of SDN-based IIoT system.For validating the supremacy of the proposed PPBDL-IIoT technique, a series of simulations was executed on ICSCA dataset and the outcomes were discussed under distinct parameters.

    The paper is structured in the following.Section 2 shows a comprehensive review of security based solutions for IIoT environment and Section 3 elaborates the detailed processes involved in the proposed model.Next, Section 4 shows a brief experimental validation process and Section 5 draws the conclusion.

    2 Related Works

    Derhab et al.[8] proposed a security framework in which SDN and blockchain techniques are incorporated.The presented method was made up of an IDS scheme that integrates RSL and KNN algorithms to defend against the forged command and BICS so that the misrouting attacks can be avoided.Rahmadika et al.[9] proposed a smart system incorporating 5G edge network, FL,and blockchain technology so as to generate an effective and secured architecture for performing transaction.FL allows the UE to train AI paradigm without revealing the beneficial data of UEs to either model or the public provider.The recorded transaction could not be maliciously changed or in other words, they remain the same.Further, the researcher also proposed few dynamic verification protocols to UE for the purpose of interacting with different BSs.In addition to this, blockchain was also employed as a rewarding method in FL to allow the computation offloading in wireless networks.

    Qu et al.[10] proposed an architecture that is composed of BCS, layers, and intersect.In this novel structure, every BCS is configured using blockchain technique.The authors described about the credibility authentication approach in this study and also showed how it can authenticate the procedure.Further, security analyses was conducted to test the efficiency based on authentication,response time, and storage efficacy.Alkadi et al.[11] proposed the development of a DBF to provide security-based distributed IDS and security-based blockchain with the help of an intelligent contract in IoT network.IDS scheme makes use of BiLSTM DL method to handle consecutive network data and is measured by BoT-IoT and UNSW-NB15 datasets.Both privacy-based blockchain and smart contracts method were a result of Ethereum library and it aims at providing privacy to the distributed IDS.

    Ferrag et al.[12] proposed a new DL and blockchain-based energy structure i.e., enabled DeepCoin for Smart Networks.DeepCoin structure employs two systems such as a DL based and blockchain based systems.The blockchain-based system includes five stages such as agreement, setup,viewchange, consensus making phases, and creating a block.It integrates a new consistent peer-to-peer energy scheme i.e., depending on real-time Byzantine fault-tolerance approach.Further, the method attains a high throughput.In Guha Roy et al.[13], a decentralized security method was proposed with the incorporation of an SDN using blockchain for IoT in fog computing and mobile edges.Here, SDN repeatedly analyses and monitors the traffic system to render an attack detection method.Blockchain was employed in this study to overcome the problems faced in the present model.This was done so, by sending the decentralized attack detection system that identifies attacks in fog and decreases the same in edge node.

    Skwarek [14] proposed an approach for IoT devices to attain industry-grade consistency in data transfer from WSN to production system via blockchain technology.It is possible to obtain improved reliability and security of the acquired data within sensor networks on application levels.Hence, a lightweight, higher-level transmission protocol, depending on blockchain technology, has been developed in this study.Blockchain mechanisms could protect the wireless transmission of IoT in a scalable, lightweight model.Lee et al.[15] proposed a novel method that leverages the distributed watchdog using blockchain system in defending the software supply chain.Furthermore,the researchers introduced a set of thorough specifications to determine the behavior of system in a clear manner and employ module checking.

    3 The Proposed Model

    The current study presented a new framework for ICS that allows network virtualization by transferring the control layer to cloud so that a centralized management is enabled.The proposed PPBDL-IIoT technique is a two-stage process namely intrusion detection and BEIC.For a long known time, WAN is generally considered as complex and costly only.The proposed framework decreases the network cost by providing zero-touch placement, viz., it is not necessary to construct the network device by plugging it in.Rather, the devices are constructed from SDN controller.The presented framework is made up of the modules discussed herewith.

    Private cloud: It hosts every component that provides a central control for ICS alike VMs such as SDN controller, SCADA, and DCS servers.

    IP network: Rather than employing a devoted WAN to replace the ICS, public internet connections are used among SDN and distinct substations.Every device is validated and end-to-end encryptions are performed through networks.

    SDN controller: It can be considered as an application that handles the flow control with the help of protocols like OpenFlow.In this setup, the switches are used for the transmission of data packets.Open Flow protocol is a southbound interface amongst the forwarding and controller mechanisms like switches.Here, the northbound interface assumes the communication amongst the applications and controller.

    Virtual Switch: It can be an application that interconnects different VMs of different/same hypervisor(s).Furthermore, it interconnects the VMs with other physical switch.

    3.1 Design of Intrusion Detection Technique

    In intrusion detection process, CGO-BiGRNN technique is applied to determine the occurrence of intrusions in the network.Multilayer RNNs are amongst one of the generally utilized techniques in DL.These kinds of networks are capable of memorizing the realized data.Further, RNN is a powerful technique for sequential data (time series) analysis.It utilizes the preceding output to predict the next output.These loops i.e., hidden neurons allow one to save the preceding input data.But, the model is forecasted based on the future output.The hidden layer outcome is retransmitted forttimes to the hidden layer [16].Then the output becomes complete and the preceding data is retained for too long.Eventually, the errors are returned back for updating the weight.During this case, four obtainable RNN structures are utilized.

    Simple RNN is fundamentally a group of general NNs which have the same task to accomplish i.e., transfer the message from one node to another.Specifically, this is a memory network that saves knowledge on the data server.However, its memory is short-term and could not continue as a longterm time series.A simple recurrent network is the one internal memory-ht- calculated through Eq.(1):

    whereg() refers to activation function,UandWdenote the variables in the weight matrices of h layer,brepresents the bias andXsignifies the input vector.

    GRU is an easy LSTM, proposed by Cho et al.[17].The difference with LSTM is that GRU combines both input as well as output gates and upgrades the same.The output value ofhtin GRU is calculated as follows.

    whererimplies reset gate, andzrepresents the upgrade gate.Reset gate represents the integration of novel input with earlier memory.An upgrade gate also represents the retention of the preceding memory.When the update gate is one, the preceding memory gets completely preserved.When it can be zero, the preceding memory goes totally forgotten.There is a forget gate in LSTM which manually defines the preceding memory as continuous.But in case of GRU, every preceding memory is either preserved or entirely forgotten.Though GRU can be compared with LSTMs in terms of efficiency, it requires low memory.

    Bi-directional framework methods have the capability of learning data from preceding and following data, if the current data is managed well.Fig.2 illustrates the framework of GRU technique.BiGRNN method has been defined as a dependent method that relies upon the state of two unidirectional and oppositely-placed GRUs [18].This allows the data in both future as well as past to impact the current state.bi-GRU is determined as follows.

    Figure 2: GRU structure

    In order to fine tune the hyperparameters of BiGRNN technique, CGO algorithm is applied.The key concept of CGO algorithm is depending on chaos theory, in which the fractals are configured by using chaos game concept.The theory of chaos game is used as the key principle behind CGO algorithm method.A quantity of solution candidates (X), represent the seed, is considered as CGO approach.Every (X) candidate solution includes a few tuneable variables (X) which represent the location of the seed.The triangle of Sierpinski is assumed as the exploration fields on the lookout for solution.The mathematical form for the above scenario is represented herewith.

    In whichndenotes the population size of exploration, whereasdindicates dimensional problem.The seed is randomly initialized by [19]

    where(0) represent the first population,andindicates min.and max.values ofjthdimension, and rand represent an arbitrary value in the range of zero and one.To finalise the structure of a triangle, the arithmetical method makes distinct seeds in their upper and lower bounds.A temporary triangle using three seeds are shown below.

    ?The Global Best (GB) established until now;

    ?The Mean Group (MGi) is the mean value of arbitrary seed;

    ?The ith solution candidate (Xj).

    The vertices of the triangle areXj.GB, and MGi A temporary triangle is created for every first seed so that it can produce novel seed to finish a novel triangle.The ith iteration comprises 3 corners of a Sierpinski triangle and the seed attained in prior iterations.For novel seed, the temporary triangle is employed and the dice is again rolled.The seeds inXiare transferred to GB or MGi based on the resultant colour.It can be modelled by an arbitrary integer value.It creates two integers i.e., zero/one,that allows the selection of green or red faces.As per the chaos game, seed motion must be minimum.in this regard, some of the factorials generated, are employed.The determined procedure for the initial seed can be mathematically expressed as

    whereasXidenotesithcandidate,αisignifies an arbitrary factorial that restricts the motion of seeds,and βjand γjdenotes arbitrary integers of zero/one to represent the probabilities of rolling some dice.The dice of three red and blue faces are employed to the succeeding seeds (GB).The dice is rolled while the GB is moved toXiorMGi, based on the resultant colour.The seed move to theXi, once a blue face is attained; when red face is attained, the seeds move toMGi.The 2nd seeds go toward a point on linking lines amongXi&MGi.It can be expressed mathematically by

    Based on the 3rd seeds, the dice is rolled.As per the resultant colour, the seed move toXiorGB,that generates 2 integrals i.e., zero and one.As per the certain factor, the seed travel to the connected lines amongXi& GB can be formulated by

    Also, alternative technique is employed for creating the 4th seed to present a mutation step in the position update of seeds.The position is modified according to specific changes in the chosen decision.The arithmetical expression for the abovementioned procedure can be given in the following.

    whereaskdenotes an arbitrary integer among 1 &d, as well asRsignifies the range of arbitrary value between zero and one.

    The four distinct formulations for αithat control the movement limitation of the grain are proposed herewith to change and monitor the exploitation and exploration rates of CGO method.

    where Rand denotes a random number in the range of zero and one, as well asδ&ε represent arbitrary factors among [0, 1].

    The fitness of novel candidate is related to the existing one and higher fitness are saved.In the meantime, the candidate with lower fitness gets rejected.

    3.2 Design of BEIC Scheme

    Blockchain is a kind of chain formation that integrates data blocks in a sequential order.It is non-forgeable, tamper-proof distributed ledger that is assured by cryptography.Each blockchain participant maintains the data for blockchain node.Hence, every data in blockchain is open and transparent.When the data is published, it is retained permanently and could not be tampered with.Tamper-proof and open authentication features of the blockchain enable one to perform as a trusted 3rd parties for addressing the concern of a user in Cloud Computing (CC) platform.Each result could be released to blockchain for verification and can be retained by every user in the blockchain.Thus,when blockchain is incorporated in CC environment, it benefits by resolving the drawbacks of CC platform and could increase the efficiency of services to users with high data security.

    Being a significant model of Bitcoin and a decentralized database, blockchain serves as a primary framework of Bitcoin simultaneously.A blockchain is a sequence of data blocks that are produced with the help of cryptography.Fig.3 illustrates the block diagram of blockchain technology.Every block has data about bitcoin network transaction which is employed to verify the validity of the data and produce succeeding blocks.In other terms, blockchain is a chained data structure that successively integrates data blocks based on time series.Being a distributed ledger, it could not be forged through or tampered with cryptography.Blockchain technologies, in general, use blockchain data structure to store and verify the information.Further, it employs distributed node consensus to update and generate the information.In addition, it also employs cryptography for the purpose of data access and transmission.It also employs smart contract, made up of automatic script code, for programming and data manipulation purposes.Blockchain contains few features such as immutability,decentralization, collective maintenance, traceability, transparency, and openness.This characteristic ensures the transparency and honesty of blockchain and lays the basis for blockchain to create trust.Blockchain-rich application scenario depends on the fact that blockchain could resolve data asymmetry and attain concerted action and collaborative trust amongst different subjects.

    Figure 3: Framework of blockchain

    Before determining the security solutions, some assumptions are to be made.

    ?Consider that ICS isn’t compromised (viz., free from malicious codes prior to the installation of BEIC scheme).Or else, forged rule could be deliberated as legitimate.

    ?BEIC scheme focuses only on southbound transmission.Consider that the northbound communications are secure between DCS and SCADA servers and IDSs from one side as well as SDN controllers from another side.

    ?Consider that SDN controllers are placed in a private cloud and are available from an individual host using access control and authentication method.

    Blockchain [20] is an essential component in the proposed integrity checking system.The fundamental concept is to offer a solution in which every flow rule that can be produced from the controller,is saved in immutable and verifiable databases.Blockchain is a series of blocks i.e., connected with hash values.In blockchain networks, every user contains: single public key signifies its exclusive address and single private key to sign the blockchain transactions.After authenticating the transmission blocks, the transaction is processed and then added to the blockchain.After recording, the data in the provided blocks could not be altered without changing every succeeding block.Additionally, the data exist in several hosts in a gradual manner.So, any modifications will be denied by hosts of the peers.In this study, a private (or permission) blockchain is presented.Therefore, consensus methods like Proof of Work are not essential herewith.This blockchain is made up of SDN firewall and controller.SDN controllers create the block and share it with firewall through blockchain.The initial nodes have full permission, viz., to send, read, and write, whereas the firewall could only receive and read.The BEIC systems are brought in the succeeding series.

    ?Once the requests are received from northbound applications, the SDN controllers are developed to transmit the respective flow rules to vSwitches.In this model, SDN controllers are also considered as members of a blockchain.It hashes the flow rule and places it in block i.e.,dispersed to another node of the blockchain.The SDN controllers are the node in blockchain that had the authority to construct a block whereas the remaining nodes could only read the blockchain.

    ?Once the flow rule reaches vSwitch nodes, save the rule in record files and latter upgrades its flow table.

    ?The firewall gathers the vSwitch log and access the BlockChain for obtaining the flow rules,transmitted through the controller.

    ?When the firewall detects the 2 rules that are present in blockchain and vSwitch, it informs the management to take a proper counter measure for fixing this mismatch.

    4 Performance Validation

    The present subsection validates the efficiency of the proposed PPBDL-IIoT technique against benchmark dataset [21].In this study, binary and multi-class datasets were considered for simulation purposes.The dataset has 128 attributes in total.Besides, the binary class dataset has two classes of instances namely,Natural and Attack classes.The binary class dataset has 15 sub-datasets out of which only five sub-datasets are used in the experimentation procedure.Sub-dataset 1 comprises of 1100 instances under natural class and 3866 instances under attack class.Besides, sub-dataset 2 has 1544 instances under natural class and 3525 instances under attack class.Similarly, sub-dataset 3 comprises of 1604 instances under natural class and 3811 instances under attack class.Likewise, sub-dataset 4 comprises of 1800 instances under natural class and 3402 instances under attack class.Lastly, subdataset 5 comprises of 1481 instances under natural class and 3680 instances under attack class.On the other hand, multi-class dataset includes 15 sub-datasets under three categories namely, No event,Natural, and Attack.

    Fig.4 shows the confusion matrix generated by PPBDL-IIoT technique under five sub-datasets present in the binary dataset.The proposed PPBDL-IIoT technique classified data-1 with 1067 instances under natural class and 3834 instances under attack class.Further, PPBDL-IIoT approach classified data-2 with 1488 instances under natural class and 3793 instances under attack class.Besides,PPBDL-IIoT manner classified data-3 with 1541 instances under natural class and 3764 instances under attack class.Additionally, PPBDL-IIoT method has classified data-4 with 1737 instances under natural class and 3350 instances under attack class.Lastly, PPBDL-IIoT algorithm classified data-5 with 1429 instances under natural class and 3642 instances under attack class.

    Figure 4: Confusion matrix analysis of PPBDL-IIoT model on binary dataset

    Tab.1 provides the classification results of PPBDL-IIoT technique under different sub-datasets of binary class datasets.The results demonstrate that PPBDL-IIoT technique gained effectual outcomes on the applied sub-datasets.For instance, on the applied data-1, PPBDL-IIoT technique achieved a high precision of 0.971, recall of 0.970, specificity of 0.992, accuracy of 0.987, and F-score of 0.970.Moreover, on the applied data-3, the proposed PPBDL-IIoT method gained a high precision of 0.970,recall of 0.961, specificity of 0.988, accuracy of 0.980, and F-score of 0.966.Furthermore, on the applied data-5, PPBDL-IIoT algorithm accomplished a superior precision of 0.974, recall of 0.965,specificity of 0.990, accuracy of 0.983, and F-score of 0.970.

    Fig.5 demonstrates the results from ROC analysis of PPBDL-IIoT method on the applied binary class dataset.The figure portrays that the proposed PPBDL-IIoT technique produced a high ROC of 99.0933.

    Table 1: Results of the analysis of the proposed PPBDL-IIoT model for binary class dataset

    Figure 5: ROC analysis of PPBDL-IIoT model on binary class dataset

    Fig.6 showcases the results of confusion matrices produced by PPBDL-IIoT technique on the applied multi-class dataset.The proposed PPBDL-IIoT technique classified a set of 162 instances into No events class, 3381 instances into Attack class, and 770 instances into Natural class.Meanwhile,PPBDL-IIoT approach classified a set of 312 instances into No events class, 3063 instances into Attack class, and 1038 instances into Natural class.Eventually, the proposed PPBDL-IIoT method classified a set of 338 instances into No events class, 3337 instances into Attack class, and 1080 instances into Natural class.Concurrently,PPBDL-IIoT method classified a set of391 instances into No events class,2956 instances into Attack class, and 1182 instances into Natural class.Simultaneously, the proposed PPBDL-IIoT algorithm classified a set of 253 instances into No events class, 3224 instances into Attack class, and 1035 instances into Natural class.

    Figure 6: Confusion matrix analysis of PPBDL-IIoT model on multi-class dataset

    Tab.2 shows the classification results attained by PPBDL-IIoT approach on the applied multiclass dataset.The proposed PPBDL-IIoT technique obtained the maximum classification performance on the applied sub-datasets.For sample, with data-1, the PPBDL-IIoT technique reached a recall of 0.889, precision of 0.750, specificity of 0.922, accuracy of 0.926, and F-score of 0.761.Likewise, with data-2, PPBDL-IIoT method gained a precision of 0.789, accuracy of 0.927, recall of 0.905, specificity of 0.953, and F-score of 0.833.Similarly, with data-3, PPBDL-IIoT approach obtained a recall of 0.907, precision of 0.828, accuracy of 0.931, specificity of 0.935, and F-score of 0.861.In line with this, for data-4, PPBDL-IIoT algorithm obtained a accuracy of 0.927, recall of 0.905, precision of 0.820, specificity of 0.934, and F-score of 0.839.In line with this, for data-5, the proposed PPBDL-IIoT approach achieved a recall of 0.916, precision of 0.856, accuracy of 0.955,specificity of 0.946, and F-score of 0.873.

    Table 2: Results analysis of proposed PPBDL-IIoT model on multiclass dataset

    Table 2: Continued

    Fig.7 exhibits the result of ROC analysis accomplished by PPBDL-IIoT methodology on the employed multi class datasets.The figure exhibits that the proposed PPBDL-IIoT method produced a high ROC of 97.1620.

    Figure 7: ROC analysis of PPBDL-IIoT model on multi-class dataset

    For ensuring the enhanced classification performances of PPBDL-IIoT process, a thorough comparison study was conducted and the results are shown in Tab.3.Fig.8 illustrates the results of binary class outcome analysis accomplished by PPBDL-IIoT technique against existing techniques.The figure illustrates that NB-K, BN, and LSVM techniques achieved low performance with its accuracy values being 75.7%, 77.9%, and 78.3% respectively.Along with that, DT and Bagging methods showcased slightly enhanced performance with accuracy values being 92.9% and 92.9%respectively.Moreover,AdaBoostM1,KNN, and RF techniques accomplished moderately reasonable outcomes with its accuracy values being94.1%,95.5%,and96%.However,the proposed PPBDL-IIoT technique produced superior outcomes with a high accuracy of 98.2%.

    Table 3: Results analysis of existing with proposed model with respect to accuracy

    Figure 8: Accuracy analysis of PPBDL-IIoT model on binary class dataset

    Fig.9 depicts the results of multi-class outcome analysis of the proposed PPBDL-IIoT approach against existing methods.The figure demonstrates that NB-K, AdaBoostM1, LSVM, and BN methods achieved less performance with its accuracy values being 19.7%, 29.4%, 30.4%, and 74.6%correspondingly.Likewise, RF and bagging techniques outperformed other techniques and achieved the same accuracy of 79.6%.Followed by,DT and KNN methods accomplished moderately reasonable outcomes with accuracy values such as 83.2% and 87.7%.Eventually, the proposed PPBDL-IIoT method produced a high accuracy of 91.5%.

    Figure 9: Accuracy analysis of PPBDL-IIoT method on multiclass dataset

    From the abovementioned figures and tables and respective discussion, it is clear that the proposed PPBDL-IIoT methodology is an appropriate tool to provide security in 6G-enabled IIoT environment.

    5 Conclusion

    In current study, a new PPBDL-IIoT technique has been proposed for the detection of intrusions in 6G-enabled IIoT environment.The proposed PPBDL-IIoT technique encompasses a two-stage process namely, intrusion detection and BEIC.DL-based BiGRNN technique with CGO-based hyperparameter optimization is used to detect and classify the intrusions in network.CGO algorithm is applied to optimally adjust the learning rate, epoch count, and weight decay so as to improve the intrusion detection performance of BiGRNN model up to a considerable manner.Moreover, BEIC technique is designed to circumvent the misrouting attacks that tamper the OpenFlow rules of SDN based IIoT system.For ensuring the supremacy of PPBDL-IIoT technique, a series of simulations was executed on ICSCA dataset.The experimental results highlight the supremacy of PPBDL-IIoT technique over an advanced technique with the higher accuracy of 91.50%.In future, the design of PPBDL-IIoT methodology can be extended further to performfault diagnosis and power management systems in blockchain-enabled IIoT environment.

    Acknowledgement:The authors would like to acknowledge the support of Prince Sultan University,Riyadh, Saudi Arabia for partially supporting this project to paying the Article Processing Charges(APC) of this publication.

    Funding Statement:The authors extend their appreciation to the Deanship of Scientific Research at King Khalid University for funding this work under Grant Number (RGP 2/23/42).www.kku.edu.sa.

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

    亚洲国产精品999| 汤姆久久久久久久影院中文字幕| 久久久精品94久久精品| 丝袜脚勾引网站| 十八禁高潮呻吟视频| 在线精品无人区一区二区三| 天天躁夜夜躁狠狠久久av| av片东京热男人的天堂| 午夜91福利影院| 晚上一个人看的免费电影| 欧美老熟妇乱子伦牲交| 国产成人影院久久av| 777米奇影视久久| 色播在线永久视频| 99久久人妻综合| 男女床上黄色一级片免费看| 国产一区二区三区av在线| 男女边吃奶边做爰视频| 欧美精品亚洲一区二区| 久久久精品免费免费高清| 久久中文字幕一级| 国产成人免费观看mmmm| 亚洲国产欧美网| 欧美日韩亚洲高清精品| 久久鲁丝午夜福利片| 两性夫妻黄色片| 国产片内射在线| 精品福利观看| 赤兔流量卡办理| 交换朋友夫妻互换小说| 婷婷丁香在线五月| netflix在线观看网站| 夜夜骑夜夜射夜夜干| 国产欧美日韩一区二区三区在线| 在线观看一区二区三区激情| 黄频高清免费视频| 大码成人一级视频| 在线 av 中文字幕| 在线观看免费视频网站a站| 男女免费视频国产| 熟女少妇亚洲综合色aaa.| 在现免费观看毛片| 国产主播在线观看一区二区 | 亚洲av电影在线观看一区二区三区| 亚洲国产精品成人久久小说| 另类亚洲欧美激情| 最新的欧美精品一区二区| 一级毛片电影观看| 一边摸一边做爽爽视频免费| 99国产精品免费福利视频| 后天国语完整版免费观看| 久久热在线av| 女人被躁到高潮嗷嗷叫费观| av有码第一页| 爱豆传媒免费全集在线观看| 欧美精品啪啪一区二区三区 | 熟女av电影| 99热网站在线观看| 国产精品一区二区免费欧美 | 久久精品国产亚洲av涩爱| 如日韩欧美国产精品一区二区三区| 久久人人爽av亚洲精品天堂| 亚洲三区欧美一区| 亚洲成av片中文字幕在线观看| 男人操女人黄网站| 久久 成人 亚洲| 男人舔女人的私密视频| 99国产综合亚洲精品| 国产成人91sexporn| 精品国产国语对白av| 丝瓜视频免费看黄片| videosex国产| 久久久久视频综合| 亚洲一区中文字幕在线| av视频免费观看在线观看| a级毛片在线看网站| 黑人猛操日本美女一级片| 久久人人爽人人片av| 国产一卡二卡三卡精品| 波多野结衣一区麻豆| 黄色毛片三级朝国网站| 丝瓜视频免费看黄片| 99精国产麻豆久久婷婷| 国产男人的电影天堂91| 少妇 在线观看| 国产91精品成人一区二区三区 | a级毛片在线看网站| 国产视频首页在线观看| 亚洲精品中文字幕在线视频| 美女脱内裤让男人舔精品视频| 亚洲免费av在线视频| 中文字幕精品免费在线观看视频| 国产av一区二区精品久久| 中国美女看黄片| 中国国产av一级| 亚洲三区欧美一区| 久久久国产精品麻豆| 精品一区二区三区四区五区乱码 | 亚洲中文av在线| 亚洲av成人精品一二三区| 9191精品国产免费久久| 亚洲第一av免费看| 视频在线观看一区二区三区| 中文字幕av电影在线播放| 成年av动漫网址| 国产男女超爽视频在线观看| 国产一区亚洲一区在线观看| 日韩一区二区三区影片| 看十八女毛片水多多多| 大片电影免费在线观看免费| 在线观看www视频免费| 老汉色av国产亚洲站长工具| 欧美黄色淫秽网站| 一二三四在线观看免费中文在| 蜜桃在线观看..| 成年人黄色毛片网站| 日日夜夜操网爽| 老熟女久久久| 天天躁夜夜躁狠狠久久av| 国产一区二区 视频在线| 午夜视频精品福利| 欧美激情高清一区二区三区| 99热网站在线观看| 大片免费播放器 马上看| 999久久久国产精品视频| 国产一区二区三区综合在线观看| 欧美日韩成人在线一区二区| 2018国产大陆天天弄谢| av电影中文网址| 成年av动漫网址| 亚洲视频免费观看视频| 亚洲成色77777| 国产黄频视频在线观看| 成年女人毛片免费观看观看9 | 精品高清国产在线一区| 97精品久久久久久久久久精品| 丝袜喷水一区| 亚洲精品美女久久av网站| 一级,二级,三级黄色视频| 欧美黄色淫秽网站| netflix在线观看网站| 三上悠亚av全集在线观看| 一本一本久久a久久精品综合妖精| 波多野结衣一区麻豆| av国产久精品久网站免费入址| 亚洲第一av免费看| 波多野结衣一区麻豆| av国产久精品久网站免费入址| 男女无遮挡免费网站观看| 这个男人来自地球电影免费观看| 久久精品国产综合久久久| 亚洲第一av免费看| 成人黄色视频免费在线看| 欧美乱码精品一区二区三区| 黄片小视频在线播放| 黄色片一级片一级黄色片| 又黄又粗又硬又大视频| 80岁老熟妇乱子伦牲交| 99热国产这里只有精品6| 一本综合久久免费| 国产91精品成人一区二区三区 | 成年人黄色毛片网站| 王馨瑶露胸无遮挡在线观看| 国产xxxxx性猛交| 久久久久久久久久久久大奶| 大片电影免费在线观看免费| 亚洲av片天天在线观看| 黄色 视频免费看| 考比视频在线观看| 超色免费av| 国产1区2区3区精品| 日本91视频免费播放| 一级片免费观看大全| xxxhd国产人妻xxx| av在线老鸭窝| 如日韩欧美国产精品一区二区三区| 久久 成人 亚洲| 欧美97在线视频| 中文欧美无线码| 亚洲av片天天在线观看| 18禁国产床啪视频网站| 亚洲av男天堂| 晚上一个人看的免费电影| 亚洲欧美精品自产自拍| 精品少妇久久久久久888优播| 国产野战对白在线观看| 一区二区三区精品91| 人人澡人人妻人| 国产日韩欧美视频二区| 国产免费又黄又爽又色| 国产成人精品无人区| 亚洲一码二码三码区别大吗| 亚洲欧美成人综合另类久久久| 午夜日韩欧美国产| 国产不卡av网站在线观看| 精品国产乱码久久久久久小说| 又黄又粗又硬又大视频| 岛国毛片在线播放| 国产精品人妻久久久影院| 超碰97精品在线观看| 叶爱在线成人免费视频播放| 久久精品久久精品一区二区三区| 两个人免费观看高清视频| 美女主播在线视频| 精品卡一卡二卡四卡免费| 涩涩av久久男人的天堂| 国产欧美亚洲国产| 在线观看免费午夜福利视频| a级毛片黄视频| 91精品三级在线观看| av又黄又爽大尺度在线免费看| 欧美日韩黄片免| 国产有黄有色有爽视频| 日本av手机在线免费观看| 好男人电影高清在线观看| 欧美黑人欧美精品刺激| 久久久久久久久久久久大奶| 日本wwww免费看| 亚洲国产中文字幕在线视频| 国产精品久久久久成人av| 韩国精品一区二区三区| 亚洲黑人精品在线| 国产99久久九九免费精品| 亚洲第一av免费看| 极品人妻少妇av视频| 亚洲三区欧美一区| 亚洲七黄色美女视频| 国产高清国产精品国产三级| 涩涩av久久男人的天堂| 国产精品成人在线| 国产精品久久久久久精品电影小说| 亚洲精品美女久久av网站| 国产男女内射视频| 极品少妇高潮喷水抽搐| 日韩伦理黄色片| 亚洲精品日本国产第一区| 国产欧美亚洲国产| 精品第一国产精品| 国产精品香港三级国产av潘金莲 | 国产在线一区二区三区精| 免费看十八禁软件| a级毛片黄视频| 90打野战视频偷拍视频| 黄色a级毛片大全视频| 青春草视频在线免费观看| 视频区图区小说| 97精品久久久久久久久久精品| 国产精品av久久久久免费| 国产精品一区二区免费欧美 | 伊人亚洲综合成人网| 国产成人免费无遮挡视频| 久久精品aⅴ一区二区三区四区| 国产极品粉嫩免费观看在线| 另类精品久久| 最新的欧美精品一区二区| 无限看片的www在线观看| www.999成人在线观看| www.精华液| 日本五十路高清| 国产精品成人在线| 下体分泌物呈黄色| 精品国产一区二区三区久久久樱花| 黄片播放在线免费| 1024香蕉在线观看| 国产男女内射视频| 免费少妇av软件| 亚洲人成电影免费在线| 国产精品久久久久久精品古装| 亚洲欧美精品自产自拍| 国产在视频线精品| 90打野战视频偷拍视频| 高清不卡的av网站| 中文字幕人妻丝袜一区二区| 亚洲欧洲精品一区二区精品久久久| 91成人精品电影| 日韩伦理黄色片| 亚洲成人免费电影在线观看 | 另类精品久久| 欧美日韩成人在线一区二区| 一边摸一边抽搐一进一出视频| 岛国毛片在线播放| 黄色a级毛片大全视频| 人妻 亚洲 视频| 一级毛片 在线播放| 黄色 视频免费看| 视频在线观看一区二区三区| 欧美精品啪啪一区二区三区 | 久久精品成人免费网站| 亚洲欧美清纯卡通| 国产熟女欧美一区二区| 精品国产国语对白av| 久久鲁丝午夜福利片| 青春草亚洲视频在线观看| 国产xxxxx性猛交| 99热网站在线观看| 97在线人人人人妻| 99热网站在线观看| 午夜精品国产一区二区电影| 免费观看a级毛片全部| 久热爱精品视频在线9| 免费观看av网站的网址| 一区二区三区乱码不卡18| 捣出白浆h1v1| 精品一区二区三区av网在线观看 | 久热爱精品视频在线9| 亚洲七黄色美女视频| 丰满少妇做爰视频| 夫妻性生交免费视频一级片| xxxhd国产人妻xxx| 国产一区二区三区综合在线观看| 丰满迷人的少妇在线观看| 成人午夜精彩视频在线观看| 美女扒开内裤让男人捅视频| 精品国产一区二区三区四区第35| 亚洲国产欧美在线一区| 国产成人系列免费观看| 国产精品亚洲av一区麻豆| 亚洲专区中文字幕在线| 亚洲人成电影免费在线| 天天躁夜夜躁狠狠躁躁| 亚洲欧美成人综合另类久久久| 免费在线观看影片大全网站 | 曰老女人黄片| av网站免费在线观看视频| 久久精品亚洲熟妇少妇任你| 在线看a的网站| 午夜91福利影院| 看十八女毛片水多多多| 大片免费播放器 马上看| 九草在线视频观看| 国产成人av激情在线播放| 久久毛片免费看一区二区三区| 欧美日韩亚洲国产一区二区在线观看 | 国产又爽黄色视频| 又粗又硬又长又爽又黄的视频| 极品少妇高潮喷水抽搐| 首页视频小说图片口味搜索 | 丰满人妻熟妇乱又伦精品不卡| 美女中出高潮动态图| 首页视频小说图片口味搜索 | 丰满人妻熟妇乱又伦精品不卡| 国产精品香港三级国产av潘金莲 | 久久精品国产亚洲av涩爱| 黑人欧美特级aaaaaa片| 亚洲成av片中文字幕在线观看| 久久国产精品人妻蜜桃| 99国产精品免费福利视频| 丰满迷人的少妇在线观看| 国产不卡av网站在线观看| 丰满人妻熟妇乱又伦精品不卡| 国产主播在线观看一区二区 | 亚洲国产日韩一区二区| 婷婷丁香在线五月| 国产欧美日韩一区二区三区在线| 亚洲黑人精品在线| 丝袜美足系列| 国产不卡av网站在线观看| 国产成人欧美在线观看 | 黄色一级大片看看| 久久久国产一区二区| 黑人巨大精品欧美一区二区蜜桃| 免费高清在线观看视频在线观看| 高清欧美精品videossex| 在线精品无人区一区二区三| 午夜久久久在线观看| 人人妻,人人澡人人爽秒播 | 国产免费视频播放在线视频| 免费少妇av软件| 亚洲欧洲国产日韩| 欧美性长视频在线观看| 久久av网站| 十分钟在线观看高清视频www| 99热全是精品| 大香蕉久久网| 国产男女内射视频| 99久久人妻综合| 日韩大码丰满熟妇| 91九色精品人成在线观看| 深夜精品福利| 午夜激情久久久久久久| 亚洲国产欧美日韩在线播放| 欧美在线一区亚洲| 精品高清国产在线一区| 国产又色又爽无遮挡免| 操美女的视频在线观看| 欧美成人午夜精品| 久久人妻福利社区极品人妻图片 | 大型av网站在线播放| a级毛片黄视频| 一级片免费观看大全| 亚洲五月色婷婷综合| 久久精品成人免费网站| 欧美日韩黄片免| 观看av在线不卡| 999久久久国产精品视频| 一区二区av电影网| av在线app专区| 女人精品久久久久毛片| 久久狼人影院| 欧美大码av| 亚洲成av片中文字幕在线观看| 久久青草综合色| 女人久久www免费人成看片| 91精品伊人久久大香线蕉| 成人亚洲欧美一区二区av| 91九色精品人成在线观看| 黄频高清免费视频| 亚洲精品一卡2卡三卡4卡5卡 | 午夜老司机福利片| av在线老鸭窝| 国产精品欧美亚洲77777| 国产不卡av网站在线观看| 91老司机精品| 女人久久www免费人成看片| 久久精品国产综合久久久| 黄频高清免费视频| 一二三四在线观看免费中文在| 嫁个100分男人电影在线观看 | 日韩大片免费观看网站| 日韩精品免费视频一区二区三区| 国产精品 欧美亚洲| 午夜免费男女啪啪视频观看| 免费在线观看日本一区| 波野结衣二区三区在线| 国产老妇伦熟女老妇高清| 久久久久久久久久久久大奶| 妹子高潮喷水视频| 日韩制服骚丝袜av| 亚洲国产看品久久| 搡老乐熟女国产| 欧美人与善性xxx| 国产1区2区3区精品| 久久中文字幕一级| 亚洲欧美精品综合一区二区三区| 成人黄色视频免费在线看| 91字幕亚洲| 亚洲精品乱久久久久久| 午夜免费男女啪啪视频观看| 亚洲,一卡二卡三卡| 大陆偷拍与自拍| 五月天丁香电影| 亚洲一区二区三区欧美精品| 美女视频免费永久观看网站| 精品一区二区三区四区五区乱码 | 菩萨蛮人人尽说江南好唐韦庄| 男女午夜视频在线观看| 欧美另类一区| 黄色视频不卡| 精品久久久精品久久久| 日韩av不卡免费在线播放| 免费看av在线观看网站| 操美女的视频在线观看| 80岁老熟妇乱子伦牲交| 啦啦啦在线免费观看视频4| 亚洲九九香蕉| 国产精品免费视频内射| 欧美日韩福利视频一区二区| 亚洲成人免费av在线播放| 欧美中文综合在线视频| 深夜精品福利| 一级毛片我不卡| 国产成人精品久久二区二区91| 一边摸一边做爽爽视频免费| 亚洲欧美成人综合另类久久久| 搡老岳熟女国产| 国产人伦9x9x在线观看| 久久久国产精品麻豆| 国产精品熟女久久久久浪| 伦理电影免费视频| 中文字幕人妻丝袜一区二区| 亚洲国产最新在线播放| 亚洲国产欧美一区二区综合| 黄色 视频免费看| e午夜精品久久久久久久| 精品一区二区三卡| 天堂8中文在线网| 老司机影院毛片| 下体分泌物呈黄色| 欧美人与性动交α欧美精品济南到| 亚洲图色成人| 国产精品免费大片| 国产一区二区三区综合在线观看| 又黄又粗又硬又大视频| 嫁个100分男人电影在线观看 | 国产精品三级大全| www.999成人在线观看| 一级毛片我不卡| 久久久亚洲精品成人影院| 飞空精品影院首页| av又黄又爽大尺度在线免费看| 91麻豆精品激情在线观看国产 | 久久精品国产a三级三级三级| 成人三级做爰电影| 亚洲男人天堂网一区| 午夜91福利影院| 国产精品 国内视频| 悠悠久久av| 两人在一起打扑克的视频| 在线观看免费视频网站a站| xxxhd国产人妻xxx| 亚洲七黄色美女视频| 国产亚洲一区二区精品| 极品人妻少妇av视频| 欧美黑人精品巨大| 久久 成人 亚洲| 国产福利在线免费观看视频| 国产无遮挡羞羞视频在线观看| av福利片在线| 国产淫语在线视频| 亚洲精品久久午夜乱码| 男女之事视频高清在线观看 | 欧美精品人与动牲交sv欧美| 在线亚洲精品国产二区图片欧美| 国产亚洲欧美精品永久| 黄色视频不卡| 下体分泌物呈黄色| 99热国产这里只有精品6| 男女无遮挡免费网站观看| 免费在线观看日本一区| 国产成人精品久久二区二区91| 999精品在线视频| 各种免费的搞黄视频| 1024香蕉在线观看| 欧美乱码精品一区二区三区| 欧美日韩成人在线一区二区| 在线观看人妻少妇| 搡老乐熟女国产| 成人亚洲欧美一区二区av| 亚洲五月色婷婷综合| 菩萨蛮人人尽说江南好唐韦庄| 国产精品久久久人人做人人爽| 少妇猛男粗大的猛烈进出视频| 亚洲一码二码三码区别大吗| 午夜精品国产一区二区电影| 多毛熟女@视频| 色94色欧美一区二区| 精品福利观看| 日本wwww免费看| 欧美变态另类bdsm刘玥| 男男h啪啪无遮挡| 久久天躁狠狠躁夜夜2o2o | 2018国产大陆天天弄谢| av有码第一页| 久久久久国产精品人妻一区二区| 久久久久精品人妻al黑| 一边摸一边抽搐一进一出视频| av福利片在线| 80岁老熟妇乱子伦牲交| 久久精品国产综合久久久| 日韩大码丰满熟妇| 丝袜喷水一区| 一级毛片女人18水好多 | 欧美日韩黄片免| 精品欧美一区二区三区在线| www.av在线官网国产| a级毛片黄视频| 视频区图区小说| 久久精品熟女亚洲av麻豆精品| 久久久精品94久久精品| 人人妻人人爽人人添夜夜欢视频| 精品熟女少妇八av免费久了| 青青草视频在线视频观看| 久久久精品免费免费高清| 久久影院123| 欧美黄色片欧美黄色片| 久久影院123| 人妻一区二区av| 久久人人爽av亚洲精品天堂| 在线精品无人区一区二区三| 国产99久久九九免费精品| 美女高潮到喷水免费观看| 欧美乱码精品一区二区三区| 亚洲一码二码三码区别大吗| 亚洲精品成人av观看孕妇| 国产黄频视频在线观看| 中文字幕制服av| 成人国语在线视频| 久久久精品免费免费高清| 久久人人爽av亚洲精品天堂| 国产亚洲欧美在线一区二区| 婷婷色综合www| 日韩一本色道免费dvd| 欧美日韩黄片免| 精品国产一区二区久久| 国产亚洲av片在线观看秒播厂| 精品欧美一区二区三区在线| 午夜免费男女啪啪视频观看| 日日夜夜操网爽| 香蕉丝袜av| 亚洲专区国产一区二区| 波多野结衣一区麻豆| 新久久久久国产一级毛片| 人人妻人人添人人爽欧美一区卜| 国产成人精品久久二区二区91| 日韩制服丝袜自拍偷拍| 伦理电影免费视频| 精品免费久久久久久久清纯 | 日本av免费视频播放| 成年动漫av网址| 亚洲av片天天在线观看| 欧美成狂野欧美在线观看| 午夜福利乱码中文字幕| 午夜福利影视在线免费观看| 99香蕉大伊视频| 欧美变态另类bdsm刘玥| 高清欧美精品videossex| 免费在线观看黄色视频的| 热99久久久久精品小说推荐| 不卡av一区二区三区| 一区二区av电影网| 欧美人与善性xxx| 国产精品一区二区精品视频观看| 成人国语在线视频| 美女高潮到喷水免费观看| 天堂俺去俺来也www色官网|