• <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 ?

    Correlation Composition Awareness Model with Pair Collaborative Localization for IoT Authentication and Localization

    2024-05-25 14:41:08KranthiAlluriandGopikrishnan
    Computers Materials&Continua 2024年4期

    Kranthi Alluri and S.Gopikrishnan

    School of Computer Science and Engineering,VIT-AP University,Amaravathi,Andhra Pradesh,522241,India

    ABSTRACT Secure authentication and accurate localization among Internet of Things(IoT)sensors are pivotal for the functionality and integrity of IoT networks.IoT authentication and localization are intricate and symbiotic,impacting both the security and operational functionality of IoT systems.Hence,accurate localization and lightweight authentication on resource-constrained IoT devices pose several challenges.To overcome these challenges,recent approaches have used encryption techniques with well-known key infrastructures.However,these methods are inefficient due to the increasing number of data breaches in their localization approaches.This proposed research efficiently integrates authentication and localization processes in such a way that they complement each other without compromising on security or accuracy.The proposed framework aims to detect active attacks within IoT networks,precisely localize malicious IoT devices participating in these attacks,and establish dynamic implicit authentication mechanisms.This integrated framework proposes a Correlation Composition Awareness (CCA)model,which explores innovative approaches to device correlations,enhancing the accuracy of attack detection and localization.Additionally,this framework introduces the Pair Collaborative Localization (PCL) technique,facilitating precise identification of the exact locations of malicious IoT devices.To address device authentication,a Behavior and Performance Measurement(BPM)scheme is developed,ensuring that only trusted devices gain access to the network.This work has been evaluated across various environments and compared against existing models.The results prove that the proposed methodology attains 96%attack detection accuracy,84%localization accuracy,and 98%device authentication accuracy.

    KEYWORDS Sensor localization;IoT authentication;network security;data accuracy;precise location;access control;security framework

    1 Introduction

    The Internet of Things (IoT) is a network of interconnected electronic and mechanical devices that can communicate autonomously without human intervention.IoT is dependent on connectivity to function,enabling sensors,actuators,and controllers to communicate with one another and the cloud.Connected devices are changing industries,homes,and lives,paving the way for an intelligent,connected future.IoT devices possess diverse internet connections during real-time deployment.The connectivity of IoT devices is impacted by various issues,including security vulnerabilities,reliability concerns,limited scalability,and the need for seamless integration of connectivity technologies to ensure accurate position estimation.Additionally,the deployment of localization devices such as anchors or beacons poses challenges.Sensor localization improves IoT security,dependability,and efficiency.IoT applications can improve authentication,access control,data accuracy,and resource allocation by knowing sensor locations[1].Sensor localization uses many methods and technologies.The Global Positioning System(GPS)is used to accurately measure latitude,longitude,and altitude.GPS works best outdoors and may struggle indoors or in obstructed areas with weak satellite signals[2].

    Network Grouping:IoT authentication groups devices by parameters using network grouping.Effective management and security of large IoT networks require this grouping strategy.Clustering devices help network managers manage authentication,access,and security.IoT devices include sensors,actuators,gateways,and controllers.Authentication and security processes tailored to each device type’s demands and vulnerabilities can be applied by administrators.Another major network grouping factor is device proximity [3].IoT devices are commonly spread between zones.Administrators can implement context-aware access controls and authentication by grouping devices by location.

    Sensor Identification:Each IoT sensor has a unique identity,either a globally unique identifier(GUID)[4]or a device-specific key.This digital fingerprint distinguishes the sensor from other network devices.An IoT sensor’s unique identifier is registered with the network or a central authentication authority during setup or deployment.Registering the sensor with the network verifies its presence and involvement.

    Sensor Authentication:Sensor authentication usually involves a challenge-response system [5] while connecting to the network or other devices.The sensor must correctly answer a cryptographic inquiry like a nonce or random value using its identification credentials.IoT sensor authentication often uses public key cryptography.A sensor may have a private key that matches the network’s public key.

    2 Background and Literature Review

    The fundamentals of sensor discovery,localization algorithms,and IoT authentication schemes employed in our proposed scheme are covered in this section.To distinguish themselves from other devices,sensor devices should have Media Access Control (MAC) addresses or digital certificates[6].During authentication,this identification verifies the device.To authenticate the sensor and IoT platform before data sharing,robust authentication protocols [7],such as mutual authentication,should be used.Role-Based Access Control(RBAC)should be used to design and enforce permissions and access restrictions for each sensor device[8]to ensure they only execute authorized actions and access relevant data.

    2.1 Localization Techniques

    IoT authentication relies on device geolocation[9].Many IoT devices position themselves using GPS or Wi-Fi triangulation [10].The IoT platform can use this data to verify the device’s location before giving access.Virtual “fences”surround geographic locations in geofencing.IoT devices can perform actions or authentication checks while entering or exiting specific zones.Place Bluetooth beacons in recognized places to transmit signals.IoT devices with Bluetooth can detect beacons for proximity-based authentication.Nearby Wi-Fi SSIDs and access point signal intensity are identified.Radio-Frequency Identification (RFID) and Near Field Communication (NFC) offer proximitybased verification.Nearby readers or devices can read IoT RFID or NFC tags.As a summary,the magnetic calibration technique is employed in [11] to attain energy efficiency.However,the incorporation of cloud computing is necessary in this model.The paper[12]introduces convolutional neural networks,highlighting their offline preprocessing advantage.However,it necessitates the exploration of optimising algorithms.The concept of dead reckoning is introduced in[13],highlighting the advantage of magnetic calibration.However,there is still a need to improve its performance in localization.The Time Difference of Arrival (TDOA) technique,as described in [14],is utilised to attain a high level of precision.However,it is necessary to further investigate and employ advanced positioning techniques.The Rivest-Shamir-Adleman cryptosystem technique,proposed in[15],aims to leverage blockchain technology for trust management.However,optimisation is necessary in order to decrease the computational cost.

    2.2 Authentication Schemes

    Device authentication and secure communication in IoT environments depend on authentication schemes[16].Based on IoT deployment requirements and constraints,many approaches and schemes can be used[17].Password-Based Authentication is a simple IoT platform authentication mechanism that uses a username and password[18].However,improper security can make this strategy vulnerable to password-guessing attacks.Certificate-based authentication,which uses X.509 certificates,is safer.A trusted Certificate Authority(CA)issues a unique certificate for each device.Certificates are more secure than passwords and resistant to common attacks.The desired level of security,IoT device capabilities,and use case determine the authentication strategy[19].To strengthen security,many IoT implementations use various authentication mechanisms.The Rivest Shamir Adleman cryptosystem is proposed in[15].Blockchain technology improves wireless sensor network(WSN)security and trust.Optimisation for scalability and computational cost.Based on mobile terminal radio signal quality,the paper[20]locates Narrowband IoT(NB-IoT)nodes using RF fingerprinting and machine learning.The location of NB-IoT nodes using mobile terminal radio signal quality without hardware is proposed in the article.Reference[21]proposes blockchain-based cascade encryption and trust evaluation.Malicious node localization and detection increase network security and service.Consider blockchainbased secure IoT-WSNs with hybrid machine learning for large,safe deployments.Reference [22]proposes machine learning-based correlation analysis for feature selection.Faster Message Queuing Telemetry Transport (MQTT) anomaly detection helps.Advanced correlation analysis can improve research efficiency.The model[23]leverages blockchain for authentication,data exchange,and nonrepudiation.For security and decentralisation,the proposed approach saves legitimate nodes’identities on a consortium blockchain.

    2.3 Attack Detection Techniques

    The federated intrusion detection system,(F-IDS)[24],uses federated learning.This model uses deep,convolutional,and recurrent neural networks as classifiers.Local learning allows devices to benefit from their peers’knowledge by sharing only model updates with an aggregation server to improve detection models while protecting data privacy.When one or more edge nodes are malicious,the model’s robustness can be improved.B-LSTM[25]is a hybrid IDS model that uses Map Reduce,Black Widow Optimised Convolutional-Long Short-Term Memory (BWO-CONV-LSTM),and the Artificial Bee Colony (ABC) algorithm for feature selection.Improving the B-LSTM to adapt to other public intrusion detection datasets and investigate adaptive strategies to detect higher-similarity attack classes like low-frequency Distributed Denial of Service(DDoS)attacks is still required.The paper[26]uses lightweight federated learning.This method uses ambient sensor and wearable device data from mobile user entities and equipment (UEs) to collaboratively and privately learn medical symptoms like COVID-19 with high accuracy.Investigating the seamless integration of asynchronous federated learning and blockchain can address open research issues and improve B5G network security and privacy.FC-IDS [27] is a machine learning-based,two-layer hierarchical intrusion detection method for IoT networks.This two-layer fog-cloud intrusion detection model finds intrusions into IoT networks while working with limited resources.It is better than other methods in terms of accuracy,precision,recall,and F1-score.Our security architecture relies on the CCA model,which introduces unique device correlation methods that greatly increase IoT attack detection and localization.Using coarse-grained IoT device correlations like delay and network quality,our strategy aims to detect active attacks and geolocate hostile IoT devices.Data is used to develop dynamic implicit authentication systems.Our method uses PCL to precisely locate hostile IoT devices and CCA model insights into attack detection.Faster response times and better isolation and mitigation increase IoT security.

    2.4 Requirements and Challenges

    Sensor localization in IoT authentication is difficult.IoT devices can operate in low-network environments [28].Localization algorithms may use offline or delayed processing to circumvent network difficulties.Securing it is another priority.Privacy and the integrity of location data matter.Thus,safe IoT authentication requires robust encryption and authentication.IoT sensors localise using GPS,Wi-Fi,Bluetooth,and RFID.Integrating various sensor data consistently and accurately is tough.Environmental variables like signal interference can also impair sensor data accuracy.Effective localization algorithms must withstand various environmental challenges in diverse environments.Managing and tracking many sensors in large IoT deployments is complex.Location data is crucial,making privacy a priority.IoT authentication must protect user location data and authorised access[29].Automated cars need real-time localization with low latency[30].Meeting these rigorous timing criteria is tough but critical for application safety and efficiency.

    2.5 Problem Statement

    Growing IoT use has enhanced connectivity and convenience.IoT networks face security risks as they grow.Underusing coarse IoT device correlations like delay and link quality to secure networks is troublesome.This problem demands a full security solution that uses these correlations to detect active attacks,locate rogue IoT devices,and establish dynamic implicit authentication.The framework must easily connect to existing IoT networks for practicality and scalability.This study has many goals.First,update the Correlation Composition Awareness model,which uses unique device correlation methods to detect and localise assaults.Second,localise malicious IoT devices with Pair Collaborative Localization.Behaviour and performance measurement authentication are needed to restrict network access to trusted devices.This issue shows that IoT network security requires a coarse device correlation security architecture.

    3 Proposed Identification and Authentication Algorithm

    This section covers the basic system model,identification,authentication,and verification processes for detecting assaults,malicious nodes,and suspicious devices.Fig.1 demonstrates a multilayered beacon node and signal processing system for IoT network attacker detection.Edge servers calculate and notify hostile nodes’positions to cloud servers for network protection from data theft and disturbances.Sensor nodes report signal strength.

    Figure 1: Process of localizing malicious nodes of the proposed model

    3.1 System Model

    Wireless sensor nodes are divided into ten regions and clusters as sink,cluster head,and sensor nodes based on computational capabilities in the network model.Sensor nodes lack location knowledge;thus,beacon and sink nodes find locations and identify malicious nodes with unique identifiers.Organising nodes into clusters,correct position determination,detecting and addressing rogue nodes,and using unique identities for node support are model priorities for efficient performance.The system model establishes the wireless sensor node network structure and assumptions for algorithms 1,2,and 3.The system model separates the network into sink,cluster head,and sensor node areas and clusters.The algorithms improve IoT network performance and security by addressing system model challenges.The CCA-PCL model uses coarse IoT device correlations,targeting latency,and network quality to make attack detection and localization better.Bayesian Sensor Relocation and Traffic Monitoring (BSRTM) [15],Weighted K-Nearest Neighbours (WKNN) [20],Block-Based Sparse Localization (BBSL) [21],Improved Range-Free Detection Algorithm (IRDA) [22],Adaptive Detection of Sybil Nodes (ADSN) [23],Flow-based Intrusion Detection System (F-IDS) [24],Bidirectional Long Short-Term Memory (B-LSTM) [25],Low-Frequency Logistic Model (LFLM)[26],Fuzzy-based Classification Intrusion Detection System (FC-IDS) [27],and High-Performance Boundary-based Feature Learning(HBFL)[28]offer IoT security insights,whereas CCA-PCL uses correlation composition awareness.The proposed methodology is unique in its ability to scale and evaluate the chain of trust using intelligent communication among IoT nodes in a specific area.The security performance of the system is isolated for each area,ensuring that DoS attacks are unable to manipulate and gain simultaneous access to multiple regions.

    3.2 Identification Phase

    The proposed method uses correlation analysis to detect and localise wireless network assaults using Gated Recurrent Units (GRU) machine learning.It addresses wireless channel randomnessrelated latency and link quality correlations.GRU helps the attack detection model,trained on dataset DL with pairings (G;ML),detect attacks with comparable correlations.Delays and link quality are correlated between devices.Attack localization auxiliary model is trained on HL,where labels(ML) represent adversary zones,using the detection model’s outcome.The system has an early exit feature and two auxiliary classifiers for robustness.Gated Recurrent Units overcome wireless channel randomization by using correlation information across IoT devices to identify and localise assaults.

    1.If denotes the count of IoT devices,the correlation of link quality (Q) or packet delay (D)between IoT devicesiandjis computed according to Eq.(1).Here,N{Q,D}and T represent the size of the window.

    For the training of our attack detection model,we generated a training dataset.Where the input features consist of correlations between pairs of devices in terms of delay and link quality.

    2.For training the attack detection model,a data set was constructed incorporating delay and link quality{(G,YD|G=whereYDrepresents the results from the detection model.Additionally,

    to train the auxiliary attack localization model,a separate dataset was generated leveraging the outcomes obtained from the detection model.Algorithm 1 discovers IoT nodes collaboratively.Start model optimisation by initialising training epochs and stochastic gradient descent.A training loop updates the model with paired data and calculates losses.The final model predicts dataset node IDs.Algorithms train models to detect IoT nodes in networks.

    3.3 Authentication Phase

    Continuous device authentication in IoT networks reduces battery life due to communication and computational issues.CCA-PCL has a Device Anomaly Detection module to address this.This module detects malicious IoT devices by analysing CPU and memory consumption.Based on Pair Collaborative Localization results,CCA-PCL dynamically narrows suspect devices,implicitly authenticating them and minimising system overhead for longer device life spans.It optimises authentication by targeting threats and exploiting performance metrics to improve accuracy and reduce network load.The algorithm detects anomalies in behavior and performance measurement to protect IoT devices by comparing data points to a model.Designing dynamic and adaptive implicit authentication solutions for IoT systems to address changing threats and network conditions while protecting user privacy and system integrity is difficult.Continuous authentication tracking of user and device activities is essential.With anomaly detection and machine learning to generate behavioral profiles,systems can dynamically adjust authentication criteria and identify questionable activities to trigger re-authentication.To minimise CCA model and PCL strategy effects from network heterogeneity and device variability,the proposed security architecture must be thoroughly reviewed.The CCA model adapts to IoT devices and networks via adaptive learning and feature engineering for robustness.

    3.4 Verification Phase

    Validating and confirming network transactions is called verification.To ensure transaction accuracy and integrity,many peers run the verification algorithm.Due to the many computations and communication overhead,this verification procedure can diminish operational efficiency and energy waste.The verification step authenticates people,devices,and data using various authentication protocols and cryptographic algorithms.A chaos-based privacy-preserving cryptographic algorithm and MAC can safeguard smart home data communications.Specialised smart contracts and authorization mechanisms can also help implement blockchain authentication and access control.Using lightweight anti-malware solutions to protect IoT devices from malware and developing lightweight cryptographic algorithms and efficient key management schemes to secure data while allowing authorised access can reduce computational overhead and speed up verification.The Algorithm 2 helps grid devices locate themselves like a GPS.Data is collected from reliable“anchor”points first.It then analyses patterns in paired data points,adjusts internal data representations,and minimises inconsistencies to improve its grid comprehension.This optimization process,fueled by“stochastic gradient descent,”produces a revised map that lets each device calculate its position depending on its neighbors.

    3.5 Attack Detection Phase

    The correlation composition-aware attack detection algorithm (Algorithm 3) and the performance algorithm hunt for hidden attackers in a network.It challenges target devices,analyses their responses,and uses extracted features to train a model that identifies malicious patterns.Suspicious devices are flagged with confidence scores,helping system defenders prioritise and neutralise potential threats.Real-world testing and validation of the proposed security framework across IoT domains are essential to assessing its flexibility and efficacy.The validation procedure will include many evaluations to verify robustness in various environments and attack vectors.First,we will run lab tests to simulate different assault situations and environmental conditions.This lets us assess the framework’s attack detection accuracy,malicious device localization precision,and dynamic implicit authentication effectiveness.By integrating privacy-enhancing features alongside powerful security procedures,the proposed IoT sensor localization architecture addresses changing data protection legislation and standards.To comply,the CCA model prioritises anonymized data aggregation,reducing sensitive information while using correlations for attack detection and location.PCL uses encrypted communication channels and encryption to locate rogue IoT devices,enhancing privacy.The BPM device authentication approach follows privacy-by-design principles to rigorously validate trusted devices without compromising user data.

    4 Result Evaluation

    Our method uses trust evaluation to discover rogue sensor nodes in the network.To prevent network intrusion,we added authentication.A trust value is calculated for each beacon node to determine its reliability and security.The model was tested using a large-scale network simulation model.This research needed several implementations of state-of-the-art current models to compare the proposed model’s performance to existing models.Fig.2 shows the simulation model.For the two main aims of this work,the suggested model was compared against two sets of similar models.In the initial phase,the proposed model’s attack detection performance is assessed using metrics like accuracy,F1-score,precision score,and recall score,and the results are compared to F-IDS[24],B-LSTM [25],LFLM [26],FC-IDS [27],and HBFL [28].The proposed model’s localization performance is assessed in the second phase.The proposed model was compared to BSRTM [15],WKNN [20],BBSL [21],IRDA [22],ADSN [23],and the proposed CCA-PCL model on accuracy,F1-score,precision score,and other parameters.In conclusion,the suggested localization model is evaluated for transaction delay,energy utilisation,throughput,and honesty.Comparing the suggested model to localization analysis models confirmed its improvement.

    Figure 2: Simulation model

    4.1 Simulation Model

    Simulation findings from numerous scenarios validate the suggested system model.Our models’network planning and simulation were done in MATLAB R2021a on a workstation with an Intel(R)Xeon(R)Silver 4214 CPU@2.20 GHz 2.19 GHz(2 processors),128 GB of RAM(128 GB useable),64-bit Windows 10,and an x64-based processor.Python was used for data analysis and processing.Table 1 summarizes network deployment and extensive security analysis simulation setup parameters.We used Solidity to create smart contracts and consensus processes in addition to Python and MATLAB.

    Table 1: Simulation setup for proposed network model

    4.2 Performance Metrics

    The proposed system’s performance is assessed through the examination of the following four metrics: Localization error (LE): This measure quantifies how far away the estimated position of a sensor node is from its true position.This can be calculated as LE=and average localizationerror (ALE) is calculated as ALEIn the anonymous node I,while(xi,yi) represent its computed coordinates.The variable p signifies the number of unknown nodes,and Q denotes the communication radius of the network.Localization errors are standardised by normalising them to the radio range of sensor nodes.This involves dividing the localization errors by the radio range of the sensor nodes,simplifying the comparison of performance across different localization systems[30].This research area’s average localization accuracy assesses node placement precision.This is calculated using all study nodes’average localization errors.Unknown sensor node location: It assesses how well the system finds new sensor nodes.A confusion matrix: This matrix rates the system’s attack classification.The system is tested using the publicly available CICIDS 2017 benchmark dataset with varied assaults.Localization error and ALE are calculated from predicted sensor node coordinates and dataset positions.Localization of unknown sensor nodes assesses the system’s ability to find new ones.The dataset has training and test sets.The system is taught and tested on training and test sets.The confusion matrix measures system attack classification.Compare system-predicted attack labels to ground-truth labels.Calculate precision,recall,and F1-score from the confusion matrix.The system’s effectiveness is measured by its average detection rate,accuracy,precision,and recall.The system’s attack detection and classification metrics are calculated in[21].

    4.3 Energy Utilization Performance

    Energy optimisation and node-to-node communication in wireless sensor networks are crucial.It assesses the trustworthiness of network nodes.We examined energy use in low-power IoT nodes to boost energy.For a data transmission of k bits across a distance D and a threshold distance Do,the required transmission energy is calculated as ETX(s)=k × (Ee+Ef) × D2,for beacon packets ETX(b)=0.2 × (Ee+Ef).Hence the total energy utilization for the time slice t is calculated as Et=ETX(s)+ETX(b).Here,Efis the energy being received,and Eeis the power being lost somewhere along the line.Innovative integration of correlation composition analysis awareness and pair collaborative localization into a single framework improves model efficiency at low-power IoT nodes and reduces communication overhead.Fig.3 compares Region 1’s average energy use.Energy utilisation performance of the proposed model with 10 IoT sensor nodes:BSRTM 70%,WKNN 71.5%,BBSL 72.5%,IRDA 71.86%,ADSN 72.68%,and our proposed method 71.234%.With 100 IoT sensor nodes,BSRTM uses 91.254% of energy,whereas our proposed technique uses 75.648%,a 15.606%variation,and a minimum of 5.587%.Attack detection is done internally during localization.The region’s reward function from the previous round updates peer node location process outcomes.This reduces computational costs and improves low-power IoT node location accuracy.

    Figure 3: Energy utilization comparison

    4.4 Honesty Analysis

    We analyse honesty to calculate behavioral trust for secure attack localization and detection in IoT-based wireless sensor networks.This honesty performance metric is calculated as ho=Ks/Knthat proves the behavioral trust.There areKninteractions between source and destination nodes,andKsinteractions between source and destination nodes which lead to successful results.Behavioral trust in beacons is compromised when malicious nodes are within the radius of an anchor node.Impaired beacon node trust leads to incorrect selection of anchor and cluster head nodes for accurate localization of malicious nodes in the IoT.We have calculated the honesty of each IoT node,and the average honesty performance of all the IoT nodes in a region is depicted in Fig.4.We calculated each IoT node’s honesty,and Fig.4 shows the region’s average honesty performance.When we assessed 10 IoT sensor nodes,BSRTM gave honesty as 97.0085%,WKNN 96.4379%,BBSL 96.4379%,IRDA 96.7308%,and ADSN 98.0842%.Our proposed method is 2%–3%more honest than our comparative methods.

    4.5 Transaction Delay Performance

    The transaction delay between base and sensor nodes was studied by adjusting the number of IoT nodes from 10 to 100 and setting the communication packets to 100 64 KB packets.We tested 10 IoT sensor nodes and found that BSRTM took 4 s,WKNN 7 s,BBSL 5 s,IRDA 6 s,ADSN 8 s,and our proposed solution 3 s.For 100 IoT sensor nodes,IRDA offers 88 s transaction delay while our proposed technique gives 78 s with 10 s fluctuations which is depicted in Fig.5.

    Now when friends ask him what happened out there, he wiggles his nose and gives them a stare. Of fishing I m particularly fond. But there s just too many surprises in McFeeglebee s pond.

    Figure 4: Honesty comparison

    Figure 5: Transaction delay comparison

    4.6 Attack Detection Analysis

    This method aims to detect and locate rogue nodes in IoT-based wireless sensor networks to improve security.It uses a comprehensive strategy that incorporates correlation composition and pair-wise collaborative localization.Investigation of IoT device-initiated assaults on network security and operation.Denial-of-service (DoS) attacks,unauthorised access attempts,and other malicious activities designed to disrupt the IoT network are evaluated.

    The system uses benchmark datasets like CICIDS 2017 and UNSW-NB15 to evaluate its performance,which cover normal network behavior and popular attacks.Many present and past intrusion detection experiments use the UNSW-NB15 dataset.The IXIA Perfect Storm programme created regular and anomalous network traffic packets.The tool simulates nine attack classes to update security vulnerabilities and exposures in the collected packets.Fig.6 depicts attack detection accuracy.When we evaluated 10 IoT sensor nodes,F-IDS had an 85% accuracy attack,B-LSTM had 85%,LFLM had 84%,FC-IDS had 86%,HBFL had 85%,and our proposed method had 87%.The HBFL technique delivers a 94%accuracy attack for 100 IoT sensor nodes,whereas our method gives 98.5%with 4.5%fluctuation.The assault detection F1-score is in Fig.7.When we tested 10 IoT sensor nodes,BSRTM gave F1-score attack 86.82%,WKNN 85.4%,BBSL 85.4%,IRDA 86.24%,ADSN 83.5%,and our suggested method 89.2%.The ADSN technique yields an 86.85%F1-score attack for 100 IoT sensor nodes,while our method gives 91.9%with a 5.05%variance.The attack detection precision score is presented in Fig.8.When we evaluate with 10 IoT sensor nodes,by applying BSRTM technique we got precision attack as 91.82%,WKNN technique 91.65%,BBSL technique 96.2%,IRDA technique 94.2%ADSN technique 97.6%.where our proposed method gives precision attack as 93.6%.When we evaluate this for 100 IoT sensor nodes ADSN method gives precision attack as 91.9%and our proposed method gives 96.5% with 4.6% variation.The recall score performance is presented in Fig.9.When we evaluate with 10 IoT sensor nodes,by applying BSRTM technique we got recall attack as 70.82%,WKNN technique 85.2%,BBSL technique 80.52%,IRDA technique 84.92%ADSN technique 93.5%gives recall attack as 93.5% where our proposed method gives recall accuracy as 94.6%.When we evaluate with 100 IoT Sensor nodes BSRTM gives recall attack as 85.2%and our proposed method gives the recall attack as 94.8%.

    Figure 6: Attack detection accuracy

    Figure 7: Attack detection F1-score

    Figure 8: Attack detection precision score

    Figure 9: Attack detection recall score

    4.7 Localization Performance

    In the proposed localization scheme,we calculate all node positions and deviations.Fig.1 shows how malicious nodes are organized.False negatives misclassify honest nodes as malevolent in network traffic.Along with honesty,intimacy is one technique to measure the behavioral trustworthy values of anchor nodes.It may be computed asIn=wheremais the time consumed for the node to communicate with the anchor node andmiis the collaboration time of a particular node with beacon i.

    Fig.10 shows localization precision.Localizing 10 IoT sensor nodes using BSRTM gives 92.13%accuracy.WKNN 91.52%,BBSL 91.68%,IRDA 91.78%,ADSN 91.89%,and our technique gives 92% localization accuracy.BBSL localizes 100 IoT sensor nodes with 92.84% accuracy,while our method has 94.21%,a 1.37%difference.These F1-score performance can be observed from Fig.11.The localization precision score is in Fig.12.BSRTM locates 10 IoT sensor nodes:81.54%,WKNN 91.25%,BBSL 85.45%,IRDA 90.5%,and ADSN 85.23%.Overall,our localization method is 79.25%accurate.ADSN’s 90.245% accuracy with 100 IoT sensor nodes is 5% lower than our suggested method’s 95.24%precision score localization.The localization recall score is in Fig.13.BSRTM locates 71.15% of 10 IoT sensor nodes,WKNN 70.45%,BBSL 73.5%,IRDA 71.25%,and ADSN 76.89%.This localization approach has a 79.25%recall.Localizing 100 IoT sensor nodes with BBSL yields a 75.542% recall score,4.448% lower than our suggested method’s 79.99%.Simulations show that 10 IoT sensor nodes can detect and locate rogue nodes with 100%honesty,71.234%energy use,and a 3-s transactional latency.

    Figure 10: Localization accuracy

    Figure 11: Localization F1-score

    Figure 12: Localization precision score

    Figure 13: Localization recall score

    5 Conclusion

    The correlation composition awareness model and the pair-collaborative localization technique are crucial to attack detection and localization accuracy.By simultaneously addressing numerous security objectives,the framework provides a strong defence against potential threats and improves IoT system security,dependability,and efficiency.The Behavior and Performance Measurement (BPM)scheme ensures strict device authentication,allowing only trusted devices to access the network,boosting framework effectiveness.The system’s dynamic implicit authentication algorithms adapt to changing security environments,protecting the IoT ecosystem.Overall,this research offers practical methods that advance IoT security and promise secure IoT implementations in the future.

    Acknowledgement:We acknowledge the support for the evaluation of the proposed model from Intel IoT Center for Excellence,VIT-AP University.

    Author Contributions:The authors confirm contribution to the paper as follows: study conception and design:Kranthi A;data collection:Kranthi A;analysis and interpretation of results:Kranthi A,Gopikrishnan S;draft manuscript preparation: Kranthi A.,Gopikrishnan S;supervision and result analysis:Gopikrishnan S.All authors reviewed the results and approved the final version.

    Availability of Data and Materials:Not applicable.

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

    日日啪夜夜爽| 日韩中文字幕欧美一区二区 | 亚洲欧美一区二区三区久久| 亚洲五月色婷婷综合| 新久久久久国产一级毛片| 人人妻人人爽人人添夜夜欢视频| 另类亚洲欧美激情| 亚洲专区中文字幕在线 | 日韩制服丝袜自拍偷拍| 乱人伦中国视频| 亚洲第一av免费看| 一个人免费看片子| 天天影视国产精品| 亚洲精品久久成人aⅴ小说| 亚洲第一青青草原| 大香蕉久久网| 99精国产麻豆久久婷婷| 欧美在线一区亚洲| 欧美日韩成人在线一区二区| 免费观看a级毛片全部| 性少妇av在线| 熟女少妇亚洲综合色aaa.| 国产日韩一区二区三区精品不卡| 啦啦啦中文免费视频观看日本| 欧美 亚洲 国产 日韩一| 免费黄色在线免费观看| 久热爱精品视频在线9| 国产色婷婷99| 久久久久久久久久久久大奶| 男女边吃奶边做爰视频| 亚洲三区欧美一区| 亚洲在久久综合| 免费高清在线观看视频在线观看| 欧美日韩视频高清一区二区三区二| 麻豆乱淫一区二区| 国产探花极品一区二区| 久久精品久久久久久噜噜老黄| 搡老乐熟女国产| 国产精品 欧美亚洲| 国产一卡二卡三卡精品 | videos熟女内射| 丝袜在线中文字幕| 婷婷色av中文字幕| 曰老女人黄片| 国产在视频线精品| 久久精品久久精品一区二区三区| 91老司机精品| 91老司机精品| 黄色毛片三级朝国网站| 久久 成人 亚洲| 嫩草影院入口| 我要看黄色一级片免费的| 国产老妇伦熟女老妇高清| 99热全是精品| 亚洲国产精品成人久久小说| 国产片特级美女逼逼视频| 精品国产超薄肉色丝袜足j| 国产在线一区二区三区精| 久久精品aⅴ一区二区三区四区| 老汉色av国产亚洲站长工具| 国产精品一区二区在线观看99| 在线看a的网站| 69精品国产乱码久久久| 国产精品亚洲av一区麻豆 | 波多野结衣av一区二区av| 亚洲成人国产一区在线观看 | a级毛片在线看网站| 日韩av免费高清视频| 久久精品亚洲熟妇少妇任你| 精品一区二区三卡| 日韩大码丰满熟妇| 精品少妇久久久久久888优播| 另类精品久久| 一本一本久久a久久精品综合妖精| 只有这里有精品99| 亚洲情色 制服丝袜| 国产成人a∨麻豆精品| 美女视频免费永久观看网站| 国产野战对白在线观看| 亚洲成色77777| 亚洲欧美成人综合另类久久久| 91精品国产国语对白视频| 午夜精品国产一区二区电影| 五月开心婷婷网| 日韩伦理黄色片| 大片免费播放器 马上看| 亚洲美女搞黄在线观看| 天堂中文最新版在线下载| 国产一区亚洲一区在线观看| 男人爽女人下面视频在线观看| 老熟女久久久| 男女之事视频高清在线观看 | 亚洲精品国产色婷婷电影| 秋霞在线观看毛片| 女性生殖器流出的白浆| 爱豆传媒免费全集在线观看| 最近2019中文字幕mv第一页| 日本黄色日本黄色录像| 操美女的视频在线观看| 美国免费a级毛片| av在线老鸭窝| 色精品久久人妻99蜜桃| 久久精品久久精品一区二区三区| 丝袜喷水一区| 午夜福利网站1000一区二区三区| 亚洲国产精品一区三区| 熟妇人妻不卡中文字幕| 亚洲欧美一区二区三区黑人| 国产精品欧美亚洲77777| 久久久久国产精品人妻一区二区| 91精品三级在线观看| 在线观看免费日韩欧美大片| 中文字幕高清在线视频| 午夜免费观看性视频| 超色免费av| 国产99久久九九免费精品| 18禁动态无遮挡网站| 伦理电影大哥的女人| 亚洲熟女精品中文字幕| 成人国产麻豆网| 2018国产大陆天天弄谢| 可以免费在线观看a视频的电影网站 | 男女之事视频高清在线观看 | 久久久久久久国产电影| 熟妇人妻不卡中文字幕| 久久天堂一区二区三区四区| 亚洲在久久综合| 成人亚洲欧美一区二区av| 亚洲一区二区三区欧美精品| 精品久久久久久电影网| 激情五月婷婷亚洲| av电影中文网址| 中文字幕亚洲精品专区| 国产午夜精品一二区理论片| 欧美激情高清一区二区三区 | 悠悠久久av| 午夜免费男女啪啪视频观看| 亚洲第一区二区三区不卡| 久久这里只有精品19| 人人澡人人妻人| av女优亚洲男人天堂| 国产亚洲精品第一综合不卡| 午夜福利免费观看在线| 制服诱惑二区| 欧美日韩视频高清一区二区三区二| 熟妇人妻不卡中文字幕| 国产成人精品久久久久久| 国产精品 欧美亚洲| 蜜桃国产av成人99| 男女之事视频高清在线观看 | 熟女少妇亚洲综合色aaa.| 我的亚洲天堂| 视频区图区小说| 色婷婷av一区二区三区视频| 极品人妻少妇av视频| 国产成人精品无人区| 国产成人免费观看mmmm| 精品一区在线观看国产| 久久青草综合色| 国产精品免费视频内射| 日韩 欧美 亚洲 中文字幕| 叶爱在线成人免费视频播放| 久久精品久久久久久噜噜老黄| 丰满迷人的少妇在线观看| av女优亚洲男人天堂| 国产一区有黄有色的免费视频| 夫妻性生交免费视频一级片| 国产 一区精品| 国产精品女同一区二区软件| 久久久久国产精品人妻一区二区| 日韩一区二区视频免费看| av网站免费在线观看视频| 国产一区亚洲一区在线观看| 免费在线观看视频国产中文字幕亚洲 | 美女主播在线视频| 高清欧美精品videossex| 考比视频在线观看| 国产精品秋霞免费鲁丝片| 色婷婷久久久亚洲欧美| 国产一区二区激情短视频 | 免费在线观看黄色视频的| 国产在视频线精品| 久久鲁丝午夜福利片| 精品久久蜜臀av无| 女的被弄到高潮叫床怎么办| 免费女性裸体啪啪无遮挡网站| 精品一区在线观看国产| 在线观看人妻少妇| 最新的欧美精品一区二区| 嫩草影院入口| 国产精品二区激情视频| xxx大片免费视频| 亚洲精品国产区一区二| 久久人妻熟女aⅴ| av天堂久久9| 国产乱人偷精品视频| 伊人亚洲综合成人网| 丁香六月欧美| 欧美激情 高清一区二区三区| 少妇人妻 视频| 777久久人妻少妇嫩草av网站| 亚洲精品美女久久久久99蜜臀 | 99久国产av精品国产电影| 人人妻人人爽人人添夜夜欢视频| 成人国语在线视频| 在线观看www视频免费| 成人亚洲欧美一区二区av| 国产视频首页在线观看| 丁香六月天网| 美女主播在线视频| 国产老妇伦熟女老妇高清| 亚洲久久久国产精品| 女人精品久久久久毛片| 蜜桃国产av成人99| 亚洲一区二区三区欧美精品| 国产免费一区二区三区四区乱码| 久久久精品免费免费高清| 亚洲成人一二三区av| 女人久久www免费人成看片| 一级片'在线观看视频| 国产成人欧美| 亚洲激情五月婷婷啪啪| 高清欧美精品videossex| 一本一本久久a久久精品综合妖精| 一本色道久久久久久精品综合| 亚洲精品日本国产第一区| 久久精品国产亚洲av涩爱| 亚洲av福利一区| 亚洲国产欧美日韩在线播放| 国产精品久久久久久人妻精品电影 | 91成人精品电影| 午夜日本视频在线| 一边摸一边抽搐一进一出视频| 黄片小视频在线播放| 色吧在线观看| 黄色 视频免费看| 人成视频在线观看免费观看| 午夜91福利影院| 免费在线观看完整版高清| 亚洲免费av在线视频| 狂野欧美激情性bbbbbb| 免费高清在线观看日韩| 一二三四中文在线观看免费高清| 黄片小视频在线播放| 国产一区亚洲一区在线观看| 97在线人人人人妻| 日韩精品有码人妻一区| 啦啦啦啦在线视频资源| 久久免费观看电影| 亚洲四区av| 国产在线视频一区二区| 欧美少妇被猛烈插入视频| 久久久国产欧美日韩av| 人人妻人人澡人人看| 久热爱精品视频在线9| 国产有黄有色有爽视频| 午夜福利视频精品| 在线天堂中文资源库| www.熟女人妻精品国产| avwww免费| 日韩一卡2卡3卡4卡2021年| 在线观看www视频免费| 永久免费av网站大全| 亚洲婷婷狠狠爱综合网| 亚洲中文av在线| 男女高潮啪啪啪动态图| 日韩视频在线欧美| 最近最新中文字幕免费大全7| 精品国产超薄肉色丝袜足j| 天天操日日干夜夜撸| 男男h啪啪无遮挡| 亚洲第一青青草原| 国产人伦9x9x在线观看| 在线观看免费视频网站a站| 操美女的视频在线观看| 乱人伦中国视频| 制服诱惑二区| 热re99久久精品国产66热6| 满18在线观看网站| 又黄又粗又硬又大视频| 天天添夜夜摸| 久久 成人 亚洲| 亚洲国产精品999| 国产精品三级大全| 无遮挡黄片免费观看| 99热国产这里只有精品6| 亚洲精品第二区| 国产免费福利视频在线观看| 亚洲自偷自拍图片 自拍| 黄色一级大片看看| 校园人妻丝袜中文字幕| 欧美人与善性xxx| 免费av中文字幕在线| 新久久久久国产一级毛片| 91aial.com中文字幕在线观看| 狂野欧美激情性bbbbbb| 天天躁夜夜躁狠狠久久av| 高清欧美精品videossex| 色94色欧美一区二区| 黑人欧美特级aaaaaa片| 亚洲美女搞黄在线观看| 捣出白浆h1v1| 日本欧美视频一区| 中文天堂在线官网| 精品第一国产精品| www.自偷自拍.com| 亚洲第一区二区三区不卡| 久久久久精品国产欧美久久久 | 国产一卡二卡三卡精品 | 丰满迷人的少妇在线观看| 亚洲精品一区蜜桃| 色精品久久人妻99蜜桃| 校园人妻丝袜中文字幕| 国产精品二区激情视频| 我的亚洲天堂| 国产探花极品一区二区| 看免费av毛片| 亚洲精品美女久久久久99蜜臀 | 国产精品国产三级国产专区5o| 中文字幕另类日韩欧美亚洲嫩草| 欧美日韩国产mv在线观看视频| 捣出白浆h1v1| 国精品久久久久久国模美| 亚洲第一青青草原| 人人澡人人妻人| 国产成人午夜福利电影在线观看| 国产精品 欧美亚洲| 美女午夜性视频免费| 操美女的视频在线观看| 乱人伦中国视频| 操出白浆在线播放| 中文字幕制服av| 如日韩欧美国产精品一区二区三区| 久久久久国产精品人妻一区二区| 亚洲精品日韩在线中文字幕| 少妇人妻精品综合一区二区| 日韩av在线免费看完整版不卡| 下体分泌物呈黄色| 欧美精品一区二区大全| 美女国产高潮福利片在线看| 男的添女的下面高潮视频| 国产精品嫩草影院av在线观看| 欧美日韩一区二区视频在线观看视频在线| 男男h啪啪无遮挡| 欧美激情极品国产一区二区三区| 久久精品久久精品一区二区三区| 岛国毛片在线播放| 日日摸夜夜添夜夜爱| 亚洲精品美女久久av网站| 亚洲国产精品一区三区| 亚洲在久久综合| 人人澡人人妻人| 人妻一区二区av| 欧美精品av麻豆av| 亚洲精华国产精华液的使用体验| 国产精品 欧美亚洲| 精品国产一区二区三区四区第35| 欧美中文综合在线视频| 两个人免费观看高清视频| 午夜久久久在线观看| 日韩电影二区| 精品一区在线观看国产| 国产毛片在线视频| 久久青草综合色| 亚洲精品美女久久av网站| 久久青草综合色| 欧美日本中文国产一区发布| 啦啦啦在线免费观看视频4| 中文字幕人妻丝袜一区二区 | 亚洲免费av在线视频| 青春草国产在线视频| 亚洲一区二区三区欧美精品| 性色av一级| 国产精品国产三级专区第一集| 秋霞在线观看毛片| 国产精品一国产av| 欧美 日韩 精品 国产| 中文字幕人妻熟女乱码| 国产在线一区二区三区精| 极品人妻少妇av视频| 极品少妇高潮喷水抽搐| 777米奇影视久久| 男人添女人高潮全过程视频| www.精华液| 成人午夜精彩视频在线观看| 免费高清在线观看日韩| 一本久久精品| 亚洲七黄色美女视频| 亚洲一区中文字幕在线| 电影成人av| 久久久久国产精品人妻一区二区| 久久性视频一级片| 狠狠精品人妻久久久久久综合| 日韩不卡一区二区三区视频在线| 在线观看人妻少妇| 色综合欧美亚洲国产小说| 一区福利在线观看| 成人18禁高潮啪啪吃奶动态图| 亚洲精华国产精华液的使用体验| 亚洲精品久久午夜乱码| 亚洲中文av在线| 好男人视频免费观看在线| 免费高清在线观看日韩| 免费日韩欧美在线观看| 视频在线观看一区二区三区| 性色av一级| 69精品国产乱码久久久| 美女视频免费永久观看网站| 久久这里只有精品19| 国产亚洲精品第一综合不卡| 熟女少妇亚洲综合色aaa.| 欧美精品高潮呻吟av久久| av又黄又爽大尺度在线免费看| 美女中出高潮动态图| 男女高潮啪啪啪动态图| 久久久久精品人妻al黑| 毛片一级片免费看久久久久| 美女脱内裤让男人舔精品视频| 午夜福利影视在线免费观看| 男的添女的下面高潮视频| 在线免费观看不下载黄p国产| 又大又爽又粗| 中文字幕亚洲精品专区| 国产97色在线日韩免费| 秋霞伦理黄片| 观看美女的网站| 国产精品二区激情视频| 丝袜美足系列| 男女之事视频高清在线观看 | 亚洲国产欧美日韩在线播放| 日本vs欧美在线观看视频| 丁香六月天网| 国产国语露脸激情在线看| 亚洲三区欧美一区| 国产精品国产三级国产专区5o| 肉色欧美久久久久久久蜜桃| 男女边吃奶边做爰视频| 亚洲av在线观看美女高潮| 婷婷色麻豆天堂久久| 99久国产av精品国产电影| 欧美激情极品国产一区二区三区| 国产精品欧美亚洲77777| 国产成人系列免费观看| 欧美日韩福利视频一区二区| 国产精品一区二区精品视频观看| 美女高潮到喷水免费观看| 久久午夜综合久久蜜桃| 在线观看一区二区三区激情| 亚洲美女黄色视频免费看| 午夜激情av网站| 免费女性裸体啪啪无遮挡网站| 天美传媒精品一区二区| 十分钟在线观看高清视频www| 少妇 在线观看| 91国产中文字幕| 熟妇人妻不卡中文字幕| 可以免费在线观看a视频的电影网站 | 亚洲av日韩在线播放| 亚洲欧美一区二区三区久久| 久久99一区二区三区| 99热全是精品| 国产精品无大码| 老熟女久久久| 久久久国产一区二区| 熟女av电影| 卡戴珊不雅视频在线播放| 男女之事视频高清在线观看 | 飞空精品影院首页| 亚洲免费av在线视频| 日韩欧美一区视频在线观看| 9色porny在线观看| 五月开心婷婷网| 亚洲自偷自拍图片 自拍| 日韩av在线免费看完整版不卡| 老司机靠b影院| 亚洲免费av在线视频| 久久久精品免费免费高清| 久久久久久久精品精品| 国产有黄有色有爽视频| 免费黄色在线免费观看| 精品国产乱码久久久久久小说| 欧美av亚洲av综合av国产av | 国产免费现黄频在线看| 亚洲四区av| av在线观看视频网站免费| 日韩制服丝袜自拍偷拍| 大片电影免费在线观看免费| 国产精品二区激情视频| 一区二区三区乱码不卡18| 久久天躁狠狠躁夜夜2o2o | 捣出白浆h1v1| 我要看黄色一级片免费的| 亚洲精品日韩在线中文字幕| 天天躁夜夜躁狠狠躁躁| 久久久久视频综合| 1024香蕉在线观看| 大片免费播放器 马上看| 欧美在线一区亚洲| 悠悠久久av| 久久久久视频综合| 国产伦理片在线播放av一区| 91精品伊人久久大香线蕉| 纵有疾风起免费观看全集完整版| 久久精品久久久久久久性| 色婷婷av一区二区三区视频| 赤兔流量卡办理| 国产免费视频播放在线视频| 纵有疾风起免费观看全集完整版| 婷婷成人精品国产| 少妇的丰满在线观看| 亚洲,欧美精品.| 咕卡用的链子| 国产极品天堂在线| 嫩草影视91久久| 国产精品三级大全| 美女高潮到喷水免费观看| 一边摸一边抽搐一进一出视频| 成人黄色视频免费在线看| 国产野战对白在线观看| 成人漫画全彩无遮挡| 99国产综合亚洲精品| 九色亚洲精品在线播放| 少妇人妻 视频| 日韩欧美一区视频在线观看| 在线观看免费高清a一片| 国产福利在线免费观看视频| 女人被躁到高潮嗷嗷叫费观| 午夜激情久久久久久久| 黄色视频不卡| 国产老妇伦熟女老妇高清| 日本午夜av视频| 国产免费又黄又爽又色| 国产成人av激情在线播放| 狠狠婷婷综合久久久久久88av| 侵犯人妻中文字幕一二三四区| 欧美日韩福利视频一区二区| 日韩精品免费视频一区二区三区| 日日爽夜夜爽网站| 天天躁夜夜躁狠狠久久av| 嫩草影视91久久| 久久久久精品国产欧美久久久 | 精品人妻一区二区三区麻豆| 在线观看国产h片| 午夜福利在线免费观看网站| 99热全是精品| 女人精品久久久久毛片| 深夜精品福利| 亚洲自偷自拍图片 自拍| 亚洲精品久久久久久婷婷小说| 婷婷成人精品国产| 又大又黄又爽视频免费| 纯流量卡能插随身wifi吗| 国产一区二区在线观看av| 国产免费又黄又爽又色| 男女国产视频网站| 国产成人一区二区在线| 久久99热这里只频精品6学生| 国产精品一区二区在线观看99| 亚洲欧美日韩另类电影网站| 亚洲国产av影院在线观看| 蜜桃在线观看..| av国产久精品久网站免费入址| 亚洲精品视频女| 亚洲伊人色综图| 久久热在线av| 免费女性裸体啪啪无遮挡网站| 黄色 视频免费看| 免费久久久久久久精品成人欧美视频| 国产在线免费精品| 少妇人妻精品综合一区二区| 一级毛片 在线播放| 欧美激情 高清一区二区三区| 亚洲国产欧美网| 成年美女黄网站色视频大全免费| 狂野欧美激情性bbbbbb| 亚洲熟女精品中文字幕| 日日啪夜夜爽| 精品少妇黑人巨大在线播放| 热re99久久国产66热| 18在线观看网站| av在线老鸭窝| 国精品久久久久久国模美| 一级,二级,三级黄色视频| 男女午夜视频在线观看| 黄片播放在线免费| 亚洲精品aⅴ在线观看| 国产深夜福利视频在线观看| 丰满迷人的少妇在线观看| 国产免费现黄频在线看| 青春草国产在线视频| 国产男人的电影天堂91| 久久精品人人爽人人爽视色| 亚洲三区欧美一区| 中文精品一卡2卡3卡4更新| 啦啦啦 在线观看视频| 下体分泌物呈黄色| 99久久精品国产亚洲精品| 久久av网站| 亚洲国产精品成人久久小说| 极品人妻少妇av视频| 亚洲成av片中文字幕在线观看| 国产精品久久久久久精品电影小说| 日韩制服骚丝袜av| 国产伦理片在线播放av一区| 久久久久网色| 少妇人妻久久综合中文| 99国产精品免费福利视频| 国产精品一区二区精品视频观看| 国产成人免费观看mmmm| 蜜桃在线观看..| 色婷婷久久久亚洲欧美| 精品一区二区三区四区五区乱码 | 美女高潮到喷水免费观看| 一边摸一边做爽爽视频免费| 精品午夜福利在线看|