• <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 ?

    Variational Bayesian Based IMM Robust GPS Navigation Filter

    2022-08-24 12:58:06DahJingJwoandWeiYehChang
    Computers Materials&Continua 2022年7期

    Dah-Jing Jwoand Wei-Yeh Chang

    1Department of Communications, Navigation and Control Engineering, National Taiwan Ocean University Keelung,202301, Taiwan

    2TDK Taiwan Corp., Yangmei, Taoyuan, 326021, Taiwan

    Abstract: This paper investigates the navigational performance of Global Positioning System (GPS) using the variational Bayesian (VB) based robust filter with interacting multiple model (IMM) adaptation as the navigation processor.The performance of the state estimation for GPS navigation processing using the family of Kalman filter (KF) may be degraded due to the fact that in practical situations the statistics of measurement noise might change.In the proposed algorithm, the adaptivity is achieved by estimating the timevarying noise covariance matrices based onVB learning using the probabilistic approach, where in each update step, both the system state and time-varying measurement noise were recognized as random variables to be estimated.The estimation is iterated recursively at each time to approximate the real joint posterior distribution of state using the VB learning.One of the two major classical adaptive Kalman filter (AKF) approaches that have been proposed for tuning the noise covariance matrices is the multiple model adaptive estimate (MMAE).The IMM algorithm uses two or more filters to process in parallel, where each filter corresponds to a different dynamic or measurement model.The robust Huber’s M-estimation-based extended Kalman filter (HEKF) algorithm integrates both merits of the Huber M-estimation methodology and EKF.The robustness is enhanced by modifying the filter update based on Huber’s M-estimation method in the filtering framework.The proposed algorithm, referred to as the interactive multi-model based variational Bayesian HEKF (IMM-VBHEKF), provides an effective way for effectively handling the errors with time-varying and outlying property of non-Gaussian interference errors, such as the multipath effect.Illustrative examples are given to demonstrate the navigation performance enhancement in terms of adaptivity and robustness at the expense of acceptable additional execution time.

    Keywords: GPS; variational bayesian; Huber’s M-estimation; interacting multiple model; adaptive; outlier; multipath

    1 Introduction

    Non-Gaussian noise is often encountered in many practical environments and the estimation performance deteriorates dramatically.Known to be one of the dominant error sources in high accuracy global navigation satellite systems (GNSS) [1] positioning systems, such as the Global Positioning System (GPS) [2,3], multipath effects [2,4] occur when GPS signals arrive at a receiver site via multiple paths due to reflections from nearby objects, such as the ground and water surfaces,buildings, vehicles, hills, trees, etc.Many estimation algorithms have been studied to eliminate the positioning error caused by multipath.Since multipath errors are among uncorrelated errors that are not cancelled out during observation differencing, the performance of high precision GPS receivers are mostly limited by the multipath induced errors.One of the most important issues in GPS system performance improvement is the interference suppression techniques.

    The well-known Kalman filter (KF) [5] provides optimal estimate, namely minimum mean square error (MSE), of the system state vector and has been recognized as one of the most powerful state estimation techniques.As a recursive minimum norm estimation technique, the KF exhibits sensitivity to deviations from the assumed underlying probability distributions.Due to its simple structure, stable performance and low computational complexity, the conventional adaptive filtering algorithm where the least MSE is involved has been widely used in a variety of applications in the fields of adaptive signal processing and machine learning.However, the MSE criterion is limited to the assumption of linearity and Gaussianity while most of the noise in real word is non-Gaussian.The performance deteriorates significantly in the non-Gaussian noise environment.The traditional Kalman-type filter provides the best filter estimate when the noise is Gaussian, but most noise in real life is unknown,uncertain and non-Gaussian.The extended Kalman filter (EKF) is a nonlinear version of the KF and has been widely employed as the GPS navigation processor.The fact that EKF highly depends on a predefined dynamics model forms a major drawback.Furthermore, it is not optimal when the system is disturbed by heavy-tailed (or impulsive) non-Gaussian noises.To solve the performance degradation problem with non-Gaussian errors or heavy-tailed non-Gaussian noises, some robust Kalman filters have been developed by using non-minimum MSE criterion as the optimality criterion.

    Essentially based on modification of the quadratic cost function in the filter framework, the Huber M-estimation methodology [6-9] provides between smooth norm properties for small residuals and robust norm properties for large residuals.The technique that relies on Huber’s generalized maximum likelihood (ML) methodology exhibits robustness against deviations from the commonly assumed Gaussian probability density functions and can solve the non-Gaussian distribution problem efficiently and has been successfully employed for robust state estimation, inertial navigation system and visual tracking applications.The robust Huber’s M-estimation-based EKF (HEKF) algorithm integrates both merits of the Huber M-estimation methodology and EKF.For the signals contaminated with non-Gaussian noises or outliers, a robust scheme combining the Huber M-estimation methodology and the EKF framework is beneficial where the Huber M-estimation methodology is used to reformulate the measurement information of the EKF to handle outliers and provide robustness against deviations from Gaussianity.The measurement update can be modified to enhance robustness using the Huber M-estimation methodology.

    The variational Bayesian (VB) approach [10-15] is an inference method that utilizes a simple distribution to approximate the true posterior distribution of hidden variables, usually assuming that the hidden variables are independent of each other.The VB has been developed for a wide range of models to perform approximate posterior inference at low computational cost in comparison with the sampling methods.These methods assume a simpler, analytically tractable form for the posterior.The purpose of the variation is to find a variational distribution of the posterior probability density function (pdf) to minimize the Kullback-Leibler divergence (KLD) between the true posterior pdf.Two main approaches are either to derive a factored free form distribution or to assume a fixed form posterior distribution (e.g., a multivariate Gaussian, with possibly a suitable parametrization of the model).

    The classical way of solving the problem of uncertain parameters is to use adaptive filters [16]where the model parameters or the noise statistics are estimated together with the dynamic states.The classical noise adaptive filtering approaches can be divided into Bayesian, maximum likelihood,correlation and covariance matching methods.The Bayesian approach is the most general of these and the other approaches can often be interpreted as approximations to the Bayesian approach.Examples of Bayesian approaches to noise adaptive filtering include state augmentation based methods,multiple model methods such as the interacting multiple model (IMM) algorithm [17-21].The IMM algorithm has the configuration that runs in parallel several model-matched state estimation filters, which exchange information (namely interact) at each sampling time.The IMM algorithm has been originally applied to target tracking, and recently extended to navigation application.A model probability evaluator calculates the current probability of the vehicle being in each of the possible modes.A global estimate of the vehicle’s state is computed using the latest mode probabilities.The algorithm carries out a soft-switching between the various modes by adjusting the probabilities of each mode, which are used as weightings in the combined global state estimate.The covariance matrix associated with this combined estimate takes into account the covariances of the mode-conditioned estimates as well as the differences between these estimates.

    The algorithm proposed in this paper intends to provide an effective way for effectively handling the errors with time-varying and outlying property of non-Gaussian interference errors.The method utilizes the VB learning to approximate the noise strength for time-varying noise covariances, the Huber M-estimation methodology to enhance robustness especially for overcoming the problem of contamination distribution or outliers, and the IMM algorithm to furtherly tune the noise covariance matrices.The remaining of this paper is organized as follows.In Section 2, preliminary background on the Huber’s M-estimation-based extended Kalman filter is reviewed.The variational Bayesian algorithm is discussed in Section3.In Section4,the interacting multiple model algorithm is presented.In Section 5, simulation experiments are carried out to evaluate the performance and effectiveness.Conclusions are given in Section 6.

    2 Huber’s M-Estimation-Based Extended Kalman Filter

    For the nonlinear single model equations in discrete-time form, we have

    where the state vector xk∈Rn, process noise vector wk∈Rm, measurement vector zk∈Rm,measurement noise vector vk∈Rm,Qkis the process noise covariance matrix and Rkis the measurement noise covariance matrix.The vectors wkand vkin Eqs.(1) and (2) are zero mean Gaussian white sequences has zero cross-correlation with each other:

    whereE[·]represents expectation, and superscript “ T”denotes matrix transpose.The symbol δikstands for the Dirac function:

    Based on robust strategy, the Huber M-estimation methodology possesses the ability to solve pollution distribution or outliers by improving filter update.The essence of Huber’s robust estimation method is to solve the filter estimation value when the minimum cost function is obtained.For further applications,based on the generalized maximum likelihood estimation, amore general object function is given by Huber

    where ρ(·) is an appropriate positive symmetric function,υk,iis the residual.The cost function given by Huber is

    where γ is the default parameter.Amounts of experiences show that the performance is best when γ= 1.345.For Huber’s filtering robustness framework, reweighted average is used by

    where υk,iis the residual,φ(υk,i) =ρ′(υk,i) is the influence function.

    The Huber M-estimation methodology makes use of the weighting matrix to recast the measurement information.The weighting matrix can be interpreted in terms of reweighting the residual error covariance and constructing the measurement process.The modified covariance has the form

    The HEKF algorithm is summarized as follow:

    Stage 1: Prediction steps/time update equations

    (1) Predict the state vector

    (2) Predict the state error covariance matrix

    Stage 2: Correction steps/measurement update equations

    (3) Compute the residual

    (4) Compute the reweighting covariance matrix for measurement noise

    (5) Compute Kalman gain matrix

    (6) Update state vector

    (7) Update state error covariance matrix

    3 The Variational Bayesian Approach

    The variational Bayesian approach is an inference method that utilizes a simple distribution to approximate the true posterior distribution of hidden variables to find a variational distribution of the posterior probability density function (pdf) to minimize the KLD between the true posterior pdf.The variable decibel parameter is estimated by iterative calculations in which the gradient descent algorithm is involved.The difference between the lower bounds of two adjacent repeated calculations is used as the basis for judging convergence.

    3.1 Variational Bayesian Extended Kalman Filter

    Based on the Bayesian rule, the posterior distribution of the system state and the measurement noise covariance can be represented as

    which can be simplified top(zk|xk,Rk).For the first-order Markov process

    the Chapman-Kolmogorov equation can be obtained through the integration.

    Given the next measurement zk, the predictive distribution is updated to a posterior distribution using the Bayes rule

    The conditional probability distribution can be approximated as a product of Gaussian and Inverse-Gamma distributions

    whereN(?k|k-1,Pk|k-1) is the Gaussian distribution for prediction xkwith parameters?k|k-1and Pk|k-1.αk|k-1,i,βk|k-1,iare two parameters of the Inverse-Gamma distribution for, thei-th diagonal element of Rk.

    The variational Bayesian extended Kalman filter (VBEKF) employes a heuristic approach in the calculation process and predicts the distribution parameter by the first-order approximation

    The parameters of the new distribution is updated through

    and the measurement noise variance can be estimated

    The VBEKF algorithm is summarized as follows:

    Stage 1: Prediction steps/time update equations

    (1) Predict the state vector:

    (2) Predict the state error covariance matrix

    (3) Predict the distribution parameter

    Stage 2: Correction steps/measurement update equations

    (4) Compute measurement noise variance

    (5) Compute Kalman gain matrix

    (6) Update state vector

    (7) Update state error covariance matrix

    (8) Update the measurement distribution parameter

    3.2 Variational Bayesian Huber’s M-Estimation Based Extended Kalman Filter

    The positioning accuracy is degraded due to the presence of outliers and time-varying noise strength.Incorporation of the VB and Huber M-estimation methodologies into the EKF yielding the variational Bayesian Huber’s M-estimation based extended Kalman filter (VBHEKF) algorithm can effectively overcome the outliers and time-varying variance in the measurement noise and improve the positioning accuracy.The VBHEKF algorithm is summarized as follows:

    Stage 1: Prediction steps/time update equations

    (1) Predict the state vector

    (2) Predict the state error covariance matrix

    (3) Predict the distribution parameters

    Stage 2: Correction steps/measurement update equations

    (4) Compute measurement noise variance:

    (5) Compute the residual

    (6) Compute the reweighting covariance matrix for measurement noise

    (7) Compute Kalman gain matrix

    (8) Update state vector

    (9) Update state error covariance matrix

    (10) Update the measurement distribution parameter

    4 Interacting Multiple Model Extended Kalman Filter

    Based on pseudo-Bayesian theory, the IMM algorithm employs two or more filters to process in parallel where each filter corresponds to a different state space or measurement model.It is an adaptive mechanism which dynamically identifies uncertainties or modeling errors can be adopted and is mainly an algorithm involving Markov chain switching coefficients between different models.In each model an EKF is running, and the IMM algorithm makes uses of model probabilities to weight the inputs and output of a bank of parallel filters at each time instant.

    Assuming that a target with r states, corresponding to r models, set the system state equation and system observation equation represented by thej-th model are as follows:

    where the state vector xk∈Rn, process noise vector wk∈Rm, measurement vector zk∈Rm,measurement noise vector vk∈Rm,Qkis the process noise covariance matrix and Rkis the measurement noise covariance matrix.The vectors wkand vkin Eqs.(35) and (36) are zero mean Gaussian white sequences has zero cross-correlation with each other:

    whereE[·]represents expectation, and superscript “ T”denotes matrix transpose.The symbol δikstands for the Dirac function:

    The IMM-EKF algorithm is summarized as follows:

    Stage 1: Model interaction/mixing

    Compute mixed state and corresponding covariances of modelj

    Stage 2: Model filtering

    Stage 3: Model probabilities update

    Compute the likelihood function

    wherevj= zk-and.The model probability is updated according to the model likelihood and model transition probability

    Stage 4: Combination of model estimates and covariances

    The algorithm performs real-time detection by setting model filters corresponding to the number of possible models, setting weight coefficients and model update probabilities for each filter, and finally performing weighted fusion to calculate the current optimal system state estimation to achieve the purpose of model adaptation.

    5 Illustrative Examples

    To validate the effectiveness of the proposed approaches, simulation experiments have been carried out to evaluate the performance of the proposed approach in comparison with the other conventional methods for GPS navigation processing.The computer codes were developed by the authors using the Matlab?software.The commercial software Satellite Navigation (SatNav) Toolbox by GPSoft LLC [22] was employed for generation of the GPS satellite orbits/positions and thereafter, the satellite pseudoranges, carrier phase measurement, and constellation, required for simulation.

    The simulated pseudorange error sources corrupting GPS measurements include ionospheric delay, tropospheric delay, receiver noise and multipath.It is assumed that the differential GPS(DGPS)mode is available and therefore most of the receiver-independent common errors can be corrected,while the multipath and receiver thermal noise cannot be eliminated.The multipath interferences are added into the GPS pseudorange observation data during the vehicle moving.Two scenarios dealing with two types of interferences in pseudorange observables are carried out, which covers pseudorange errors involving (1) outlier type of multipath interferences during the vehicle moving, and (2) timevarying variance in the measurement noise with additional outliers.

    Fig.1 shows the test trajectory for the simulated vehicle and the skyplot during the simulation.In the simulation, there are 8 GPS satellites available.Since the research focus on the mitigation of multipath errors, the influence of measurement noise is relatively critical.A vehicle is designed to perform the uniform accelerated motion to reduce the impact caused by unmodeling system dynamic errors.First of all, performance comparison for the four estimator/filter including EKF,VBEKF, HEKF and VBHEKF is shown.Secondly, performance enhancement based on the IMM configuration is presented.

    Figure 1: Test trajectory for the simulated vehicle (left) and the skyplot (right) during the simulation

    5.1 Scenario 1: Pseudorange Observable Involving Outlier Type of Multipath Errors

    In Scenario 1, mitigation of the errors for pseudorange observable involving outlier type of multipath interferences is investigated.There are totally seven time durations where additional randomly generated errors are intentionally injected into the GPS pseudorange observation data during the vehicle moving.Tab.1 provides the information of the standard deviation for the 5 outliers.

    Table 1: Information of the standard deviation for the five outliers

    5.1.1 Performance Comparison for EKF and its Variants

    Comparison of GPS navigation accuracy for the four schemes: EKF, VBEKF, HEKF and VBHEKF is shown in Fig.2.The results show that either the VB or Huber’s algorithms can assist EKF to effectively deal with the outliers in the pesudorange observables individually and combination of the two algorithms can furtherly enhance the performance.

    Figure 2: Comparison of positioning accuracy for extended Kalman filter (EKF) and its variants-Scenario 1

    5.1.2 Performance Enhancement Based on the Interacting Multiple Model Configuration

    Incorporation of the IMM into the various filters is utilized for further performance enhancement.Fig.3 illustrates the positioning accuracy when the IMM configuration is involved for various algorithms: IMM-EKF, IMM-VBEKF, IMM-HEKF and IMM-VBHEKF.For the case that the VB does not possess sufficient capability to resolve the outlier type of interference, the IMM-VBHEKF demonstrates substantial performance improvement in navigation accuracy with acceptable extra computational expense.Tab.2 summarizes the estimation performance and execution time for various algorithms.

    Figure 3: Comparison of positioning accuracy when the interacting multiple model (IMM) configuration is involved for Scenario 1

    Table 2: Performance comparison for various algorithms-Scenario 1

    5.2 Scenario 2: Errors Involving Time-Varying Variance in Measurement Noise with Additional Outliers

    Scenario 2 is designed for investigating the performance when dealing with the time-varying Gaussian measurement noise with additional outliers.Description of time varying measurement variances in the five time intervals is shown in Tab.3.The variances of time-varying measurement noiserρiin the five time intervals for this scenario are assumed to follows the variation:rρi= 1→10→1→17→1 (in unit of m2), where the‘a(chǎn)rrows (→)’is employed for indicating the time-varying trajectory of variances.The set of unknown time-varying statistical parameters of noise needs to be estimated with the system states and the error covariance.

    Table 3: Description of the variation of noise strength and number of additional outliers in the five time intervals for Scenario 2

    5.2.1 Performance Comparison for EKF and its Variants

    Comparison of GPS navigation accuracy for the EKF, VBEKF, HEKF and VBHEKF is shown in Fig.4 where comparison of positioning accuracies without outlier and with outlier, respectively is provided.The results show that both VB and Huber based EKF can effectively improve the positioning performance.As can be seen, both the VB and Huber algorithms can be adopted into the EKF to improve GPS navigation accuracy in the environment of time-varying Gaussian noise.When additional outliers are involved in addition to the time-varying interferences, neither VB nor Huber only algorithm possesses sufficient capability to resolve the problem for such kind of challenge.The VBHEKF demonstrates substantial performance improvement in navigation accuracy at the expense of acceptable extra computational cost.

    Figure 4: Comparison of positioning accuracy for EKF and its variants-Scenario 2 (a) without outlier(b) with outlier

    5.2.2 Performance Enhancement Based on the Interacting Multiple Model Configuration

    Comparison of positioning accuracies for the four algorithms: IMM-EKF, IMM-VBEKF, IMMHEKF and IMM-VBHEKF is shown in Fig.5.Positioning accuracies for the cases without and with outliers, respectively, are presented.The Huber’s approach does not catch up the variation of noise strength very well.With the assistance of VB, the VBHEKF can further improve the performance.From the other view point, the adaptation capability of noise variance for the VBHEKF has been improved with the assistance of the IMM mechanism.Tabs.4 and 5 summarize the performance for various algorithms without and with outliers, respectively.Among the various approaches, the IMMVBHEKF provides the best positioning accuracy with only limited amount of extra execution time.

    Table 4: Performance comparison for Scenario 2-without outliers

    Table 5: Performance comparison for Scenario 2-with outliers

    Figure 5: Comparison of positioning accuracy for Scenario 2 when the IMM configuration is involved when IMM algorithm is incorporated (a) without outlier (b) with outlier

    6 Conclusions

    In this paper, an interacting multiple model based variational Bayesian robust extended Kalman filter is adopted for processing GPS navigation states.The method utilizes the variational Bayesian learning to approximate the noise strength and to provide a strong tracking capability for time-varying noise covariances; the Huber M-estimation methodology based on the robust strategy is employed to enhance robustness especially for overcoming the problem of contamination distribution or outliers;the IMM algorithm is introduced for further tuning of the noise covariance matrices.Combination the three merits leads to the estimator referred to as the IMM-VBHEKF.Illustrative examples have been provided for validation of the method.Two scenarios are presented representing two possible peudorange error patterns, including (1) observable involving outlier type of multipath errors;(2) observable involving time-varying variance in measurement noise with additional outliers.The proposed method has demonstrated performance enhancement in terms of adaptivity and robustness at the expense of acceptable additional execution time.

    Funding Statement:This work has been partially supported by the Ministry of Science and Technology,Taiwan [Grant Numbers MOST 108-2221-E-019-013 and MOST 109-2221-E-019-010].

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

    99热网站在线观看| 午夜免费成人在线视频| 搡老乐熟女国产| 人人澡人人妻人| 精品久久久精品久久久| 又紧又爽又黄一区二区| 久热爱精品视频在线9| 晚上一个人看的免费电影| 新久久久久国产一级毛片| 丝瓜视频免费看黄片| 午夜视频精品福利| 亚洲精品在线美女| 亚洲精品中文字幕在线视频| 悠悠久久av| 精品亚洲成国产av| 宅男免费午夜| 99国产精品免费福利视频| 亚洲伊人色综图| 亚洲精品在线美女| 妹子高潮喷水视频| 亚洲自偷自拍图片 自拍| 国产在视频线精品| 久久精品国产综合久久久| 午夜福利视频精品| 国产99久久九九免费精品| 国产成人精品久久二区二区免费| 王馨瑶露胸无遮挡在线观看| 亚洲国产精品一区三区| 多毛熟女@视频| 亚洲av男天堂| 日韩中文字幕欧美一区二区 | 中文字幕色久视频| 99香蕉大伊视频| 操美女的视频在线观看| 丰满人妻熟妇乱又伦精品不卡| 婷婷色综合www| 国产欧美日韩一区二区三区在线| 国产午夜精品一二区理论片| 欧美日韩综合久久久久久| 国产一区二区三区av在线| 欧美中文综合在线视频| 国产精品熟女久久久久浪| 国产精品一区二区在线观看99| 青春草视频在线免费观看| 老司机影院毛片| 免费观看人在逋| 天天操日日干夜夜撸| 亚洲黑人精品在线| 精品国产乱码久久久久久男人| 日本a在线网址| 日本a在线网址| 1024视频免费在线观看| 亚洲精品自拍成人| 成人黄色视频免费在线看| 丰满人妻熟妇乱又伦精品不卡| 亚洲精品日本国产第一区| 亚洲少妇的诱惑av| 精品国产乱码久久久久久男人| 亚洲av成人精品一二三区| 黄色怎么调成土黄色| 欧美成人午夜精品| 人妻一区二区av| 日韩av不卡免费在线播放| 一边亲一边摸免费视频| 校园人妻丝袜中文字幕| 久久精品久久精品一区二区三区| www.精华液| 久久午夜综合久久蜜桃| 亚洲情色 制服丝袜| 久久久久久久精品精品| 国产成人免费无遮挡视频| 日韩免费高清中文字幕av| 一边摸一边抽搐一进一出视频| 777久久人妻少妇嫩草av网站| 国精品久久久久久国模美| 国产成人免费无遮挡视频| 午夜91福利影院| 五月天丁香电影| 极品少妇高潮喷水抽搐| 午夜福利在线免费观看网站| 操出白浆在线播放| 午夜日韩欧美国产| 人人妻,人人澡人人爽秒播 | 爱豆传媒免费全集在线观看| 女人爽到高潮嗷嗷叫在线视频| 成人国产一区最新在线观看 | 日韩人妻精品一区2区三区| 黄色怎么调成土黄色| 在线观看国产h片| 欧美日韩亚洲国产一区二区在线观看 | 一区二区av电影网| 久久久亚洲精品成人影院| 只有这里有精品99| 最黄视频免费看| 18禁裸乳无遮挡动漫免费视频| 免费av中文字幕在线| 久久人人爽av亚洲精品天堂| 高清不卡的av网站| 97人妻天天添夜夜摸| 99热全是精品| 狠狠精品人妻久久久久久综合| 亚洲少妇的诱惑av| 亚洲成av片中文字幕在线观看| 99国产精品免费福利视频| 黑丝袜美女国产一区| av网站免费在线观看视频| 亚洲色图 男人天堂 中文字幕| 性少妇av在线| 天堂8中文在线网| 热re99久久国产66热| 老司机亚洲免费影院| 国产精品一二三区在线看| 韩国精品一区二区三区| 亚洲成色77777| 国产97色在线日韩免费| 国产精品久久久久成人av| 免费黄频网站在线观看国产| 90打野战视频偷拍视频| 美女大奶头黄色视频| 一级黄片播放器| 中文字幕精品免费在线观看视频| 老司机亚洲免费影院| 久久精品久久久久久久性| 久久人人97超碰香蕉20202| 搡老岳熟女国产| 午夜免费男女啪啪视频观看| 黄色一级大片看看| 满18在线观看网站| 国产亚洲精品久久久久5区| 欧美黑人精品巨大| 国产女主播在线喷水免费视频网站| 啦啦啦 在线观看视频| 午夜福利在线免费观看网站| 免费女性裸体啪啪无遮挡网站| 欧美精品人与动牲交sv欧美| 91国产中文字幕| 99久久精品国产亚洲精品| 制服人妻中文乱码| 久久人人爽av亚洲精品天堂| 国语对白做爰xxxⅹ性视频网站| 亚洲精品成人av观看孕妇| av线在线观看网站| 尾随美女入室| 亚洲国产欧美一区二区综合| 日本午夜av视频| 国产日韩一区二区三区精品不卡| 在线观看www视频免费| 国产在线视频一区二区| 午夜福利在线免费观看网站| 久久精品亚洲av国产电影网| 最新在线观看一区二区三区 | 精品少妇久久久久久888优播| 国产男人的电影天堂91| 午夜老司机福利片| 日本av免费视频播放| 18在线观看网站| 亚洲精品久久久久久婷婷小说| 免费观看av网站的网址| 国产精品二区激情视频| 欧美在线黄色| 国产日韩欧美亚洲二区| 久热这里只有精品99| 国产在线免费精品| 成人手机av| 久久久久久亚洲精品国产蜜桃av| 欧美久久黑人一区二区| 一本综合久久免费| 黑人巨大精品欧美一区二区蜜桃| 热99久久久久精品小说推荐| 国产精品av久久久久免费| 汤姆久久久久久久影院中文字幕| 免费看十八禁软件| 久久人妻熟女aⅴ| 久久毛片免费看一区二区三区| 18禁观看日本| 国产日韩欧美视频二区| 国产精品一区二区精品视频观看| 热re99久久国产66热| 老司机深夜福利视频在线观看 | 秋霞在线观看毛片| 青春草视频在线免费观看| 麻豆av在线久日| 久久久久国产一级毛片高清牌| 国产av国产精品国产| 亚洲专区国产一区二区| 国产视频一区二区在线看| 亚洲欧美日韩另类电影网站| 中文字幕制服av| 久久精品aⅴ一区二区三区四区| 丝袜美腿诱惑在线| 亚洲av成人精品一二三区| 亚洲色图 男人天堂 中文字幕| 婷婷丁香在线五月| 校园人妻丝袜中文字幕| 在现免费观看毛片| 自线自在国产av| 日本黄色日本黄色录像| 亚洲av男天堂| 另类精品久久| 日本a在线网址| 欧美大码av| 午夜精品国产一区二区电影| 国产精品av久久久久免费| 午夜久久久在线观看| 999精品在线视频| 午夜福利在线免费观看网站| 国产男女内射视频| 热re99久久精品国产66热6| 人人妻人人澡人人爽人人夜夜| 一级片'在线观看视频| 精品少妇一区二区三区视频日本电影| 国产一区二区三区综合在线观看| 日韩一卡2卡3卡4卡2021年| 9热在线视频观看99| 精品人妻熟女毛片av久久网站| 精品久久久久久电影网| 在线观看www视频免费| 大陆偷拍与自拍| 777米奇影视久久| 婷婷色综合www| 午夜视频精品福利| 久久人人97超碰香蕉20202| 男女无遮挡免费网站观看| 欧美精品高潮呻吟av久久| 黄色毛片三级朝国网站| 2021少妇久久久久久久久久久| 热99国产精品久久久久久7| 91字幕亚洲| 美女中出高潮动态图| 成人亚洲精品一区在线观看| 成年人黄色毛片网站| 制服诱惑二区| 精品国产一区二区三区久久久樱花| 色精品久久人妻99蜜桃| 欧美日韩一级在线毛片| 激情五月婷婷亚洲| 欧美精品啪啪一区二区三区 | 中国国产av一级| 又紧又爽又黄一区二区| 成人免费观看视频高清| 国产精品一区二区在线不卡| 乱人伦中国视频| 日韩人妻精品一区2区三区| 国产av国产精品国产| 青春草视频在线免费观看| 久9热在线精品视频| 婷婷色综合www| 亚洲av美国av| 精品久久蜜臀av无| 男人舔女人的私密视频| 国产男女超爽视频在线观看| 人人妻人人澡人人看| 脱女人内裤的视频| 狠狠精品人妻久久久久久综合| 在线精品无人区一区二区三| 欧美亚洲 丝袜 人妻 在线| 激情五月婷婷亚洲| 欧美日韩黄片免| 亚洲av美国av| bbb黄色大片| 啦啦啦中文免费视频观看日本| 国产无遮挡羞羞视频在线观看| 亚洲人成网站在线观看播放| 母亲3免费完整高清在线观看| 亚洲综合色网址| www日本在线高清视频| 亚洲国产毛片av蜜桃av| 亚洲激情五月婷婷啪啪| 久久青草综合色| 中文字幕亚洲精品专区| 成人亚洲欧美一区二区av| 色婷婷av一区二区三区视频| 亚洲国产欧美网| 乱人伦中国视频| 久久精品国产亚洲av高清一级| 日日摸夜夜添夜夜爱| 精品国产一区二区三区四区第35| 日韩伦理黄色片| 高清欧美精品videossex| 老汉色av国产亚洲站长工具| 国产亚洲午夜精品一区二区久久| 精品亚洲乱码少妇综合久久| 如日韩欧美国产精品一区二区三区| 在现免费观看毛片| 人体艺术视频欧美日本| 免费一级毛片在线播放高清视频 | 丰满饥渴人妻一区二区三| 深夜精品福利| 亚洲av国产av综合av卡| 极品少妇高潮喷水抽搐| 我的亚洲天堂| 热re99久久精品国产66热6| 菩萨蛮人人尽说江南好唐韦庄| 少妇人妻久久综合中文| 午夜激情久久久久久久| 色播在线永久视频| 中文字幕精品免费在线观看视频| 热re99久久国产66热| www日本在线高清视频| 亚洲av综合色区一区| 建设人人有责人人尽责人人享有的| 午夜精品国产一区二区电影| 亚洲国产精品国产精品| 免费看不卡的av| 中文字幕人妻丝袜制服| 91国产中文字幕| 成人国语在线视频| 久久综合国产亚洲精品| 亚洲欧美清纯卡通| 亚洲欧美日韩另类电影网站| 麻豆乱淫一区二区| av在线老鸭窝| 欧美人与善性xxx| 你懂的网址亚洲精品在线观看| 两性夫妻黄色片| 欧美日韩福利视频一区二区| 欧美另类一区| 99久久综合免费| 亚洲欧美激情在线| 国产日韩欧美亚洲二区| 欧美人与善性xxx| 狠狠婷婷综合久久久久久88av| 亚洲欧美日韩另类电影网站| 桃花免费在线播放| 欧美日韩成人在线一区二区| 看十八女毛片水多多多| 精品国产一区二区三区四区第35| 亚洲av综合色区一区| 99国产精品一区二区三区| 国产1区2区3区精品| 一区二区三区四区激情视频| 2018国产大陆天天弄谢| 日韩伦理黄色片| 天天躁夜夜躁狠狠久久av| 亚洲中文av在线| 亚洲成人国产一区在线观看 | 久久精品aⅴ一区二区三区四区| 亚洲第一青青草原| 你懂的网址亚洲精品在线观看| 欧美日韩一级在线毛片| 中文字幕色久视频| 一级毛片女人18水好多 | 国产三级黄色录像| 亚洲九九香蕉| 亚洲色图 男人天堂 中文字幕| xxx大片免费视频| 日日爽夜夜爽网站| a级毛片黄视频| 亚洲国产成人一精品久久久| 精品人妻在线不人妻| 精品国产一区二区三区久久久樱花| 国产黄频视频在线观看| 国产精品免费大片| 国产精品免费视频内射| 黄色一级大片看看| 这个男人来自地球电影免费观看| 在线av久久热| 久久综合国产亚洲精品| 欧美日韩av久久| 精品亚洲成a人片在线观看| 日韩,欧美,国产一区二区三区| 欧美精品一区二区大全| 七月丁香在线播放| 精品少妇久久久久久888优播| 又紧又爽又黄一区二区| 青春草视频在线免费观看| 国产一区二区三区综合在线观看| 日韩一本色道免费dvd| 国产一区亚洲一区在线观看| 国语对白做爰xxxⅹ性视频网站| 亚洲av综合色区一区| 国产不卡av网站在线观看| 一级毛片黄色毛片免费观看视频| 99香蕉大伊视频| 国产高清videossex| 欧美性长视频在线观看| 精品亚洲乱码少妇综合久久| 在线天堂中文资源库| 一级,二级,三级黄色视频| www.精华液| 一区二区三区乱码不卡18| 18禁观看日本| 另类亚洲欧美激情| 国产精品 国内视频| 又粗又硬又长又爽又黄的视频| 丝袜脚勾引网站| 搡老岳熟女国产| 国产男女超爽视频在线观看| 成人黄色视频免费在线看| 国产高清videossex| 国产黄色视频一区二区在线观看| 国产成人系列免费观看| 老司机影院毛片| 亚洲人成电影免费在线| 一区二区三区四区激情视频| 一区二区三区激情视频| 在线观看人妻少妇| 丝袜美腿诱惑在线| 两人在一起打扑克的视频| 91精品国产国语对白视频| 妹子高潮喷水视频| 亚洲成av片中文字幕在线观看| 天天躁夜夜躁狠狠久久av| 国产片内射在线| 一二三四在线观看免费中文在| 国产亚洲av片在线观看秒播厂| 国产男女超爽视频在线观看| 日本色播在线视频| 我的亚洲天堂| 欧美成人精品欧美一级黄| 一本一本久久a久久精品综合妖精| 亚洲精品自拍成人| 日本猛色少妇xxxxx猛交久久| 在线 av 中文字幕| 国产精品久久久av美女十八| 国产av精品麻豆| 亚洲av男天堂| 一本久久精品| av在线老鸭窝| 日韩av在线免费看完整版不卡| 中国国产av一级| 在线观看www视频免费| 一二三四社区在线视频社区8| 欧美日韩视频精品一区| 制服人妻中文乱码| 免费在线观看视频国产中文字幕亚洲 | 男女床上黄色一级片免费看| 亚洲色图 男人天堂 中文字幕| 国产免费视频播放在线视频| 91成人精品电影| 免费观看人在逋| 成人三级做爰电影| 欧美精品亚洲一区二区| avwww免费| 天天躁夜夜躁狠狠久久av| 欧美大码av| 欧美成人午夜精品| 亚洲专区中文字幕在线| 成人亚洲欧美一区二区av| 日韩一卡2卡3卡4卡2021年| 丝袜美足系列| 老司机深夜福利视频在线观看 | 国产男人的电影天堂91| 老司机影院成人| 精品视频人人做人人爽| 欧美精品人与动牲交sv欧美| 一边亲一边摸免费视频| 国产精品免费视频内射| 捣出白浆h1v1| 国产国语露脸激情在线看| 亚洲av片天天在线观看| 新久久久久国产一级毛片| 男女免费视频国产| 丰满人妻熟妇乱又伦精品不卡| 国产在视频线精品| 黄色 视频免费看| 久久人妻福利社区极品人妻图片 | 国产野战对白在线观看| 国产精品 国内视频| 亚洲成人手机| 国产精品国产av在线观看| 人成视频在线观看免费观看| 91九色精品人成在线观看| 欧美在线黄色| 青草久久国产| 一区二区三区乱码不卡18| 蜜桃在线观看..| 久久久久久亚洲精品国产蜜桃av| 一级黄色大片毛片| 精品人妻熟女毛片av久久网站| 最新在线观看一区二区三区 | 丝袜喷水一区| 精品免费久久久久久久清纯 | 亚洲情色 制服丝袜| 波多野结衣av一区二区av| 高清黄色对白视频在线免费看| 亚洲精品国产色婷婷电影| 男人舔女人的私密视频| 亚洲国产精品一区二区三区在线| 我的亚洲天堂| av在线老鸭窝| 中文欧美无线码| 欧美性长视频在线观看| 老汉色av国产亚洲站长工具| 18禁裸乳无遮挡动漫免费视频| 国产亚洲av高清不卡| 色94色欧美一区二区| 丰满人妻熟妇乱又伦精品不卡| 中文字幕制服av| 一级,二级,三级黄色视频| 欧美乱码精品一区二区三区| 80岁老熟妇乱子伦牲交| 在线观看免费午夜福利视频| 亚洲欧美精品综合一区二区三区| 国产一区二区激情短视频 | 亚洲av成人精品一二三区| 麻豆乱淫一区二区| 叶爱在线成人免费视频播放| 黄片播放在线免费| 极品少妇高潮喷水抽搐| 香蕉国产在线看| 久久精品亚洲熟妇少妇任你| 久久精品人人爽人人爽视色| 国产男女内射视频| 中文精品一卡2卡3卡4更新| 国产亚洲av高清不卡| 亚洲七黄色美女视频| 国精品久久久久久国模美| 国产精品久久久人人做人人爽| 亚洲欧美成人综合另类久久久| 手机成人av网站| 大香蕉久久网| 亚洲精品国产一区二区精华液| 乱人伦中国视频| 日韩av在线免费看完整版不卡| 夜夜骑夜夜射夜夜干| 老汉色av国产亚洲站长工具| 亚洲欧洲日产国产| 女人高潮潮喷娇喘18禁视频| 国产精品成人在线| 在线 av 中文字幕| 波野结衣二区三区在线| 久久青草综合色| 热99久久久久精品小说推荐| 七月丁香在线播放| 亚洲精品国产区一区二| 国产成人精品久久二区二区91| 亚洲国产欧美网| 亚洲av美国av| 成年女人毛片免费观看观看9 | 成人国产av品久久久| 久久久国产欧美日韩av| 日本色播在线视频| 久热爱精品视频在线9| 国产精品一区二区在线观看99| 国产一区二区 视频在线| 国产真人三级小视频在线观看| 欧美人与性动交α欧美精品济南到| 狂野欧美激情性bbbbbb| 免费观看人在逋| 日日爽夜夜爽网站| 久久天堂一区二区三区四区| 久久久久国产一级毛片高清牌| 伊人久久大香线蕉亚洲五| 香蕉国产在线看| 九色亚洲精品在线播放| 亚洲七黄色美女视频| 久久精品国产综合久久久| 欧美激情极品国产一区二区三区| 中国国产av一级| 日韩电影二区| 国产精品亚洲av一区麻豆| 狠狠精品人妻久久久久久综合| 国产精品成人在线| 狂野欧美激情性bbbbbb| 欧美在线一区亚洲| 欧美性长视频在线观看| 精品国产一区二区三区久久久樱花| 久久久久久久久免费视频了| 国产不卡av网站在线观看| 午夜久久久在线观看| av天堂在线播放| 汤姆久久久久久久影院中文字幕| 欧美精品av麻豆av| 亚洲成人手机| 久久国产精品影院| 每晚都被弄得嗷嗷叫到高潮| 狠狠精品人妻久久久久久综合| 中国国产av一级| av片东京热男人的天堂| 校园人妻丝袜中文字幕| 女人久久www免费人成看片| 成人18禁高潮啪啪吃奶动态图| 丝袜喷水一区| 成年女人毛片免费观看观看9 | 国产成人免费无遮挡视频| 国产av一区二区精品久久| av片东京热男人的天堂| 久久精品国产a三级三级三级| 日本av免费视频播放| 免费看av在线观看网站| 成人手机av| 亚洲成人免费电影在线观看 | 精品国产国语对白av| 国产在线免费精品| 国产成人免费无遮挡视频| 在线天堂中文资源库| 国产成人av激情在线播放| 国产91精品成人一区二区三区 | 波多野结衣av一区二区av| 日韩一卡2卡3卡4卡2021年| 亚洲av在线观看美女高潮| 99香蕉大伊视频| 亚洲国产欧美日韩在线播放| 免费人妻精品一区二区三区视频| 久热爱精品视频在线9| 丝袜人妻中文字幕| 少妇 在线观看| 免费人妻精品一区二区三区视频| 性色av一级| 国产精品一国产av| 亚洲中文字幕日韩| 久久精品亚洲av国产电影网| 少妇粗大呻吟视频| 久久精品成人免费网站| 国产精品一二三区在线看| 麻豆国产av国片精品| 欧美亚洲日本最大视频资源| 一区在线观看完整版| 亚洲国产欧美日韩在线播放| 99精品久久久久人妻精品| 久久久久久久大尺度免费视频| 欧美成狂野欧美在线观看| 每晚都被弄得嗷嗷叫到高潮| 黄色片一级片一级黄色片|