• <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 ?

    A Physical Layer Algorithm for Estimation of Number of Tags in UHF RFID Anti-Collision Design

    2019-11-07 03:13:06ZhongHuangJianSuGuangjunWenWenxianZhengChuChuYijunZhang4andYiboZhang
    Computers Materials&Continua 2019年10期

    Zhong HuangJian SuGuangjun WenWenxian ZhengChu ChuYijun Zhang4 and Yibo Zhang

    Abstract:A priori knowledge of the number of tags is crucial for anti-collision protocols in slotted UHF RFID systems.The number of tags is used to decide optimal frame length in dynamic frame slotted ALOHA(DFSA)and to adjust access probability in random access protocols.Conventional researches estimate the number of tags in MAC layer based on statistics of empty slots,collided slots and successful slots.Usually,a collision detection algorithm is employed to determine types of time slots.Only three types are distinguished because of lack of ability to detect the number of tags in single time slot.In this paper,a physical layer algorithm is proposed to detect the number of tags in a collided slot.Mean shift algorithm is utilized,and some properties of backscatter signals are investigated.Simulation results verify the effectiveness of the proposed solution in terms of low estimation error with a high SNR range,outperforming the existing MAC layer approaches.

    Keywords:UHF RFID,anti-collision,cluster algorithm.

    1 Introduction

    Anti-collision algorithms are carried out in multi-access UHF RFID systems to reduce collisions as well as to increase channel efficiency.As one of the most popular anticollision algorithms for RFID system,dynamic framed slotted ALOHA(DFSA)algorithm employs a mechanism similar with time division multiple access(TDMA).Synchronized frame is divided into several time slots for random access.Reader dynamically adjusts the number of time slots based on the estimated number of tags.It is well known that DFSA reaches the optimal system throughput when the number of slots equals to the number of tags waiting to be identified.There are two main problems in DFSA design,estimation of the number of tags and frame length adjustment[Chen,Liu,Ma et al.(2018)].

    Various researches are proposed to improve the estimation accuracy so as to improve the channel efficiency of RFID system.The number of tags is estimated by multiplying the number of collided slots in a frame by the expected number of tags per collided slot(=2.39),which is a constant for all frames regardless of number of tags and frame size[Schoute(1983)].In the mechanisms proposed by Vogt et al.[Vogt(2002);Chen and Lin(2006);Cha and Kim(2005)],estimation of the number of tags is based on successful,collided,and empty probability in a frame.Probability theory is utilized in these works.The number of tags is estimated in Chen[Chen(2009)]by multiplying the number of collided time slots by a well-defined factor,which is found by an iterative algorithm.A posteriori probability distribution-based method is proposed in Eom et al.[Eom and Lee(2010)]to further improve the accuracy.However,the computation complexity is higher than the others.A Bayesian method[Annur,Srichavengsup,Nakpeerayuth et al.(2015)]is used to update the posterior probability distribution of number of users' slot by slot so as to estimate the number of tags.

    Since RFID readers usually adopt in-phase and quadrature information of tag signals,tag recovery and estimation methods of the number of tags based on signal processing are developed to solve collision problem.Tag recovery is able to turn a collided slot into a successful slot while estimation of the number of tags based on physical layer process is able to enhance performance of anti-collision algorithm in upper layer.Most researches require multiple antennas receiver[Angerer,Langwieser and Rupp(2010)],specific tag signal strength[Fyhn,Jacobsen,Popovski et al.(2011)],modified coding mechanisms[Parks,Liu,Gollakota et al.(2014)],etc.Bipartite Grouping(BiGroup)[Ou,Li and Zheng(2017)]is the first one proposed to parallelly decode multiple CTOS tags with the help of both time domain and constellation domain information in physical layer.Algorithms based on signal processing in time domain are proposed to reduce SNR requirements.Collided signals are transformed to time-scale domains and LS criterion is utilized for tag signal separation in Zeng et al.[Zeng,Wu,Yang et al.(2017)].An edge transition scheme is proposed to recover collision and decode tag signals in Benbaghdad et al.[Benbaghdad,Fergani and Tedjini(2016)].These works focus on tag signal recovery issues in physical layer and pay less attention on MAC layer design.Tan et al.[Tan,Wang,Fu et al.(2018)]proposed a collision detection and signal recovery method and combine them with DFSA algorithm.Optimal frame length is calculated based on a collision recovery probability coefficient,which is obtained by simulations of its physical layer design.A novel closed-form solution is further proposed in Ahmed et al.[Ahmed,Salah,Robert et al.(2018)]for optimal FSA frame length decision,in which collision recovery probabilities are provided arbitrarily.An accurate tag number estimation algorithm is still needed and not addressed.

    In this paper,we propose to estimate the number of tags in a collided time slot in physical layer.Different from the existing work,we focus on improving estimation accuracy under CTOS tag assumption.Estimation error is reduced compared with MAC layer design and SNR range is expanded compared with current physical layer design.In this scheme,the number of tags in single time slot is determined by the number of clusters located in constellation domain.Signal samples are firstly scaled based on the baseband noise level.After that,mean shift algorithm is utilized to divide the data into several clusters.In the end,a cluster adjustment process is carried out to get better performance.Simulation results verify the effectiveness of the proposed solution in terms of low estimation errors and high SNR range,outperforming the existing MAC layer-based approaches.This paper is organized as following.Section II describes signal model of UHF RFID system and drawback of clustering algorithms.After that,design consideration and cluster-based algorithm is proposed in Section III,which is evaluated with numeric simulations in Section IV.Finally,conclusion follows.

    2 Signal model and cluster algorithm

    2.1 Backscatter signal and its distribution

    In a UHF RFID system,reader energizes tags by transmitting continuous wave.Passive tags backscatter the radio to communicate with the reader.After that,backscatter signals are down converted in the reader side.As shown in Fig.1(a),baseband signal in receive path of a reader consists of three components,backscatter signal of tags,self-jammer and noise.Eq.(1)shows the detailed format of those signals.

    The first one indicates signals of multiple tags,the second one indicates the signals of self-jammer caused by RFID system.Where i indicates the index of collided tags in the same time slot,Aiandθiare respective transmitted data and initial phase of the i-th tag signal.θ0andμis the amplitude and phase of self-jammer.n(t)represents white noise in the receiver.

    Figure 1:RFID framework and backscatter signal model

    Backscatter signals have both in-phase and quadrature components,with which we could plot the signal samples in a two-dimension coordinates.Every sample is represented by a point on the I-Q plane.Samples of the same transmitting status are dispersed and scattered around a centroid position,forming a cluster.As shown in Fig.1(b),there are 4 clusters,representing 4 transmitting status of 2 tags.The number of clusters is decided by the number of tags.Every tag has two transmission status,the size of the full status space is 2n,where n is the number of tags.Three properties could be obtained from Eq.(1).

    1.Since noise in in-phase axis and in quadrature axis are different due to the existence of self-jammer,every cluster is shaped as an eclipse.

    2.We assume that noise in both axes obeys gaussian distribution.As a result,around 95% samples of one cluster are located in a circle of radius of two times standard deviation of noise.The higher the noise is,the bigger the circle is.When SNR goes down,the radius gets bigger,and when it is larger than the distance between cluster centers,the clusters overlap.Fig.1(c)shows an example when clusters are overlapped.

    3.The clusters are always located in pairs,symmetrizing to the center of all samples.For every cluster,reverse all the tag status,a symmetric cluster is obtained.Their symmetric center is located at the center of the whole graph.Eq.(2)shows the coordinates of the center point,wherensam,xcenterandycenterdenotes the number of samples,x axis coordinate and y axis coordinate.

    2.2 Density based clustering algorithm

    A straight forward way to find the number of tags is to divide signal samples into clusters with a cluster algorithm.Mean shift algorithm and DBSCAN algorithm are both widely used method to identify multiple dimension data without a prior knowledge of the number of clusters.They both make use of sample density to make decisions.Mean shift algorithm updates every cluster center based on the vector sum of all samples in it until convergence.Bandwidth is setup to decide the range of clusters.On the other hand,DBSCAN choose samples with large number of neighbors as core samples.Connection between core samples are calculated and clusters are divided.Parameters distance is required to decide neighbors and minimum points are required to decide core samples.

    Table 1:Parameter settings of cluster algorithms

    Both algorithms are tested with simulations.Parameter settings are shown in Tab.1.Fig.(2)shows the cluster division of two algorithms.Samples in one cluster are encircled by a circle.DBSCAN divides the signals into 6 clusters while mean shift divides them into more than 20 clusters.

    Figure 2:Performance of two algorithm in different SNR scenarios

    3 Physical layer algorithm for estimation of the number of tags in single time slot

    3.1 Design consideration

    Mean shift algorithm and DBSCAN are available for clustering in some cases with proper parameter settings.However,both algorithms are not designed for detection of the number of tags.DBSCAN requires uniform distributed samples in one cluster,which is not preferred in our case.Density of samples in one cluster decreases along with distance to the center.Furthermore,it is not able to identify overlapped clusters.Mean shift algorithm outperforms DBSCAN in low SNR scenario.Cluster centers are updated towards a denser direction,it always finds the densest points.However,isolated samples are identified as a cluster in some cases.In this paper,we make use of samples distribution information and its properties to improve performance of mean shift algorithm.The improvement is based on the following considerations.

    First of all,our purpose is to find number of clusters,not accurate cluster division.Wrong assignment to clusters may cause bit error in signal recovery case,but not in number detection case.As a result,we only consider a small core area of clusters,which is defined as samples within distance of two times standard deviation of noise.Find it and we get a valid cluster center.The noise samples are ignored naturally.

    Second,as described in Section II,all clusters are shaped like an eclipse.On the meantime,mean shift algorithm calculates updated vector based on a Euclid distance,which means samples in a perfect circle are all considered.It brings a big performance decrease.It is fitting and proper to adjust the scale of in-phase and quadrature signal magnate by noise level.After that,the core area of clusters is formed as a perfect circle.It is better for identification.

    Finally,clusters are distributed in pairs.Every cluster have a symmetrical one with similar number of samples.Their cluster centers are also symmetrical to the whole center of samples.This property makes us able to discard isolated noise cluster identified by mean shift algorithm.

    3.2 Algorithm details

    The detailed algorithm is shown as Algorithm.1,we first transform samples to make clusters shaped as perfect circle other than eclipse.After that,mean shift algorithm is carried out.Some random points are selected as initial centers.These centers are updated based on samples in their neighborhood.After they converge,an adjustment scheme is carried out to discard isolated noise clusters.

    4 Simulation and performance evaluation

    In this section,we evaluate the performance of proposed algorithm under different scenarios.Success rate is firstly proposed for accuracy of detecting number of tags to evaluate performance in single time slot scenario.After that,total success rate and estimation error are derived based on probability to evaluate performance in multiple time slots scenario.Success rate is compared with DBSCAN algorithm and estimation error is compared with probability-based methods.

    4.1 Performance in single time slot scenario

    The performance of proposed algorithm is evaluated by accuracy or success rate,which is defined as the number of successful experiments over the total number of experiments.In order to improve reliability of simulations,simulations are carried out multiple times in different SNR scenarios.Furthermore,scenarios with different number of tags are evaluated separately due different performance in these cases.Scenarios when the number of tags is larger than 4 is not considered here because it rarely happens in practice.

    The simulation runs as the following steps.

    In the first step,we initialize the system parameters,i.e.,number of tags and SNR.

    In the second step,10000 experiments are executed.In each experiment,pseudo signals are generated based on Eq.(1).Signal strength and initial phase of tags are randomly selected in every experiment.Proposed algorithm and DBSCAN are used to determine the number of tags.Both actual number of tags and determined number of tags are recorded for performance evolution.

    In the third step,switch to the next parameter and execute Step 2 for another time.

    In the fourth step,success rate in each system parameter set are calculated.Before that,a performance indicatorbased on conditional probability is calculated by real number and determined number,as shown in Eq.(5).Wheredenotes detected number of tags whilendenotes actual number of tags.It is apparent that success rate when the number of tags is n equals probability ofp(n|n).

    Fig.3 shows success rate of proposed algorithm in different conditions of SNR and the number of tags.Success rate of proposed algorithm is greater than DBSCAN in all conditions.When SNR is larger than 18 dB,success rate of proposed algorithm is larger than 0.9 in 2 and 3 tags scenarios.Success rate is relatively lower when there are 4 tags,still over 0.8 when SNR is large enough.

    Figure 3:Success rate comparison in different conditions

    4.2 Performance of multiple time slots

    Performance of proposed algorithm in multiple time slots is evaluated by two indicators,total success rate and estimation error.When there are multiple time slots and unknown number of unidentified tags,number of tags in one time slot follows a binomial distribution,as shown in Eq.(6).

    whereB(r)denotes the probability ofrtags in one slot,ndenotes number of tags to be identified in the read range,Ldenotes frame length,i.e.,number of time slots.

    Total success rate takes distribution of number of tags in one slot into consideration,which is defined as Eq.(7).Total success rate shows an average performance under specific condition of frame length and the number of tags.

    Similar with other estimation researches,estimation error is a good indicator for performance evaluation.Here it is defined in a probabilistic way in Eq.(8).

    whereE(n)denotes average number of tags in one time slot,whiledenotes average estimated number of tags in every time slot.It is calculated by Eq.(9).

    wheredenotes expectation of estimated number of tags on condition of the number of tags in one time slot,shown as Eq.(10).

    Figure 4:Total success rate in different conditions of the number of tags

    Fig.4 shows total success rate in different conditions of the number of tags when number of time slots are set to 128.Success Rate decreases along with the number of tags increases because possibility is larger when the number of tags is higher.Apparently,SNR effects total success rate.Total success rate decreases to 0.5 when SNR is 15 dB and the number of tags reaches 300.However,total success rate is higher than 0.8 in most high SNR conditions(higher than 20).

    Figure 5:Estimation error under different number of tags and SNR conditions

    Fig.5 shows estimation error comparison between proposed algorithm and two other methods.Estimation error of our proposal decreases while SNR goes up.In 15 dB SNR condition,our proposal beats S+2.39C method when the number of tags is more than 220.In 20dB SNR condition,estimation error of our proposal is close to method proposed in Chen.(2009).Estimation error gets lower when SNR is higher than 20 and outperforms both two methods.

    5 Conclusion

    This paper focuses on tag number estimation method for RFID anti-collision purpose.A clustering algorithm is proposed to detect the number of tags in physical layer.The proposed algorithm makes full use of in-phase and quadrature information to get better performance in low SNR scenarios.Simulation results show that it is good in a large range of SNR conditions.It is better than DBSCAN in terms of success rate and better than probability-based methods in terms of estimation errors.

    Acknowledgement:This work was supported in part by the National Natural Science Foundation of China under project contracts[NOS.61601093,61791082,61701116,61371047],in part by Sichuan Provincial Science and Technology Planning Program of China under project contracts No.2016GZ0061 and No.2018HH0044,in part by Guangdong Provincial Science and Technology Planning Program of China under project contracts No.2015B090909004 and No.2016A010101036,in part by the fundamental research funds for the Central Universities under project contract No.ZYGX2016Z011,and in part by Science and Technology on Electronic Information Control Laboratory.

    久久人妻熟女aⅴ| 国产亚洲精品久久久久5区| 久久 成人 亚洲| 精品免费久久久久久久清纯 | 男女免费视频国产| 日韩中文字幕欧美一区二区| 国产精品免费大片| 久久亚洲真实| 在线永久观看黄色视频| 亚洲熟女毛片儿| 欧美日韩视频精品一区| 日韩欧美一区视频在线观看| 人人澡人人妻人| 中文欧美无线码| 国产成人啪精品午夜网站| 看片在线看免费视频| 黄色片一级片一级黄色片| 亚洲性夜色夜夜综合| 日韩欧美一区二区三区在线观看 | av不卡在线播放| 国产高清国产精品国产三级| 波多野结衣一区麻豆| 一区二区日韩欧美中文字幕| 咕卡用的链子| 国产精品 国内视频| netflix在线观看网站| 黄频高清免费视频| 国产99久久九九免费精品| 午夜福利在线免费观看网站| 91字幕亚洲| 久久午夜综合久久蜜桃| 国产精品久久视频播放| 在线免费观看的www视频| 久热爱精品视频在线9| 亚洲综合色网址| 19禁男女啪啪无遮挡网站| 一二三四在线观看免费中文在| 久久天躁狠狠躁夜夜2o2o| 少妇 在线观看| 黄色怎么调成土黄色| 久热爱精品视频在线9| 国产成人av教育| 久久久国产精品麻豆| netflix在线观看网站| 午夜福利免费观看在线| 午夜两性在线视频| 国产日韩一区二区三区精品不卡| 久久久久久久久久久久大奶| 国产精品久久久人人做人人爽| 免费观看精品视频网站| 日本撒尿小便嘘嘘汇集6| 99国产精品一区二区蜜桃av | 老鸭窝网址在线观看| 亚洲伊人色综图| 黑人操中国人逼视频| 久久久国产一区二区| 欧美色视频一区免费| 99re6热这里在线精品视频| 成人国产一区最新在线观看| 国产成人一区二区三区免费视频网站| 捣出白浆h1v1| 国产精品免费视频内射| 久久人妻熟女aⅴ| 精品一品国产午夜福利视频| 天堂俺去俺来也www色官网| 亚洲精品久久午夜乱码| a在线观看视频网站| 亚洲av电影在线进入| www.自偷自拍.com| 夜夜躁狠狠躁天天躁| 精品一区二区三区四区五区乱码| 别揉我奶头~嗯~啊~动态视频| 久久久国产精品麻豆| 成人国产一区最新在线观看| 黄色毛片三级朝国网站| 香蕉国产在线看| 国产一区有黄有色的免费视频| 色婷婷久久久亚洲欧美| 国产精品久久电影中文字幕 | 精品免费久久久久久久清纯 | 久久久精品国产亚洲av高清涩受| 中亚洲国语对白在线视频| 99在线人妻在线中文字幕 | 99国产综合亚洲精品| tube8黄色片| 侵犯人妻中文字幕一二三四区| 9色porny在线观看| svipshipincom国产片| 性色av乱码一区二区三区2| 国产91精品成人一区二区三区| av在线播放免费不卡| 亚洲少妇的诱惑av| 精品国产亚洲在线| 亚洲国产欧美网| 国产麻豆69| 视频区欧美日本亚洲| 少妇被粗大的猛进出69影院| 国产精品亚洲一级av第二区| 最新在线观看一区二区三区| 免费看十八禁软件| 岛国毛片在线播放| 午夜福利一区二区在线看| 熟女少妇亚洲综合色aaa.| 黄频高清免费视频| 日韩成人在线观看一区二区三区| 手机成人av网站| 中文亚洲av片在线观看爽 | 亚洲精品在线美女| 色综合欧美亚洲国产小说| 国产精品久久视频播放| 女人久久www免费人成看片| 超碰97精品在线观看| 欧美大码av| 久久草成人影院| 母亲3免费完整高清在线观看| 精品人妻在线不人妻| 中文亚洲av片在线观看爽 | 性色av乱码一区二区三区2| 99国产极品粉嫩在线观看| 三上悠亚av全集在线观看| 亚洲va日本ⅴa欧美va伊人久久| 99国产精品99久久久久| 日本撒尿小便嘘嘘汇集6| 天天添夜夜摸| 女人被躁到高潮嗷嗷叫费观| 中文字幕av电影在线播放| 高清av免费在线| 人成视频在线观看免费观看| 国产在线一区二区三区精| 成人国产一区最新在线观看| 男女午夜视频在线观看| 老熟妇仑乱视频hdxx| 757午夜福利合集在线观看| xxxhd国产人妻xxx| 欧美日韩视频精品一区| 欧美激情高清一区二区三区| 国产淫语在线视频| 亚洲精品在线观看二区| 国产亚洲一区二区精品| 岛国在线观看网站| 窝窝影院91人妻| 色尼玛亚洲综合影院| 十分钟在线观看高清视频www| 母亲3免费完整高清在线观看| 天堂√8在线中文| 中文字幕人妻丝袜制服| 中亚洲国语对白在线视频| 国产麻豆69| 午夜精品在线福利| 91av网站免费观看| 日韩熟女老妇一区二区性免费视频| 1024香蕉在线观看| 中文字幕制服av| 亚洲va日本ⅴa欧美va伊人久久| av不卡在线播放| 亚洲一区中文字幕在线| 亚洲中文日韩欧美视频| 国产精品一区二区在线不卡| 亚洲欧美日韩高清在线视频| 欧美乱码精品一区二区三区| 久久影院123| 91麻豆av在线| 夜夜夜夜夜久久久久| 后天国语完整版免费观看| 人成视频在线观看免费观看| 欧美激情极品国产一区二区三区| 精品久久久久久,| 建设人人有责人人尽责人人享有的| 午夜免费鲁丝| 欧洲精品卡2卡3卡4卡5卡区| 热99re8久久精品国产| 中文字幕人妻丝袜制服| 亚洲精品粉嫩美女一区| 99re6热这里在线精品视频| 国产精品香港三级国产av潘金莲| 免费在线观看影片大全网站| 黄网站色视频无遮挡免费观看| 一级毛片高清免费大全| 人妻久久中文字幕网| 大片电影免费在线观看免费| 色播在线永久视频| 在线十欧美十亚洲十日本专区| 不卡一级毛片| av中文乱码字幕在线| 国产精品二区激情视频| 日本a在线网址| 人人妻人人澡人人看| 免费在线观看完整版高清| 欧美老熟妇乱子伦牲交| 久久国产精品大桥未久av| 99国产精品一区二区蜜桃av | 亚洲 欧美一区二区三区| 一级毛片女人18水好多| 久久精品国产综合久久久| 免费一级毛片在线播放高清视频 | 热99久久久久精品小说推荐| 丝袜在线中文字幕| 美女视频免费永久观看网站| 中文亚洲av片在线观看爽 | 另类亚洲欧美激情| 成人国产一区最新在线观看| 两人在一起打扑克的视频| 男人的好看免费观看在线视频 | 亚洲 欧美一区二区三区| 亚洲第一欧美日韩一区二区三区| 免费不卡黄色视频| 18在线观看网站| 在线看a的网站| 欧美激情高清一区二区三区| 51午夜福利影视在线观看| 大香蕉久久网| 国产精品国产av在线观看| 午夜激情av网站| 99国产精品99久久久久| av电影中文网址| 热99国产精品久久久久久7| 国产精品一区二区精品视频观看| 超碰成人久久| 亚洲一区中文字幕在线| 亚洲专区国产一区二区| 国产精品久久久久久精品古装| 国产亚洲精品第一综合不卡| 久久国产乱子伦精品免费另类| 成人国产一区最新在线观看| 嫩草影视91久久| 黄色成人免费大全| 亚洲三区欧美一区| 亚洲三区欧美一区| 黄色怎么调成土黄色| 亚洲自偷自拍图片 自拍| 欧美亚洲日本最大视频资源| 女人被躁到高潮嗷嗷叫费观| 亚洲精品成人av观看孕妇| 亚洲国产欧美日韩在线播放| 12—13女人毛片做爰片一| 亚洲专区中文字幕在线| 啦啦啦免费观看视频1| 欧美午夜高清在线| 国产一卡二卡三卡精品| 女性生殖器流出的白浆| 啪啪无遮挡十八禁网站| 午夜福利在线观看吧| 久久香蕉国产精品| 久久国产精品影院| 成年人午夜在线观看视频| 青草久久国产| 国产精品久久久人人做人人爽| 国产精品99久久99久久久不卡| 国产亚洲精品一区二区www | 99热网站在线观看| 国产精品偷伦视频观看了| 亚洲一码二码三码区别大吗| 欧美午夜高清在线| 日韩 欧美 亚洲 中文字幕| 国产无遮挡羞羞视频在线观看| 国产1区2区3区精品| 婷婷丁香在线五月| bbb黄色大片| 久久久国产精品麻豆| 国产一区二区三区综合在线观看| 另类亚洲欧美激情| 午夜福利欧美成人| 亚洲专区中文字幕在线| 国产亚洲一区二区精品| 9191精品国产免费久久| 亚洲avbb在线观看| 啪啪无遮挡十八禁网站| 岛国毛片在线播放| 国产精品免费一区二区三区在线 | 天堂中文最新版在线下载| 国产淫语在线视频| 亚洲精品中文字幕在线视频| av电影中文网址| 久久精品成人免费网站| 欧美日本中文国产一区发布| 国产精品.久久久| 国产亚洲精品久久久久5区| 久久精品人人爽人人爽视色| 欧美精品高潮呻吟av久久| 十八禁高潮呻吟视频| 欧美日韩av久久| 国产高清国产精品国产三级| 午夜精品久久久久久毛片777| 久久性视频一级片| 狠狠婷婷综合久久久久久88av| cao死你这个sao货| 国产日韩欧美亚洲二区| 十八禁高潮呻吟视频| 麻豆成人av在线观看| 成人特级黄色片久久久久久久| 精品高清国产在线一区| 又黄又粗又硬又大视频| 亚洲中文日韩欧美视频| 日日摸夜夜添夜夜添小说| 精品国内亚洲2022精品成人 | 久久精品国产亚洲av高清一级| 国产一卡二卡三卡精品| 99国产极品粉嫩在线观看| 丁香欧美五月| 成年动漫av网址| 国产在线一区二区三区精| 亚洲全国av大片| 男人舔女人的私密视频| 50天的宝宝边吃奶边哭怎么回事| 亚洲avbb在线观看| 国产单亲对白刺激| 女警被强在线播放| 免费观看a级毛片全部| 一区二区三区精品91| 成人手机av| 亚洲av美国av| 国产精品久久久av美女十八| 建设人人有责人人尽责人人享有的| 黄色成人免费大全| 亚洲欧美激情在线| 999精品在线视频| 日本黄色视频三级网站网址 | 国产精品免费视频内射| xxx96com| 最新在线观看一区二区三区| 久久亚洲真实| 成人手机av| 久久天躁狠狠躁夜夜2o2o| 天堂√8在线中文| 大香蕉久久网| 精品国产一区二区三区久久久樱花| 国产伦人伦偷精品视频| 叶爱在线成人免费视频播放| 男女床上黄色一级片免费看| 免费黄频网站在线观看国产| 80岁老熟妇乱子伦牲交| 国产91精品成人一区二区三区| videosex国产| 久久中文字幕人妻熟女| 久久中文字幕人妻熟女| 欧美精品人与动牲交sv欧美| 色精品久久人妻99蜜桃| 亚洲av成人av| 人人妻人人澡人人爽人人夜夜| 亚洲人成77777在线视频| 精品一品国产午夜福利视频| 欧美黄色片欧美黄色片| 久久国产乱子伦精品免费另类| 国产又爽黄色视频| 久久国产精品男人的天堂亚洲| 男人舔女人的私密视频| 国产精品综合久久久久久久免费 | 日本欧美视频一区| 精品一品国产午夜福利视频| 亚洲精品一二三| 十八禁网站免费在线| 久久人妻福利社区极品人妻图片| 香蕉丝袜av| 免费高清在线观看日韩| 天堂√8在线中文| 一个人免费在线观看的高清视频| 久久精品人人爽人人爽视色| videosex国产| 国产成人精品在线电影| 国产一卡二卡三卡精品| 美女午夜性视频免费| 巨乳人妻的诱惑在线观看| 免费看a级黄色片| 搡老岳熟女国产| 亚洲第一av免费看| 高清视频免费观看一区二区| 久久久久久久久久久久大奶| 日本wwww免费看| 欧美大码av| 男男h啪啪无遮挡| 色老头精品视频在线观看| 亚洲三区欧美一区| 国产精品亚洲一级av第二区| 好男人电影高清在线观看| 天天影视国产精品| 精品一区二区三卡| 另类亚洲欧美激情| 岛国毛片在线播放| 日韩欧美免费精品| 国产精品1区2区在线观看. | 久久久国产欧美日韩av| 精品少妇一区二区三区视频日本电影| 中文亚洲av片在线观看爽 | 新久久久久国产一级毛片| 波多野结衣av一区二区av| 在线观看一区二区三区激情| 搡老熟女国产l中国老女人| 老熟妇乱子伦视频在线观看| 婷婷精品国产亚洲av在线 | 久久99一区二区三区| 久久天躁狠狠躁夜夜2o2o| 制服诱惑二区| 欧美日韩福利视频一区二区| 国产精品美女特级片免费视频播放器 | 在线观看免费视频日本深夜| 亚洲成人免费电影在线观看| 18禁观看日本| 亚洲av成人不卡在线观看播放网| 一级a爱片免费观看的视频| 中文字幕人妻丝袜一区二区| 黄频高清免费视频| 男女免费视频国产| 久久精品亚洲av国产电影网| 成人黄色视频免费在线看| 欧美一级毛片孕妇| 色综合欧美亚洲国产小说| 最近最新中文字幕大全电影3 | 精品国产乱子伦一区二区三区| 手机成人av网站| 国产亚洲精品第一综合不卡| 久久国产乱子伦精品免费另类| 精品一区二区三卡| 色94色欧美一区二区| 一二三四社区在线视频社区8| 精品久久久久久久毛片微露脸| 亚洲av日韩在线播放| 国产精品二区激情视频| 亚洲av美国av| 欧美日本中文国产一区发布| 男女下面插进去视频免费观看| 老鸭窝网址在线观看| 在线观看免费日韩欧美大片| 成人黄色视频免费在线看| 精品无人区乱码1区二区| 黑人猛操日本美女一级片| 亚洲色图综合在线观看| 午夜日韩欧美国产| 女性被躁到高潮视频| 纯流量卡能插随身wifi吗| 午夜福利欧美成人| 美女高潮到喷水免费观看| 51午夜福利影视在线观看| av一本久久久久| 亚洲熟女毛片儿| 日本欧美视频一区| 男女之事视频高清在线观看| 香蕉国产在线看| 日韩一卡2卡3卡4卡2021年| 首页视频小说图片口味搜索| 高潮久久久久久久久久久不卡| 欧美精品亚洲一区二区| 极品少妇高潮喷水抽搐| 精品一品国产午夜福利视频| 日韩三级视频一区二区三区| 亚洲精品自拍成人| 精品一区二区三区四区五区乱码| 91麻豆av在线| 一区二区三区国产精品乱码| 成年版毛片免费区| 国产一区有黄有色的免费视频| 欧美激情极品国产一区二区三区| videosex国产| 桃红色精品国产亚洲av| svipshipincom国产片| 中文字幕高清在线视频| 精品人妻熟女毛片av久久网站| 视频区欧美日本亚洲| 亚洲熟女精品中文字幕| 黑人猛操日本美女一级片| 午夜日韩欧美国产| 在线观看免费视频日本深夜| 动漫黄色视频在线观看| 国产真人三级小视频在线观看| 中文字幕人妻丝袜一区二区| 亚洲精品中文字幕一二三四区| 久久久国产成人免费| 黑人猛操日本美女一级片| 欧美日韩精品网址| 妹子高潮喷水视频| 亚洲精品中文字幕在线视频| 亚洲精品在线观看二区| 久久中文看片网| 国产精品 国内视频| 午夜精品久久久久久毛片777| 法律面前人人平等表现在哪些方面| 欧美激情久久久久久爽电影 | 精品午夜福利视频在线观看一区| av天堂久久9| 成人精品一区二区免费| 久久久久国产精品人妻aⅴ院 | 天堂动漫精品| 日韩欧美一区视频在线观看| 一二三四社区在线视频社区8| 18在线观看网站| 国产区一区二久久| 精品卡一卡二卡四卡免费| 好男人电影高清在线观看| 亚洲av美国av| 欧美亚洲 丝袜 人妻 在线| 亚洲精品国产色婷婷电影| 一级黄色大片毛片| 久久中文看片网| 女性生殖器流出的白浆| 极品人妻少妇av视频| 老司机深夜福利视频在线观看| 男女午夜视频在线观看| 欧美日韩亚洲高清精品| 久久久久精品人妻al黑| 18禁国产床啪视频网站| 亚洲九九香蕉| 18禁裸乳无遮挡动漫免费视频| 亚洲av成人一区二区三| 国产精品久久久久久精品古装| 国产成人精品在线电影| 99精品久久久久人妻精品| 国产精品免费大片| 人人澡人人妻人| 国产精品99久久99久久久不卡| 久久久国产欧美日韩av| 国产有黄有色有爽视频| x7x7x7水蜜桃| 国产成人免费无遮挡视频| 精品人妻1区二区| 男女免费视频国产| 多毛熟女@视频| 国产精品一区二区精品视频观看| 国产无遮挡羞羞视频在线观看| 天堂动漫精品| 超色免费av| 高清av免费在线| 香蕉久久夜色| 波多野结衣av一区二区av| 一a级毛片在线观看| 色精品久久人妻99蜜桃| 久久久久国产一级毛片高清牌| 看片在线看免费视频| 九色亚洲精品在线播放| 日韩成人在线观看一区二区三区| 少妇裸体淫交视频免费看高清 | 少妇裸体淫交视频免费看高清 | 国产成+人综合+亚洲专区| av天堂在线播放| 亚洲一码二码三码区别大吗| 欧美黄色淫秽网站| 老熟妇乱子伦视频在线观看| 久久精品国产清高在天天线| 男人的好看免费观看在线视频 | 精品亚洲成a人片在线观看| 国产午夜精品久久久久久| 人人妻人人澡人人看| 啦啦啦视频在线资源免费观看| 男人舔女人的私密视频| 亚洲精品美女久久久久99蜜臀| 亚洲一卡2卡3卡4卡5卡精品中文| 欧美日韩亚洲国产一区二区在线观看 | 日韩欧美在线二视频 | www.熟女人妻精品国产| 咕卡用的链子| 免费日韩欧美在线观看| 日韩免费av在线播放| 亚洲性夜色夜夜综合| 女人被狂操c到高潮| 亚洲精品国产一区二区精华液| 在线国产一区二区在线| 自线自在国产av| 99国产综合亚洲精品| 两人在一起打扑克的视频| 国产精品一区二区在线不卡| 久久久久久免费高清国产稀缺| 日韩制服丝袜自拍偷拍| 99在线人妻在线中文字幕 | 岛国毛片在线播放| 日韩有码中文字幕| av电影中文网址| 三上悠亚av全集在线观看| 久久久水蜜桃国产精品网| 欧美色视频一区免费| 亚洲午夜理论影院| 亚洲欧美色中文字幕在线| 国产欧美日韩一区二区精品| 精品一区二区三区av网在线观看| 国产精品久久久久久精品古装| 女人久久www免费人成看片| 99热网站在线观看| 两性午夜刺激爽爽歪歪视频在线观看 | 91麻豆av在线| 国产精品亚洲av一区麻豆| 国产深夜福利视频在线观看| 别揉我奶头~嗯~啊~动态视频| 国产成人欧美| 老司机影院毛片| 看黄色毛片网站| videosex国产| 成人影院久久| av天堂在线播放| 电影成人av| 多毛熟女@视频| 成人特级黄色片久久久久久久| 成人免费观看视频高清| 国产精品偷伦视频观看了| 国产成人免费无遮挡视频| 精品一区二区三卡| 国产精品av久久久久免费| 日韩一卡2卡3卡4卡2021年| 一本综合久久免费| 日韩欧美一区二区三区在线观看 | 国产免费现黄频在线看| 久9热在线精品视频| 黑人操中国人逼视频| a级片在线免费高清观看视频| 香蕉国产在线看| 9191精品国产免费久久| 国产高清激情床上av| 色婷婷av一区二区三区视频| 麻豆成人av在线观看| 日韩欧美免费精品| 国产精品乱码一区二三区的特点 | 激情视频va一区二区三区| 国产精品二区激情视频| 美女高潮到喷水免费观看| 国产成人精品久久二区二区91| 久久久精品国产亚洲av高清涩受| 久久国产精品人妻蜜桃| 久久人人97超碰香蕉20202| av有码第一页|