• <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 ?

    Throughput scheduling in cognitive radio networksbased on immune optimization

    2015-03-01 09:21:56ChaiZhengyiZhengBaolinShenLianfengZhuSifeng

    Chai Zhengyi  Zheng Baolin  Shen Lianfeng  Zhu Sifeng

    (1National Mobile Communications Research Laboratory, Southeast University, Nanjing 210096, China)(2School of Computer Science and Software Engineering, Tianjin Polytechnic University, Tianjin 300384, China)(3Department of Information Engineering, Henan Vocational and Technical College, Zhengzhou 450046, China)

    ?

    Throughput scheduling in cognitive radio networksbased on immune optimization

    Chai Zhengyi1,2Zheng Baolin3Shen Lianfeng1Zhu Sifeng1

    (1National Mobile Communications Research Laboratory, Southeast University, Nanjing 210096, China)(2School of Computer Science and Software Engineering, Tianjin Polytechnic University, Tianjin 300384, China)(3Department of Information Engineering, Henan Vocational and Technical College, Zhengzhou 450046, China)

    Abstract:To study the throughput scheduling problem under interference temperature in cognitive radio networks, an immune algorithm-based suboptimal method was proposed based on its NP-hard feature. The problem is modeled as a constrained optimization problem to maximize the total throughput of the secondary users (SUs). The mapping between the throughput scheduling problems and the immune algorithm is given. Suitable immune operators are designed such as binary antibody encoding, antibody initialization based on pre-knowledge, a proportional clone to its affinity and an adaptive mutation operator associated with the evolutionary generation. The simulation results show that the proposed algorithm can obtain about 95% of the optimal throughput and operate with much lower liner computational complexity.

    Key words:cognitive radio networks; throughput scheduling; immune algorithm; interference temperature

    Received 2015-04-20.

    Biography:Chai Zhengyi (1976—), male, doctor, associate professor,super_chai@126.com.

    Foundation items:The National Natural Science Foundation of China (No.U1504613, 61202099, 61201175, U1204618), China Postdoctoral Science Foundation (No.2013M541586).

    Citation:Chai Zhengyi, Zheng Baolin, Shen Lianfeng, et al. Throughput scheduling in cognitive radio networks based on immune optimization[J].Journal of Southeast University (English Edition),2015,31(4):431-436.[doi:10.3969/j.issn.1003-7985.2015.04.001]

    The increasing growth in wireless communication demands has intensified the shortage crisis for the radio spectrum, while a significant amount of the licensed spectrum is not currently being utilized[1]. The spectrum is far more underutilized rather than naturally scarce. A cognitive radio network (CRN) is a kind of intelligent communication system, which enables the devices to opportunistically access the licensed spectrum, and thereby enhance the utilization of the existing spectrum resources[2-3]. The nodes in a cognitive radio network can be classified into primary users (PUs) and secondary users (SUs). A PU is a licensed user that has exclusive rights to the access spectrum. A SU is an unlicensed user that can utilize the spectrum opportunistically with primary users under interference restriction.

    Throughput optimal scheduling for cognitive radio networks under interference temperature constraints is an open research issue[4]. The throughput scheduling determines how many packets and with which frequency each SU will transmit in each time slot[5]. The aim of it is to maximize the total throughput of the SUs in the cell. The throughput scheduler issue in conventional networks has been widely studied[6]. Nonetheless, the cognitive radio paradigm brings new challenges into the issue because of the coexistence of the PUs and the SUs. The throughput scheduling problem considered in this paper can be distinguished from these works by its cognitive radio specific nature. That is, not only the availability of different frequencies but also the maximum allowable transmission rate of the frequency bands are time-varying[5].

    Researchers have done some work with different scenarios. In Ref.[6], a throughput scheduling algorithm was proposed which does not enable the true coexistence of the PUs and the SUs. The authors in Ref.[7] figureted a distributed heuristic to determine the channels and time slots for the cognitive nodes. However, they do not consider the interference for the PUs either in their optimization figuretion or in their suboptimal heuristic. The interference temperature model provides the true coexistence data of its licensed and unlicensed users. The throughput optimization is a binary integer programming problem, so the figureted scheduling algorithms have a high computational complexity[3-7]. In Ref.[3], its optimal solution was obtained by the branch-and-bound algorithm with very high computational complexity. In Ref.[4], the authors focused on throughput scheduling under interference temperature constraints and figureted the throughput maximization problems. Then, the authors proposed suboptimal schedulers, referred to as maximum frequency selection (MFS) and probabilistic frequency selection (PFS), with low complexity at the expense of poor throughput performance. Hence, the design of better performing suboptimal scheduling with reasonable complexity is very meaningful.

    It is known that bio-inspired methods are ideal for such nonlinear optimization problems[8]. Some bio-inspired methods have been employed in conventional (non-cognitive) schedulers, such as the genetic algorithm[9]and particle swarm optimization[10]. In this paper, an improved immune algorithm is introduced to solve the throughput scheduling problem. The inspiration comes from the fact that the immune algorithm is ideal for nonlinear optimization problems with a large feasible solution space where a quick sub-optimal solution will suffice. Also, to the best of our knowledge, the use of the immune algorithm for scheduling in cognitive radio networks has not previously been explored.

    1System Model

    Consider a time-slotted IEEE 802.22 system in which the SUs are controlled and guided by the cognitive base station (CBS)[4-5]. The scheduler is at the CBS. Assume that the interference temperature perceived by the PUs is within the interference temperature limits; reliable communication between the CBS and the SUs is achieved; and collisions among the SUs are avoided[4]. Each SUncalculates every frequencyf, and the value is denoted asUnf, which is the maximum number of packets that it can transmit for frequencyfin a time slot. The calculation procedure forUnfvalues guarantees that the interference temperature perceived by the PUs is within the predetermined limits. The CBS then constitutes a matrix called U=[Unf].

    The throughput optimal schedule can be formatted as

    (1)

    (1a)

    Xnft+Xn′ft≤1?n,n′∈1,2,…,N;n≠n′; ?f,?t

    (1b)

    whereN,F,Tare the total number of SUs, frequencies, time slots, respectively;Xnftis a binary variable such thatXnft=1 if userntransmits with frequencyfin time slottand 0 otherwise. Constraint (1a) guarantees that at least one time slot is assigned to each SU, whereas constraint (1b) makes certain that at most one user can transmit at a particular time slot and frequency combination, and consequently preventing collisions among the cognitive nodes. Moreover, the schedule lengthTis the time period in which the spectral and networking environment changes slowly enough so that theXnftvalues are not affected. For example, the TV bands used by an IEEE 802.22 network constitute a slowly altering spectral environment, and hence enableTto be large enough[4].

    2Proposed Algorithm

    2.1 Overview of immune optimization algorithm

    The artificial immune system (AIS) is inspired by the human immune system. The AIS-based algorithms typically extract ideas from the human immune system’s characteristics of learning and adaptability to solve some complicated problems[10]. Most immune system inspired optimization algorithms are based on the clone selection principle. Clone selection is a dynamic simulation process of the immune system that is self-adaptive against antigens. The clone selection algorithm for optimization has been widely used in engineering-oriented fields, such as spectrum allocation[11], job scheduling[12], and image segmentation[13]and so on. These algorithms essentially evolve solutions to problems via the repetition of a clone, affinity maturation (via mutation) and a selection cycle for a candidate solutions population, and remaining good solutions in the population[10-12].

    Some related terms are described briefly as follows:

    1) Antigen. An antigen represents one sample in the solution space of the problem. In this paper, antigen refers to the throughput schedule problem to be solved and the total constraints.

    2) Antibody. An antibody represents a candidate solution to the problem in this paper.

    3) Antibody population. The complete antibodies consist of antibody population.

    4) Affinity. Affinity is the fitness measurement for an antibody, which indicates the extent that the antigen satisfies the problem requirements.

    5) Clone. In immunology, cloning means asexual propagation so that a group of identical cells can be descended from a single common ancestor. It is used to enlarge the search region.

    6) Mutation. In immunology, mutation means the immune system recognizes external patterns from antibody gene mutation in order to obtain a higher affinity. Mutations take the search procedure out of a locally optimal region, and enable it to possibly enter a better region of the search space.

    7) Selection. An immune algorithm takes a group of antigens from a population using an operation called selection. The selection operation serves the purpose of eliminating the relatively bad solution candidates and focusing the search operation on a relatively good portion of the solution space.

    2.2 Realization of throughput scheduling based on the immune algorithm

    Our motivations for utilizing the immune algorithm for the throughput optimal scheduling problems are manifold. First, the immune algorithm is suitable for problems with large search spaces. It is equipped with many tools to reduce the computational complexity and produce a diverse set of solutions. The fact that the immune algorithm operates on a population of solutions rather than a single solution implies that the algorithm makes parallel searches in the search space. Considering that the solution space in the throughput scheduling problems is enormous (even for 5 nodes, 3 frequencies, and 3 time slots, the size of the solution space is 245), the immune algorithm appears to be a suitable tool. Secondly, the immune algorithm can be conveniently implemented. The binary decision variablesXnftcan be easily encoded to a binary string.

    Some key techniques are as follows:

    1) Antigen representation (encoding). We use the binary encoded antigen which containsXnftvalues. Thus, the antigen [X111,X211,X311,X112,X212,…,X123,X223,X323], whereas the other antigen structure is [X111,X112,X113,X121,X122,…,X322,X323].Xnftis a gene bit of antibody.

    2) Affinity evaluation. In this paper, the optimization model is described in Eq.(1). The affinity is a mapping of the value of Eq.(1) for a given antibody. SinceQis to be maximized, it can be stated that if an antibody has higher affinity, it is the better one.

    The proposed algorithm for throughput optimal scheduling schemes is implemented as follows:

    Step 1Initialization. Set the maximum iterative generationtmax. Sett=0, wheretis termed as the current iterative generation. Create an initial antibody population A(t) with sizekin accordance with antibody encoding in section 3.2. That is

    A(t)={p1(t),p2(t),…,pk(t)}

    (2)

    where pi(1≤i≤k) is a candidate throughput scheduling scheme; A(t) is a set of candidate throughput scheduling schemes.

    Here, some pre-knowledge is used to initialize the antibody piin order to accelerate algorithm convergence, which is proved by the latter simulation experiments. From constraint (1a), it is known that at least one time slot is assigned to each SU. Constraint (1b) makes certain that at most, one user can transmit at a particular time slot and frequency combination. Each antibody pi(1≤i≤k) that satisfies the constraints (1a) and (1b) will be a candidate.

    Step 2Affinity evaluation. The affinities of all antibodies in A(t) are calculated according to Eq.(1) and it is denoted as

    f(A(t))={f(p1(t),f(p2(t)),…,f(pk(t)}

    (3)

    If an antibody pi(t) (1

    Step 3Proportional cloneTc. In this paper, B(t) is obtained by applying clone proliferationTcto A(t), and it is defined as

    B(t)=Tc(A(t))={Tc(p1(t)),Tc(p2(t)),…,Tc(pk(t))}

    (4)

    Here, the clone scaleqifor each antibody pi(1≤i≤k) is proportional to its affinityf(pi(t)). That is

    (5)

    where Int() denotes the integer function, andncis a given value (nc>k). The antibody with a large affinity value (objective function value of Eq.(1)) has a largeqi.

    (6)

    Actually, clone proliferation on antibody pi(t) is to make multiple identical copies of it.

    Step 4MutationTm. In this paper, it is defined as C(t)=Tm(B(t)). An adaptive mutation which associates the mutation probabilitympwith the evolutionary generation is designed. That is

    (7)

    wheretis the current evolutionary generation;tmaxis the maximum evolutionary generation.

    The advantages of the mutation lie in its searching ability within a large scope in the early evolution process while it searches in a local scope in the latter evolution process, which can accelerate the convergence. After mutation, the population becomes

    C(t)={p″1(t),p″2(t),…,p″z(t)}

    (8)

    In this paper, the mutation is done by exchanging the element one and element zero with each other with probabilitymp. The proposed mutation is easily realized and it does not violate the constraints.

    Step 5Affinity evaluation. The affinities of all antibodies in C(t) are calculated according to Eq.(1) and it is defined as

    f(C(t))={f(p″1(t),f(p″2(t)),…,f(p″z(t)}

    (9)

    Step 6Clone selectionTsis defined as

    A(t+1)=Ts(C(t)∪A(t)=

    (p1(t+1),p2(t+1),…,pk(t+1))

    (10)

    That is,kantibodies with a high affinity are selected from C(t) and A(t) to form the next population A(t+1).

    Step 7Termination test. Iftmaxis reached, stop the algorithm. Output the antibody with the maximum affinity in A(t+1) as the result of the throughout scheme. Otherwise,t=t+1, and go to Step 3.

    2.3 Computational complexity

    Recall thatNdenotes the number of SU, andFdenotes the number of available frequencies. For the immune-based scheme, the total computational complexity is mainly composed of that for initialization, affinity evaluation, cloning, mutation, and selection. Given the population sizek, the clone scalenc(nc>k) and the maximum generationtmax, the procedure of population initialization, the affinity evaluation and proportional clone (Step 1 to Step 3) has the same computational complexity ofO(kFN) in each generation, while the procedure of mutation, affinity evaluation, selection (Step 4 to Step 6) has the computational complexity ofO(kncFN) in each generation. Hence, for each generation, the total computational complexity isO(3kFN+3kncFN). Sincenc>k, according to the properties of symbolicO[10,13-14], it can be denoted asO(ncFN). When the throughout scheduling is finished, it has the total computational complexity ofO(tmaxncFN).

    The computer simulations show thattmaximplicitly depends onFandN. The more complex the search space is, the larger the number of generations should be. Thus, for givenncandtmax, the gradual computational complexity of the proposed algorithm isO(NF) in accordance with the properties of symbolicO.

    A brief summary of the complexities of previous typical algorithms and our proposed algorithm is as follows. The complexity of algorithm in Ref.[3] isO(FN2), while the complexity of our proposed algorithm is the same as the algorithm in Ref.[4], which isO(FN).

    3Simulation Results and Discussion

    3.1 Experimental environments and parameter settings

    We simulated the suboptimal schedulers and acquired theUnfvalues in OPNET Modeler, and we solved the optimization problems in CPLEX[15]. Additive white Gaussian noise (AWGN) channels are considered. In all the simulations, each SU has three primary neighbors in its interference range. The simulation results are the average of 100 independent tests. The parameter settings are as the same as that in Refs.[4-5]. There are three frequencies with interference temperature thresholds of 1 000, 2 000, and 3 000 K.

    3.2 Sensitivity in relation to the immune algorithm parameters

    Four parameters are to be set at the initialization phase: the antibody population sizek, the clone population sizenc, the mutation probabilitymp, and the maximum number of generationstmax. The sequential experimental design method of employing a series of small experiments each with a specific objective is a common method in experimental design[16], because the experimenter can quickly learn crucial information from a small group of runs that can be used to plan the next experiment.kandncdirectly affect the computational complexity of the algorithm[10-13]. If the givenkandncare large enough, the diversity of the population can be enhanced and the prematurity can be avoided in some extent, but the computational complexity will also be very large.tmaxclearly depends onFandN. The more complex the search space is, the larger the number of generations should be.mpis very important for local search in the algorithm. A largemphas the ability to produce more new antibodies, but it also has the probability to destroy some good antibodies. Whenmpis too small, the convergence speed is not quick enough to find the best solution in appointed generations.

    Since the optimal choice is difficult to determine by theoretical analysis, it is important to analyze the performance affected by experiments in different cases. After trial and error, the parameters employed in the proposed immune algorithm are as follows: the number of generationstmaxis 100; the population sizekis 50; the clone scalencis 10; and the mutation probabilitympis 0.3.

    3.3 The performances of the proposed algorithm

    In order to evaluate the performances of the proposed algorithm, the effect of number of iterations (evolutionary generation) on throughout scheduling is studied, and the number of SUs is set to be 5. As it is evident from Fig.1, the throughout initially increases with the number of evolutionary generations and then gradually converges to the high value, close to an optimal point. It is also can be seen that the proposed method provides significant gain in throughout and fast convergence rate. The simulation results prove the effectiveness of the proposed immune operators. It also can be seen from Fig.2 that the evolutionary generation increases with the numbers of SUs. It is effective.

    The results of different numbers of SUs are shown in Tab.1. It can be seen that the proposed algorithm gives consistent good results. First of all, the proposed solution yields better results than the MFS and PFS schedulers proposed in Ref.[4], at the same time being very close to the throughput optimal scheduler performance in Ref.[3].

    Fig.1 Throughout vs. evolutionary generation

    Fig.2 Evolutionary generation vs. different numbers of SUs

    Tab.1 The results of relative algorithms for different numbers of SUs

    Fig.3 presents the optimal throughput[3], MFS, and PFS[4]scheduling schemes compared with the proposed scheduling scheme. The algorithm in Ref.[3] can be regarded as the upper-bound of the proposed heuristic algorithm.

    Fig.3 Comparison of different algorithms

    Fig.4 shows the performances comparisons of relative algorithms with the increase in the numbers of SUs. It can be seen that the proposed algorithm performs excellently. It is very close to the optimal performances in Ref.[3] and is better than the algorithm in Ref.[4].

    All in all, the proposed scheduling scheme achieves performances close to the optimal scheduling operating with much lower complexity. However, the iterations in the simulation results reveal that the proposed algorithm is computationally more costly than the MFS and the PFS schedulers in Ref.[4]. Nevertheless, when they are compared with the throughput performance, we can see that the proposed algorithm is approximately twice as good as the MFS and the PFS schedulers. Moreover, the proposed algorithm is computationally more efficient than the classical branch and bound algorithms that are used to solve binary integer programming problems. Therefore, ther proposed algorithm presents a very reasonable tradeoff between computational complexity and performance.

    Fig.4 Throughout vs. numbers of SUs

    4Conclusion

    The immune algorithm-based suboptimal scheduling for the throughput problem in cognitive radio networks is proposed. The simulation results show that the proposed algorithm is very close to optimal performance with a relatively lower complexity. Hence, it can be concluded that the proposed scheduling is more suitable for slowly varying spectral environments. Considering that IEEE 802.22 networks that operate on the TV broadcast bands that are slowly changing, it can be confidently concluded that the proposed algorithm can operate in realistic network settings, and provide useful solutions for the open research problem.

    References

    [1]Akyildiz I, Lee W, Vuran M, et al. Next generation/dynamic spectrum access/cognitive radio wireless networks[J].ASurveyComputerNetworks, 2006, 50(13): 2127-2159.

    [2]Busson A, Jabbari B, Babaei A, et al. Large overlaid cognitive radio networks: from throughput scaling to asymptotic multiplexing gain[J].JournalofCommunicationsandNetworks, 2014, 16(1): 67-80.

    [3]Tian Z, Leus G, Lottici V. Joint dynamic resource allocation and waveform adaptation for cognitive networks[J].IEEEJournalonSelectedAreasinCommunications, 2011, 29(2): 443-454.

    [4]G?züpek D, Alag?z F. Throughput and delay optimal scheduling in cognitive radio networks under interference temperature constraints[J].JournalofCommunicationsandNetworks, 2009, 11(2): 147-155.

    [5]Arshad K, Mackenzie R, Celentano U, et al. Resource management for QoS support in cognitive radio networks[J].IEEECommunicationsMagazine, 2014, 52(3): 114-120.

    [6]Chaporkar P, Kar K, Luo X, et al. Throughput and fairness guarantees through maximal scheduling in wireless networks[J].IEEETransactionsonInformationTheory, 2008, 54(2): 572-594.

    [7]Chai Zhengyi, Liu Fang. A novel immune optimization algorithm for fairness resource allocation in cognitive wireless network[J].WirelessPersonalCommunications, 2013, 69(4): 1671-1687.

    [8]He A, Kyung K B,Newman T R, et al. A survey of artificial intelligence for cognitive radios[J].IEEETransactionsonVehicularTechnology, 2010, 59(4): 1578-1592.

    [9]Zhao Jun, Liu Quanli, Wang Wei, et al. A parallel immune algorithm for traveling salesman problem and its application on cold rolling scheduling[J].InformationSciences, 2011, 181(7): 1212-1223.

    [10]Chai Zhengyi, Liu Fang. On the use of immune clonal optimization for joint subcarrier and power allocation in OFDMA with proportional fairness rate[J].InternationalJournalofCommunicationSystems, 2013, 26(10): 1273-1287.

    [11]Luh G-C, Chueh C-H. A multi-modal immune algorithm for the job-shop scheduling problem[J].InformationSciences, 2009, 179(10): 1516-1532.

    [12]Yang Dongdong, Jiao Licheng, Gong Maoguo, et al. Artificial immune multi-objective SAR image segmentation with fused complementary features[J].InformationSciences, 2011, 181(13): 2797-2812.

    [13]Tarcisio F, Maciel M, Anja K. On the performance, complexity, and fairness of suboptimal resource allocation for multi-user MIMO-OFDMA Systems[J].IEEETransactionsonVehicularTechnology, 2010, 59(1): 406-421.

    [14] Banaei A, Georghiades C N, Cui S G. Maximum-minimum throughput for MIMO systems in cognitive radio networks[J].IEEETransactionsonWirelessCommunications, 2014, 13(6): 3042-3055.

    [15]Sadr S, Anpalagan A, Raahemifar K. Radio resource allocation algorithms for the downlink of multiuser OFDM communication systems[J].IEEECommunicationsSurveys&Tutorials, 2009, 11(3): 92-106.

    doi:10.3969/j.issn.1003-7985.2015.04.001

    多毛熟女@视频| 黑人高潮一二区| xxx大片免费视频| 久久国产亚洲av麻豆专区| 国产黄色视频一区二区在线观看| 十八禁高潮呻吟视频| 亚洲精品美女久久av网站| 交换朋友夫妻互换小说| 精品久久国产蜜桃| 欧美日韩av久久| 乱码一卡2卡4卡精品| 伦理电影大哥的女人| 国产精品成人在线| 天堂8中文在线网| 狂野欧美激情性bbbbbb| 两个人免费观看高清视频| 免费观看在线日韩| 国产精品人妻久久久影院| videossex国产| 三上悠亚av全集在线观看| 亚洲无线观看免费| 久久99蜜桃精品久久| 久久久精品94久久精品| 免费黄色在线免费观看| 91精品一卡2卡3卡4卡| 国产精品 国内视频| 国产男人的电影天堂91| 老司机影院成人| tube8黄色片| 夫妻性生交免费视频一级片| 免费观看av网站的网址| 十八禁网站网址无遮挡| 精品人妻熟女av久视频| 热99国产精品久久久久久7| videossex国产| 三上悠亚av全集在线观看| 一本一本久久a久久精品综合妖精 国产伦在线观看视频一区 | 久久国内精品自在自线图片| 天天影视国产精品| 交换朋友夫妻互换小说| 国产高清国产精品国产三级| 亚洲成人一二三区av| 欧美激情 高清一区二区三区| 亚洲精品自拍成人| av天堂久久9| 亚洲精品第二区| 久久这里有精品视频免费| 天天操日日干夜夜撸| 日韩三级伦理在线观看| 国产老妇伦熟女老妇高清| 丝袜喷水一区| 母亲3免费完整高清在线观看 | av网站免费在线观看视频| 少妇的逼好多水| 在现免费观看毛片| 亚洲人成网站在线观看播放| 在线观看免费日韩欧美大片 | 免费av不卡在线播放| 草草在线视频免费看| 91精品国产国语对白视频| 国产精品蜜桃在线观看| 插逼视频在线观看| 国产亚洲欧美精品永久| 一级黄片播放器| 蜜桃在线观看..| 婷婷色麻豆天堂久久| 中文字幕最新亚洲高清| 特大巨黑吊av在线直播| 欧美日韩一区二区视频在线观看视频在线| 免费久久久久久久精品成人欧美视频 | 青春草亚洲视频在线观看| 少妇人妻精品综合一区二区| 高清视频免费观看一区二区| 亚洲美女视频黄频| 亚洲国产最新在线播放| 中文字幕人妻熟人妻熟丝袜美| 美女脱内裤让男人舔精品视频| 久久久久久伊人网av| 一区二区三区乱码不卡18| 国产亚洲精品第一综合不卡 | 啦啦啦在线观看免费高清www| 亚洲四区av| 成人综合一区亚洲| 国产精品偷伦视频观看了| 人人澡人人妻人| 成人二区视频| 人体艺术视频欧美日本| 免费av中文字幕在线| 亚洲欧美日韩另类电影网站| 伦精品一区二区三区| 哪个播放器可以免费观看大片| 久久99热这里只频精品6学生| 91精品国产九色| 国产精品.久久久| 国产伦精品一区二区三区视频9| av电影中文网址| 欧美精品高潮呻吟av久久| 国产免费福利视频在线观看| av一本久久久久| 国产视频内射| 免费av不卡在线播放| 成人国产av品久久久| 大话2 男鬼变身卡| 欧美性感艳星| 精品亚洲成国产av| av在线观看视频网站免费| 国产精品女同一区二区软件| 国产爽快片一区二区三区| 久久女婷五月综合色啪小说| 多毛熟女@视频| 国产精品一二三区在线看| 亚洲久久久国产精品| 日韩伦理黄色片| 好男人视频免费观看在线| 一个人免费看片子| 纵有疾风起免费观看全集完整版| 久久综合国产亚洲精品| 十八禁高潮呻吟视频| 大话2 男鬼变身卡| 亚洲国产精品一区三区| 精品人妻一区二区三区麻豆| 亚洲精品亚洲一区二区| 精品少妇内射三级| 夜夜爽夜夜爽视频| 成人黄色视频免费在线看| 亚洲成人手机| 久久这里有精品视频免费| 亚洲少妇的诱惑av| 男的添女的下面高潮视频| 久久久久人妻精品一区果冻| 伊人久久国产一区二区| 国产成人精品久久久久久| a级毛片免费高清观看在线播放| 丰满饥渴人妻一区二区三| 亚洲国产日韩一区二区| 国产亚洲欧美精品永久| 国产一级毛片在线| 男女免费视频国产| 日韩强制内射视频| 十八禁高潮呻吟视频| 国产免费一级a男人的天堂| 狂野欧美激情性xxxx在线观看| 亚洲精品中文字幕在线视频| 欧美xxⅹ黑人| 久久影院123| 18禁动态无遮挡网站| 九九在线视频观看精品| 一级毛片电影观看| 午夜精品国产一区二区电影| 两个人的视频大全免费| 91精品国产国语对白视频| 极品少妇高潮喷水抽搐| 欧美日韩一区二区视频在线观看视频在线| 26uuu在线亚洲综合色| 卡戴珊不雅视频在线播放| 国产亚洲一区二区精品| 午夜免费男女啪啪视频观看| 国产精品秋霞免费鲁丝片| 热99国产精品久久久久久7| 蜜桃在线观看..| 少妇丰满av| 中文字幕精品免费在线观看视频 | 日产精品乱码卡一卡2卡三| 一二三四中文在线观看免费高清| 赤兔流量卡办理| 欧美日韩精品成人综合77777| 午夜老司机福利剧场| 日韩一区二区视频免费看| 街头女战士在线观看网站| 国产免费一区二区三区四区乱码| 午夜免费男女啪啪视频观看| 成人国产av品久久久| 国产精品无大码| 午夜91福利影院| 久久这里有精品视频免费| 一区在线观看完整版| 最后的刺客免费高清国语| 久热这里只有精品99| 3wmmmm亚洲av在线观看| 国产精品久久久久久精品电影小说| 亚洲,欧美,日韩| 亚洲国产精品专区欧美| 久久精品夜色国产| 国产av精品麻豆| 男女高潮啪啪啪动态图| 国产日韩欧美亚洲二区| 国产一区二区三区av在线| 一区二区三区乱码不卡18| 免费看不卡的av| 国产精品秋霞免费鲁丝片| 曰老女人黄片| 亚洲中文av在线| 九色成人免费人妻av| 亚洲国产欧美日韩在线播放| 久久热精品热| 婷婷色综合www| 亚洲精品一二三| 国产一区二区在线观看av| 下体分泌物呈黄色| 欧美日韩综合久久久久久| 多毛熟女@视频| 99热这里只有精品一区| 日产精品乱码卡一卡2卡三| 好男人视频免费观看在线| 少妇被粗大的猛进出69影院 | 国产国拍精品亚洲av在线观看| 嘟嘟电影网在线观看| 狠狠婷婷综合久久久久久88av| 国产亚洲最大av| 久久精品国产a三级三级三级| 日韩人妻高清精品专区| 99久久综合免费| 久久毛片免费看一区二区三区| 精品国产一区二区久久| 国产精品久久久久久精品古装| av.在线天堂| 观看美女的网站| tube8黄色片| 欧美日韩在线观看h| 国产在视频线精品| 人妻一区二区av| 国产精品国产三级国产av玫瑰| 99热这里只有是精品在线观看| 九九在线视频观看精品| 久久99热这里只频精品6学生| 精品一区二区三卡| 亚洲av男天堂| 婷婷成人精品国产| 午夜福利网站1000一区二区三区| 91久久精品国产一区二区成人| 午夜免费观看性视频| 成人国产麻豆网| 在线免费观看不下载黄p国产| 欧美日韩综合久久久久久| 美女xxoo啪啪120秒动态图| 国产日韩欧美亚洲二区| 日韩伦理黄色片| 国产成人aa在线观看| 亚洲成色77777| 在线观看国产h片| 少妇精品久久久久久久| 亚洲欧洲国产日韩| 成人二区视频| 只有这里有精品99| 最近的中文字幕免费完整| 精品国产露脸久久av麻豆| 国产成人av激情在线播放 | 色婷婷av一区二区三区视频| 亚洲精品456在线播放app| 国产免费一级a男人的天堂| 热re99久久国产66热| 国产精品不卡视频一区二区| 国产精品三级大全| 亚洲精品国产av蜜桃| 天天躁夜夜躁狠狠久久av| 精品国产国语对白av| 亚洲,欧美,日韩| 九色成人免费人妻av| 亚洲av在线观看美女高潮| 免费人妻精品一区二区三区视频| 狂野欧美激情性bbbbbb| 亚洲精品aⅴ在线观看| 国产精品久久久久久av不卡| 91精品三级在线观看| 男女边吃奶边做爰视频| 男人添女人高潮全过程视频| 99久久中文字幕三级久久日本| 丝袜在线中文字幕| 大香蕉久久成人网| 亚洲五月色婷婷综合| 在线天堂最新版资源| 亚洲色图综合在线观看| 亚洲在久久综合| 免费av中文字幕在线| 国产高清三级在线| 99久久中文字幕三级久久日本| 国产精品蜜桃在线观看| 中文欧美无线码| 熟女人妻精品中文字幕| 我的老师免费观看完整版| 亚洲高清免费不卡视频| 精品酒店卫生间| 日韩av在线免费看完整版不卡| 最近手机中文字幕大全| 欧美激情极品国产一区二区三区 | 亚洲成人一二三区av| 久久久久久久久久久久大奶| 国产片特级美女逼逼视频| 波野结衣二区三区在线| 少妇猛男粗大的猛烈进出视频| 亚洲精品一二三| 日韩av免费高清视频| 91成人精品电影| 国产欧美日韩综合在线一区二区| 嘟嘟电影网在线观看| 亚洲精品视频女| 99久国产av精品国产电影| 亚洲av二区三区四区| 观看美女的网站| 97超碰精品成人国产| 国产国拍精品亚洲av在线观看| 精品亚洲成国产av| 七月丁香在线播放| 亚洲av在线观看美女高潮| 观看av在线不卡| 韩国av在线不卡| 91久久精品电影网| 欧美日韩av久久| 免费观看av网站的网址| 免费观看的影片在线观看| 久久精品人人爽人人爽视色| 国产精品久久久久久av不卡| 国产免费现黄频在线看| 久久99热这里只频精品6学生| 亚洲精品一二三| 多毛熟女@视频| 精品一区二区三卡| 少妇的逼好多水| 久久久久久久亚洲中文字幕| 国产精品国产三级国产av玫瑰| 黑人巨大精品欧美一区二区蜜桃 | 丝袜美足系列| 免费高清在线观看日韩| 一级毛片我不卡| 日韩熟女老妇一区二区性免费视频| 黄片播放在线免费| 精品人妻熟女毛片av久久网站| 91久久精品国产一区二区三区| 男人添女人高潮全过程视频| 高清av免费在线| 天天影视国产精品| 欧美精品国产亚洲| 一级毛片aaaaaa免费看小| 黑人高潮一二区| 丁香六月天网| 亚洲av男天堂| tube8黄色片| 一级毛片 在线播放| 国产在线一区二区三区精| 又大又黄又爽视频免费| 老熟女久久久| www.色视频.com| 老熟女久久久| 精品一区二区免费观看| 在线观看人妻少妇| 亚洲av福利一区| 午夜影院在线不卡| 亚洲内射少妇av| 久久狼人影院| √禁漫天堂资源中文www| 青春草视频在线免费观看| 亚州av有码| 精品亚洲成国产av| av又黄又爽大尺度在线免费看| 涩涩av久久男人的天堂| 久久久久视频综合| 亚洲三级黄色毛片| 国产精品三级大全| 日韩在线高清观看一区二区三区| 免费大片18禁| 日本黄色日本黄色录像| 久久韩国三级中文字幕| 免费高清在线观看日韩| 婷婷色综合www| 欧美日韩一区二区视频在线观看视频在线| 亚洲av免费高清在线观看| 欧美日韩一区二区视频在线观看视频在线| 99久国产av精品国产电影| 成年美女黄网站色视频大全免费 | 久久99精品国语久久久| 欧美亚洲日本最大视频资源| 最新中文字幕久久久久| av在线app专区| 国产成人免费观看mmmm| 丝袜喷水一区| 亚洲内射少妇av| 亚洲丝袜综合中文字幕| 国产片内射在线| 国产精品国产av在线观看| 欧美国产精品一级二级三级| 国产不卡av网站在线观看| 青春草亚洲视频在线观看| 久久av网站| 久久综合国产亚洲精品| 国产成人精品福利久久| 好男人视频免费观看在线| 色网站视频免费| 夜夜骑夜夜射夜夜干| 高清午夜精品一区二区三区| 精品一区二区三区视频在线| 国产精品 国内视频| 少妇人妻精品综合一区二区| 欧美精品一区二区免费开放| 午夜精品国产一区二区电影| 久久久久久伊人网av| 在线观看人妻少妇| 国产男人的电影天堂91| 免费看不卡的av| 18禁在线播放成人免费| 国产精品一区二区三区四区免费观看| 午夜91福利影院| 你懂的网址亚洲精品在线观看| 99久国产av精品国产电影| 亚洲精品aⅴ在线观看| 九色亚洲精品在线播放| 黄片无遮挡物在线观看| 中文字幕制服av| 九九久久精品国产亚洲av麻豆| 精品亚洲成国产av| 免费久久久久久久精品成人欧美视频 | av在线app专区| 亚洲精品乱久久久久久| 日韩欧美一区视频在线观看| 免费大片18禁| 人人妻人人澡人人爽人人夜夜| 精品国产国语对白av| 你懂的网址亚洲精品在线观看| 老司机影院成人| 日韩一本色道免费dvd| 国产精品一二三区在线看| 国产熟女午夜一区二区三区 | 久久久国产一区二区| 赤兔流量卡办理| 国产综合精华液| 黄色配什么色好看| 一级毛片我不卡| 久久99一区二区三区| 水蜜桃什么品种好| av播播在线观看一区| 亚洲国产成人一精品久久久| 亚洲av.av天堂| 成人综合一区亚洲| 亚洲国产精品一区三区| 日韩熟女老妇一区二区性免费视频| 能在线免费看毛片的网站| 日本黄大片高清| 最近2019中文字幕mv第一页| 制服诱惑二区| 国产乱人偷精品视频| 亚洲精品色激情综合| 久久久精品免费免费高清| 国产精品女同一区二区软件| 亚洲经典国产精华液单| av电影中文网址| 人妻夜夜爽99麻豆av| 一级黄片播放器| 哪个播放器可以免费观看大片| 老熟女久久久| 亚洲内射少妇av| 日日摸夜夜添夜夜添av毛片| 男男h啪啪无遮挡| 免费观看av网站的网址| 久久人人爽人人片av| 美女主播在线视频| 少妇精品久久久久久久| 久久ye,这里只有精品| 日韩,欧美,国产一区二区三区| 国产精品久久久久久av不卡| 国产免费一级a男人的天堂| 国产乱来视频区| 亚洲色图 男人天堂 中文字幕 | 午夜久久久在线观看| 欧美变态另类bdsm刘玥| 精品少妇久久久久久888优播| 亚洲人成77777在线视频| 亚洲精品日本国产第一区| 欧美亚洲 丝袜 人妻 在线| 国产欧美日韩一区二区三区在线 | av专区在线播放| 综合色丁香网| 高清午夜精品一区二区三区| 亚洲精品成人av观看孕妇| 免费观看a级毛片全部| 99久久人妻综合| 少妇的逼好多水| 国产 一区精品| 久久精品国产亚洲av天美| 五月玫瑰六月丁香| 尾随美女入室| 性色avwww在线观看| 22中文网久久字幕| 女的被弄到高潮叫床怎么办| 国产精品人妻久久久影院| 日韩成人伦理影院| 18禁在线无遮挡免费观看视频| 两个人的视频大全免费| 国产精品久久久久久久久免| 亚洲精品色激情综合| 91精品一卡2卡3卡4卡| av卡一久久| 人成视频在线观看免费观看| 亚洲国产欧美日韩在线播放| 国产欧美另类精品又又久久亚洲欧美| 熟女av电影| av卡一久久| 交换朋友夫妻互换小说| 特大巨黑吊av在线直播| 欧美精品国产亚洲| 国产成人freesex在线| 插阴视频在线观看视频| 18在线观看网站| 国产深夜福利视频在线观看| √禁漫天堂资源中文www| 欧美日本中文国产一区发布| 国产精品嫩草影院av在线观看| 国产精品成人在线| 日本91视频免费播放| 一级毛片电影观看| 国产成人精品在线电影| 久久精品夜色国产| 亚洲国产av影院在线观看| h视频一区二区三区| 91在线精品国自产拍蜜月| 欧美xxⅹ黑人| 丝袜在线中文字幕| 欧美精品国产亚洲| 久久久久国产网址| 国产片特级美女逼逼视频| av免费观看日本| 夜夜爽夜夜爽视频| 国产av国产精品国产| 我的老师免费观看完整版| 美女视频免费永久观看网站| 国产男人的电影天堂91| 亚洲精品一二三| 成人无遮挡网站| 男的添女的下面高潮视频| 午夜精品国产一区二区电影| 精品久久久久久久久av| 成人免费观看视频高清| 欧美xxxx性猛交bbbb| 搡女人真爽免费视频火全软件| 在线观看免费视频网站a站| av在线播放精品| 久久99一区二区三区| 欧美 日韩 精品 国产| 午夜激情久久久久久久| 大香蕉久久网| 丝袜脚勾引网站| 又大又黄又爽视频免费| 考比视频在线观看| 99九九在线精品视频| 蜜臀久久99精品久久宅男| 97在线人人人人妻| 中文字幕最新亚洲高清| 99视频精品全部免费 在线| 国产视频内射| 黑人高潮一二区| 十八禁网站网址无遮挡| 国产片特级美女逼逼视频| 日韩伦理黄色片| 久久这里有精品视频免费| 国产成人a∨麻豆精品| 亚洲精品自拍成人| videos熟女内射| 国产日韩一区二区三区精品不卡 | 日日摸夜夜添夜夜爱| 22中文网久久字幕| 精品亚洲乱码少妇综合久久| 久久久久人妻精品一区果冻| 夜夜看夜夜爽夜夜摸| 韩国高清视频一区二区三区| 精品午夜福利在线看| 日韩,欧美,国产一区二区三区| 国产精品久久久久久精品电影小说| 最新中文字幕久久久久| 国产又色又爽无遮挡免| 最后的刺客免费高清国语| 91aial.com中文字幕在线观看| 国产色婷婷99| 99热这里只有精品一区| 另类精品久久| 国产日韩欧美亚洲二区| 最新中文字幕久久久久| 99热这里只有精品一区| 国产成人91sexporn| 亚洲精品日本国产第一区| 亚洲av中文av极速乱| 久久99热6这里只有精品| 国产 一区精品| 老女人水多毛片| 亚洲国产欧美在线一区| 一级毛片aaaaaa免费看小| 午夜免费鲁丝| 在线天堂最新版资源| 国产视频内射| 日韩大片免费观看网站| 国产视频首页在线观看| 国产白丝娇喘喷水9色精品| 久久久久网色| 久久久久精品久久久久真实原创| 久久狼人影院| 一区二区三区四区激情视频| 成人毛片a级毛片在线播放| 国产无遮挡羞羞视频在线观看| 亚洲av男天堂| 久久久久久久久久成人| 免费观看的影片在线观看| 亚洲熟女精品中文字幕| 中文字幕久久专区| 高清不卡的av网站| 男女啪啪激烈高潮av片| 精品人妻一区二区三区麻豆| 久久综合国产亚洲精品| 在线看a的网站| 国产欧美日韩一区二区三区在线 | 免费高清在线观看视频在线观看| 国产在线一区二区三区精| 中国国产av一级| 99国产综合亚洲精品| 啦啦啦啦在线视频资源| 一本色道久久久久久精品综合| 久久精品国产亚洲av涩爱| 国产片特级美女逼逼视频| 日本wwww免费看|