• <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 ?

    Clustered trajectories anonymity in wireless sensor networks①

    2015-04-17 05:33:38SebleHailuDadyWangJiahao
    High Technology Letters 2015年2期
    關(guān)鍵詞:充分考慮南通原材料

    Seble Hailu Dady, Wang Jiahao

    (School of Computer Science & Engineering, University of Electronic Science and Technology of China, Chengdu 611731, P.R.China)

    ?

    Clustered trajectories anonymity in wireless sensor networks①

    Seble Hailu Dady, Wang Jiahao②

    (School of Computer Science & Engineering, University of Electronic Science and Technology of China, Chengdu 611731, P.R.China)

    This paper proposes a clustered trajectories anonymity scheme (CTA) that enhances the k-anonymity scheme to provide the intended level of source location privacy in mobile event monitoring when a global attacker is assumed. CTA applies isomorphic property of rotation to create traces of the fake sources distributions which are similar to those of the real sources. Thus anonymity of each trajectory and that of the clustered is achieved. In addition, location k-diversity is achieved by distributing fake sources around the base station. To reduce the time delay, tree rooted at the base station is constructed to overlap part of the beacon interval of the nodes in the hierarchy. Both the analytical analysis and the simulation results prove that our scheme provides perfect anonymity with improved energy overhead and time delay.

    clustered trajectories anonymity scheme (CTA), source location privacy, k-anonymity, global attackers, wireless sensor networks (WSNs)

    0 Introduction

    Recently wireless sensor networks (WSNs) are extending their applications into many areas such as health care, environmental monitoring and wild life monitoring. In these application areas source location privacy providing is critical. This is because the exposure of the location of the source node leads to the location of the sensitive object under supervision. But source location privacy providing is a challenging task especially when the attacker is a global attacker and the monitored objects are mobile. One reason is that the spatio-temporal data associated with the mobile object is rich in correlation information.

    Solutions proposed to provide source location privacy against global attacker[1,2]introduce excessive energy overhead due to network wide dummy message injection. To trade off privacy with communication cost and latency, Mehta, et al.[1]proposed source simulation. In source simulation few sensor nodes are randomly selected to simulate the mobility of the monitored object. But building and implementing the object’s mobility profile is still an open research topic[1]. Scheme proposed in Ref.[3] switch statistically strong source anonymity (SSSA) scheme in Ref.[2] to a k-anonymity scheme in Ref.[1] on demand, but mobile event monitoring which can create real trajectory is not considered. Trajectory anonymity is addressed in a distributed manner in Ref.[4] but the solution applies only for events that start at the perimeter of the network and end somewhere inside the network.

    Spatio-temporal data expose not only the movement pattern of each monitored object but also the relationship between different monitored objects and their interaction with the environment. One aspect of this is hotspot. As discussed in Ref.[5] hotspot phenomenon causes an obvious inconsistency in the network traffic pattern. This is because a large volume of packets are originating from a small area. Thus, analysis of traffic may also lead to a cluster of trajectories that expose the mobility patterns of a collection of monitored objects.

    This paper considers WSNs deployed for wild life monitoring that demand privacy but tolerate time delay in the order of few seconds. Based on this application, but not limited to it, a clustered trajectories anonymity scheme (CTA) is proposed. This scheme provides anonymity to all possible correlation information that can be deduced from the spatio-temporal data associated with event reporting such as trajectory made by monitored object or cluster of trajectories made by the collection of monitored objects. In addition, location k-diversity is achieved by selecting fake sources distributed around the base station (BS). Our scheme also satisfies that fake sources should be comparable to the real source with respect to their distances to the BS[6]. To achieve all these there is no need to know the mobility pattern and movement speed of the monitored objects as in Refs[1,4]. Instead our scheme applies isomorphic property of rotation at BS to select fake sources. To reduce energy cost dummy messages are on demand and the beacon frames at MAC layer are used to send event notification message to the BS. Time delay during event notification is reduced by constructing tree rooted at BS to overlap part of the beacon intervals of nodes in the hierarchy. The analytical analysis and the simulation results show that our scheme provides perfect source location anonymity with improved energy cost and time delay.

    1 Related work

    Several protocols to provide source location privacy have been proposed[1-7], but a solution that guarantees efficient privacy against global attackers remains hard to pin down. The threat of global adversaries is first considered by Mehta et al.[1]. The authors proposed the periodic collection and source simulation schemes. Periodic collection scheme provides maximum location privacy but the network wide dummy messages consume significant amount of sensor energy while increasing the network collisions and decreasing the packet delivery ratio. Shao, et al.[7]proposed a scheme which conveyed data to the base station only using beacon frames at the MAC layer. Since beacons are periodically broadcasted regardless of the occurrence of real events, this approach provides perfect event source unobservability at no additional energy cost. However, since the time between consecutive beacons is relatively large, the solution is impractical for nodes that have successive messages to send.

    Mehta, et al.[1]proposed a second technique called source simulation which aimed at trading off energy consumption with privacy. In this scheme the real event is hidden among k-1 fixed fake events simulating the mobility patterns of the monitored object. But randomly selecting the nodes that simulate the movement patterns of the monitored object and the constant number of fake events may leak correlation information when multiple objects are monitored. Also in applications that monitor heterogeneous objects there is a need to implement multiple models to the sensor nodes which increases complication and overhead. The solution proposed in Ref.[3] randomly selects fake sources when it switches from SSSA to k-anonymity scheme, but trajectory anonymity is not considered. The advanced form of k-anonymity scheme called l-diversity is proposed in Ref.[3]. L-diversity is used in this scheme to diversify location of fake sources in order to decrease the attacker information gain when he tries to search the object in the suspected areas. Unobservable handoff trajectory protocol is proposed in Ref.[4] to hide trajectory of events that may change frequently in a distributed environment. Even if this scheme provides solution to handoff problem for the events with the assumed movement speed, it does not provide privacy for events that start inside the network such as wild life monitoring.

    2 Problem formulation

    2.1 Network model

    Beacon-enabled IEEE 802.15.4 is proposed used in wireless sensor network platforms such as ZigBee[8]to announce node presence and exchange system parameters. Beacon interval ranges from 15.36 milliseconds to 786.432 seconds as defined in IEEE 802.15.4. The considered WSN consists of the BS and a large number of homogenous sensor nodes which are randomly deployed in an area of interest as illustrated in Fig.1. The sensor nodes are resource constrained devices with low battery power and computation capacity but equipped with sensing, data processing, and communicating components. A BS connecting to the outside infrastructure such as the Internet collects data from the network. We assume it is located at the centre of the deployment area and is resourceful. Each sensor node is assumed to know its location with Global Positioning System (GPS) or other mechanisms mentioned in Ref.[9] and share this information with the neighbouring nodes and the BS. Message between sensor node and the BS is communicated by using encryption techniques. Although key management and the packet content confidentiality are beyond the scope of this paper, we recommend[10]it for further reading.

    Fig.1 The architecture of the considered network

    2.2 Attacker model

    A global attacker that is assumed in this work can eavesdrop every packet transmission in the network, examine the encrypted packets, analyze network-wide traffic, determine traffic density on every link in the wireless sensor network and search the place where he suspects the object. However, the attacker cannot inject, modify or interrupt packet transmissions and can neither decrypt the content of captured packets.

    2.3 Problem definition

    Conventional k-anonymity schemes randomly select fake sources to provide source location privacy against global attackers. But these schemes have limitations to provide anonymity to trajectories and other correlation information from traffic analysis of spatio-temporal data collected when mobile objects are monitored. Traffic information collected from a WSN for a discrete time interval {t1, t2, …, tn} includes tuples (x1, y1, t1), (x2, y2, t2), …, (xn, yn, tn) where (xi, yi) is the location of a node that reports event occurrence and tiis the event reporting time. Analysis of these might provide the trajectory made by the monitored object defined as:

    Ti={(x1, y1, t1), (x2, y2, t2),…, (xk, yk, tk)}, where (t1

    Definition 1 (Trajectory Similarity): Two trajectories Tiand Tjare said to be similar if:

    i. for any tuple (xk, yk, tk)∈Ti, ?(xh, yh, tk)∈Tjand

    ii. for any (xk, yk, tk), (xk+1, yk+1, tk+1)∈Tiand (xh, yh, tk), (xh+1, yh+1, tk+1)∈Tj, it holds that ‖(xk, yk), (xk+1, yk+1)‖≈

    ‖(xh, yh), (xh+1, yh+1)‖

    This implies that two trajectories are similar if they are defined in the same time interval and distances travelled by the two trajectories are almost the same at each time interval.

    For a WSN applied to monitor a collection of objects, trajectory anonymity of each object to be monitored may not be enough. Because collection of trajectories can also provide correlation information such as hot spot created by group movement of the monitored objects or nature of objects to visit the same place for some purposes. Thus, distribution of the monitored objects in the monitored area needs also anonymity. This distribution is defined as cluster of trajectories.

    Definition 2 (Clustered Trajectories Similarity): A collection of trajectories TC={T0, T1,…,Tn} and TF={Tf0, Tf1,…, Tfn} are similar if

    i. ?Ti∈TC, ?Tfk∈TFsuch that Tiand Tfksatisfy trajectory similarity and

    ii. ?Th, Ti∈TC, ?TfkTfl∈TFsuch that Dist (Th, Ti)≈ Dist (Tfk, Tfl) at all corresponding points.

    Hence clustered trajectories similarity implies trajectory similarity and collection of trajectories location relationship anonymity in terms of distance between trajectories. If the real trajectories are collected in one place, the same thing will happen for fake trajectories.

    Definition 3 (Clustered Trajectories k-anonymity): A cluster of real object trajectories Tcis k-anonymous if there are k-l clusters of fake trajectories TF1, TF2,…,TF(k-1)that are similar to Tc.

    This implies clustered trajectories k-anonymity also satisfies trajectory k-anonymity. Because ?Ti∈Tc,?Tfj∈TFjwhere j=1,2,…,k-1 such that Tiis similar to Tfj.

    For an attacker that performs an onsite examination distributing the suspected nodes affects his cost of finding the source node[3]. Based on this l-diversity[3]is adopted to WSNs in order to improve location diversity of the fake sources. The idea is that suppose the network area is divided into L(L>0) partitions of almost the same size. For the total number N of the sensor nodes in the WSN, assume S denotes a set of nodes where S?N and P(S) denotes the total number of different partitions that nodes in S are from. If P(S)≥l for 0

    Definition 4 (Spatial k-diversity): A set of nodes S?N has a property of spatial k-diversity if P(S)=L=k.

    The other problem in the k-anonymity scheme is that dummy traffics are on demand for reducing energy cost. This in turn needs a message to initiate communication. In the presence of global attackers, initiating communication without scarifying anonymity, energy overhead and time delay is a challenging issue. Previous researches have only partially addressed all problems mentioned above. This has motivated us to design clustered trajectories anonymity scheme (CTA).

    3 Clustered trajectories anonymity scheme (CTA)

    In this section three phases of CTA scheme are described bootstrapping (beacon interval) phase, fake source location calculation phase, and fake source assignment and message transfer phase.

    3.1 Beacon interval

    To trade off privacy with energy efficiency, the best method is reducing the number of fake sources and initiating fake message only when there is an event to report. To make fake message generation on demand we use the beacon frame at MAC layer in beacon enabled WSN. In CTA the source node inserts event notification message that contains event id, timestamp, event location and number of messages to be sent in the beacon payload.

    In beacon enabled WSNs every node in the network has independent beacon interval and sends the beacon message at the end of this interval. This introduces cumulative delay in sending event notification to the BS. To reduce time delay we use hierarchical routing which is common in WSNs[11]. A tree rooted at the BS is constructed as shown in Fig.2. The purpose of this tree is to overlap the delay of the beacon messages for nodes in the hierarchy.

    Fig.2 Hierarchical transfer of beacon messages

    A leaf node sends beacon messages to the parent node at the end of the beacon interval. The parent node initiates a beacon message when it accepts beacon message from the child node. The parent node decrypts and processes the accepted beacon message and sends it to its parent node without additional delay. This will continue until the beacon message reaches to the BS. Thus the beacon interval for each node starts from the time it sends a beacon message until the next beacon message arrives to it which also includes part of the delay at the leaf node. To do this the BS needs only to identify the leaf nodes and assign interval on these nodes. These nodes initiate beacon messages based on their interval.

    As illustrated in Fig.3, the beacon interval starts at t0. Leaf node L sends its beacon massage at t1which is the end of the first beacon interval. Parent P1sends the beacon message only by taking time to process beacon from L, parent P2do the same after accepting the beacon from P1and finally P2sends the beacon to the BS. The process will continue with this pattern for all time intervals t1, t2…. Thus the average time delay for event notification message that is hidden in the beacon frame to reach the BS is td/2 where tdis the beacon interval. The BS calculates fake sources locations and initiates message communication at the end of each beacon interval.

    Fig.3 Beacon interval overlaps

    3.2 Fake source location calculation

    Traffic collected from WSN deployed to monitor mobile objects exposes correlation information such as the trajectory made by the monitored object and how monitored objects are distributed in the network area which may in turn lead to the current source node location. To provide real event report distribution we use isomorphism which is studied in mathematics in order to extend insights from one phenomenon to others. If two objects are isomorphic, then any property which is preserved by an isomorphism and which is true of one of the objects is also true of the other.

    The distance between the nodes that report event occurrences are the main sources of correlation information for an adversary. One example of isometric functions that preserve distance is rotation. The point (x, y) is rotated by angle β to find (x′, y′) using the following formula:

    x′=xcosβ-ysinβ, y′=xsinβ+ycosβ

    Generally rotation about a fixed point preserves distance and angle. This advantage is taken to create k-1 sets of fake reports; each set will have distances between fake sources similar to the distances between sources that report real event occurrences. Every time when the source sends event notification to BS, BS calculates the fake sources locations by rotating the source node location by angles β, 2β,…, (k-1)β. In our case β is calculated as β=360/k to diversify the k nodes around the BS. Algorithm 1 describes the details of how the BS calculates fake source locations from the real source location.

    Algorithm 1: Fake source location calculation.

    From Fig.4 for k=4 fake sources F1, F2 and F3 are located by rotating real source location R by an angle β=360/4=900, 1800, and 2700. When the new source location is R’ the new fake sources are located by the same calculation but by the coordinate of R’. Thus the distance RR’ is maintained by the fake sources distances F1F1’, F2F2’ and F3F3’.

    Fig.4 Locating fake sources around the BS

    3.3 Fake sources assignment and message transfer

    Calculating location of fake sources based on Algorithm 1 provides ideal location. That is, there may not be a sensor node at the calculated location. In this case BS selects nodes that are closer to the calculated points as fake sources, and then it broadcasts the fake sources assignment message by introducing the real source location. Following this the fake and real sources start to generate messages at the same pattern. In the case when BS accepts event notification from a large number of source nodes, it schedules the fake assignment or drops the old event notification message if more than one source node detects the same event. This contributes to the reduction of energy consumption.

    3.4 Discussion

    Random location of BS: for BS located randomly in WSN we can apply translation instead of rotation to select fake sources which also has isomorphic property to preserve distance.

    Non-beacon enabled WSNs: In this type of WSN it is possible to make every node to generate fake event notification messages at the beacon interval to provide un-observablity to the source with real event notification. To minimize energy cost, message size at this interval can be smaller than the size of the event reporting message in this interval.

    4 Analysis of the scheme

    4.1 Analytical analysis

    4.1.1 Privacy analysis

    Theorem 1: CTA satisfies clustered trajectories k-anonymity property.

    Proof: To prove that CTA satisfies clustered trajectories k-anonymity, first we have to show that CTA satisfies trajectory similarity and Ti={(x0, y0, t0), (x1, y1, t1),…,(xn, yn, tn)} is a real trajectory of the monitored object analyzed from the traffic collected at discrete time intervals {t0, t1,…,tn} where (t1

    To show the distance travelled by the fake trajectory (D′) at any point is the same as the real trajectory distance (D), suppose (xi, yi) and (xj, yj) are random locations of nodes that report real event occurrences at time tiand tj, distance D between (xi, yi) and (xj, yj) is

    公司產(chǎn)品的價(jià)格制定,一定要充分考慮收益、利潤(rùn)及價(jià)格與企業(yè)長(zhǎng)期發(fā)展的關(guān)系等因素。隨著紡織原材料價(jià)格的不斷增長(zhǎng),勞動(dòng)力成本的水漲船高,南通鵬越紡織有限公司一定要明確定價(jià)目標(biāo),創(chuàng)新定價(jià)方法,優(yōu)化定價(jià)策略。

    (1)

    Distance D′ after rotation of (xi, yi) and (xj, yj) at an angle j×β where 0

    This implies D=D′ when there is a deviation of the actual fake source location from the calculated

    D≈D′

    (2)

    Because of the dense deployment of WSN, this deviation is insignificant to differentiate the fake trajectory from the real one. Thus from Eqs(1,2) CTA satisfies trajectory similarity. And since a single source node location is rotated by angles β, 2β,…, (k-1)β, K-1 fake trajectories exist for each real trajectory, which implies CTA satisfies trajectory k-anonymity. Also for each real trajectory in trajectory set {T0, T1,…,Tn} which are analyzed from the traffic collected, it is possible to find k-1 collections of fake trajectories. When we categorize the fake trajectories by the angles of rotation we will find k-1 cluster of trajectories {C1, C2,…,Ck-1). C1contains a cluster of trajectories found by rotating points in T0, T1, …, Tnby an angle β, C2by an angle 2β,… and Ck-1by an angle (k-1)β. Thus ?Th, Ti∈{T0, T1,…,Tn}, ?Tjl, Tkl∈Clwhere l=1,2,…, k-1 such that Dist (Th, Ti)≈Dist (Tjl, Tkl). Therefore it is concluded that CTA satisfies clustered trajectories k-anonymity.

    Theorem 2: CTA satisfies the property of location k-diversity.

    4.1.2 Time delay analysis

    The time delay of CTA for sending messages is the same with that of the dynamic source anonymity after the fake sources are selected. However when the time it takes to send an event occurrence message to BS for fake source request, each node has its independent beacon interval: t=(td/2+ts)h, where h is the maximum number of hops from the BS, tsis the time to send event notification message and td/2 is the average time delay of a beacon message on each node that participates in forwarding the beacon message. But in CTA, beacon intervals overlap so that t=td/2 which is independent of h and ts. In case of dynamic source anonymity the time delay for sending fake request is t=(ts+tsd)h+tw, where tsdis the shortest on each sensor node, h is the number of hops from source to the BS and twis the waiting time at BS before it broadcasts the fake source assignment message.

    4.1.3 Energy overhead

    The energy overhead for dynamic source anonymity scheme when a single source node reports event related information is: ET=Ed+knEm, where Edis the average energy cost of sending event notification message, k is the total number of nodes that participate in sending messages at the same time interval, n is the total number of messages sent by each node, and Emis the average cost of sending a single message from source to BS. But since beacon messages have no energy overhead for CTA, ET= knEm.

    4.2 Simulation results

    A discrete simulator has been built up to evaluate the performance and security of our scheme using TinyOS. 1,000 sensor nodes are randomly deployed in a round area network with a diameter of 50. The transmission range is set to 1. The base station is located at the centre of the network. The network has one hot spot that is randomly located and fixed during each simulation run to show the hot spot created due to different objects that visit the area from different directions. In our simulation 10 objects visit the hot spot area repeatedly.

    4.2.1 Privacy analysis

    The false positive hot spot detection ratio of the attacker is compared when random selection and CTA scheme are applied. The attacker first performs rate monitoring to detect the most visited area and then studies the trajectory around the detected hot spot area to determine whether it is the real hot spot area or a fake one. In random selection, fake sources may or may not form the hot spot areas as randomly selecting fake sources which are distributed throughout the network. Even when fake sources form hot spots by chance there may not form a pattern of movement that resemble the real event movement around the hot spot. But in case of CTA isomorphic property of rotation copies any pattern observed by the real event report on fake reports.

    As it can be seen from Fig.5, when the number of traffic collected for analysis increases, the probability of detecting false hot spots increases. This is because hot spot is created when there is large volume of data in the network. When we compare the false positive hot spot detection ratio of our scheme with random selection, our scheme gives 90% of the time hot spot that is false positive for k=10. This is because there are exactly 9 false hot spots for a single hot spot which provides exact k-anonymity. But random selection of fake sources ends up in 50% probability of detecting false positive hot spot. Generaly, when we see the number of hot spots detected for different values of k, in the proposed scheme k hot spots are detected as we can see from Fig.6, which implies the probability of detecting the real hot spot is 1/k. But the number of hot spots detected in random selection is always less than k as indicated in Fig.6 and hence there is a chance to detect the real hot spot only.

    Fig.5 False positive hot spot detection ratio for k=10

    4.2.2 Time delay

    We compare the time delay for sending event notification in the case of Dynamic SSSA, CTA without beacon interval overlaps, and CTA. Dynamic SSSA has less time delay when there is less frequent fake request from the source node. But when the number of fake requests increases, the attacker may detect a consecutive short time delay which in turn implies that there are real event notifications. Thus, time delay increases for providing source location privacy. CTA without beacon interval overlap has a large time delay because each node has independent beacon interval and consequently cumulative delay is introduced for the event notification to arrive at BS. But the time delay decreases with significant amount when the beacon intervals overlap as we can see from Fig.7.

    Fig.6 Hot spots detected for different values of k

    Fig.7 Time delay comparison

    4.2.3 Energy consumption

    To send the same amount of message within a given time T, SSSA consumes more energy due to network wise dummy message injection. Dynamic SSSA reduces the energy cost by extending the delay of network wise dummy message injection. CTA consumes the least energy as fake request is sent within the beacon frames at MAC layer which have no energy overhead. Fig.8 shows this clearly.

    5 Conclusion

    This study proposes a scheme that provides perfect k-anonymity to the source that reports event occurrence in mobile events monitoring application against a global attacker. The analytical analysis and the simulation results show that our scheme is efficient and effective in providing the desired anonymity.

    Fig.8 Energy overhead

    [ 1] Mehta K, Liu D, Wright M. Location privacy in sensor networks against a global eavesdropper. In: Proceedings of the IEEE International Conference on Network Protocols, Beijing, China, 2007. 314-323

    [ 2] Shao M, Yang Y, Zhu S C, et al. Towards statistically strong source anonymity for sensor networks. In: Proceedings of the 27th IEEE Communications Society Conference on Computer Communications, Phoenix, USA, 2008. 466-474

    [ 3] Yang Y, Zhu S C, Cao G H, et al. An active global attack model for sensor source location privacy: analysis and countermeasures. Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, 2009, 19:373-393

    [ 4] Ortolani S, Conti M, Crispo B, et al. Events privacy in WSNs: A new model and its application. In: Proceedings of the IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks, Lucca, Italy, 2011. 1-9

    [ 5] Mahmoud M, Shen X M. A cloud-based scheme for protecting source-location privacy against hotspot-locating attack in wireless sensor Networks. IEEE Transactions on Parallel and Distributed Systems, 2012, 23(10):1805-1818

    [ 6] Kamat P, Zhang Y Y, Trappe W, et al. Enhancing source-location privacy in sensor network routing. In: Proceedings of the 25th IEEE international Conference on Distributed Computing Systems, Columbus, USA, 2005.599-608

    [ 7] Shao M, Hu W H, Zhu S C, et al. Cross-layer enhanced source location privacy in sensor networks. In: Proceedings of the 6th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks, Rome, Italy, 2009. 1-9

    [ 8] Li Z. ZigBee wireless sensor network in industrial applications. In: Proceedings of the 2006 SICE-ICASE International Joint Conference,Busan,Korea,2006.1067-1070

    [ 9] Avinash K, Sonu A. Location detection in wireless sensor network using classical optimization methodology. International Journal of Computer Science and Technology, 2012, 3(1):685-688

    [10] Zhao J. Research on key predistribution scheme of wireless sensor networks. In: Proceedings of the 5th International Conference on Intelligent Computation Technology and Automation, Zhangjiajie, China, 2012. 287-290

    [11] Iwanicki K, Van S M. On hierarchical routing in wireless sensor networks. In: Proceedings of the International Conference on Information Processing in Sensor Networks, San Francisco, USA, 2009. 133-144

    Seble Hailu Dady, born in 1976. She is Ph.D student in University of Electronic Science and Technology of China. She received B.Ed in Mathematics from Haramay University in 2000, B.Sc and M.S degree in Computer Science from Addis Ababa University in 2004 and 2008 respectively. Her research interest mainly focuses on privacy and security in wireless sensor networks and related applications.

    10.3772/j.issn.1006-6748.2015.02.004

    ①Supported by the National Natural Science Foundation of China (No. 60903157), the Fundamental Research funds for the Central Universities of China (No. ZYGX2011J066) and the Sichuan Science and Technology Support Project (No. 2013GZ0022).

    ②To whom correspondence should be addressed. E-mail: wangjh@uestc.edu.cn Received on Feb. 5, 2014, Qin Zhiguang, Yang Fan

    猜你喜歡
    充分考慮南通原材料
    水利工程原材料質(zhì)量檢測(cè)控制探討
    藍(lán)印花布:南通獨(dú)具特色的非遺傳承
    非遺南通
    南通職業(yè)
    觀點(diǎn)
    知識(shí)無(wú)窮盡
    意林(2019年16期)2019-09-04 21:00:12
    南通中船機(jī)械制造有限公司
    應(yīng)充分考慮醫(yī)院管理的復(fù)雜性
    含有字母的不等式(組)
    原版亦或重生
    汽車之友(2015年6期)2015-03-18 05:15:08
    一区二区日韩欧美中文字幕| 一级毛片女人18水好多| 一区二区av电影网| 人成视频在线观看免费观看| 大香蕉久久网| 亚洲av美国av| av视频免费观看在线观看| 侵犯人妻中文字幕一二三四区| 色播在线永久视频| 最近最新中文字幕大全电影3 | 首页视频小说图片口味搜索| 国产色视频综合| 亚洲黑人精品在线| 丝袜美腿诱惑在线| 人妻一区二区av| 80岁老熟妇乱子伦牲交| 天天影视国产精品| 超碰97精品在线观看| 下体分泌物呈黄色| 一本大道久久a久久精品| 看免费av毛片| 亚洲 欧美一区二区三区| 两个人看的免费小视频| 丝袜在线中文字幕| 欧美久久黑人一区二区| 亚洲国产中文字幕在线视频| 日韩大片免费观看网站| 最黄视频免费看| 欧美日韩视频精品一区| 成人手机av| 嫁个100分男人电影在线观看| 亚洲七黄色美女视频| 午夜福利免费观看在线| 国产精品.久久久| 十八禁网站网址无遮挡| 天堂动漫精品| 考比视频在线观看| 男女午夜视频在线观看| 丝袜在线中文字幕| 飞空精品影院首页| 纯流量卡能插随身wifi吗| 妹子高潮喷水视频| 久久国产精品影院| 欧美性长视频在线观看| netflix在线观看网站| 中文字幕高清在线视频| 丰满人妻熟妇乱又伦精品不卡| 国产单亲对白刺激| 国产单亲对白刺激| 欧美一级毛片孕妇| 亚洲成a人片在线一区二区| 菩萨蛮人人尽说江南好唐韦庄| 十八禁人妻一区二区| 这个男人来自地球电影免费观看| 中文字幕人妻丝袜制服| 香蕉丝袜av| 一二三四社区在线视频社区8| 久久毛片免费看一区二区三区| 久久精品熟女亚洲av麻豆精品| 国产淫语在线视频| 天天影视国产精品| 国产日韩欧美在线精品| 极品少妇高潮喷水抽搐| 三上悠亚av全集在线观看| 嫩草影视91久久| 成人国产一区最新在线观看| 悠悠久久av| 国产欧美日韩一区二区三区在线| 另类精品久久| 久久久精品区二区三区| 亚洲精品一卡2卡三卡4卡5卡| 国产精品久久久久久精品电影小说| 超碰成人久久| 久久久久网色| 亚洲精品成人av观看孕妇| 欧美精品亚洲一区二区| 777久久人妻少妇嫩草av网站| 热99国产精品久久久久久7| 午夜福利影视在线免费观看| 精品国产国语对白av| 啦啦啦视频在线资源免费观看| 久久午夜综合久久蜜桃| √禁漫天堂资源中文www| 最近最新中文字幕大全电影3 | av天堂久久9| 国产精品一区二区在线不卡| 久久ye,这里只有精品| 最近最新中文字幕大全电影3 | 欧美 日韩 精品 国产| 动漫黄色视频在线观看| 欧美精品亚洲一区二区| 国产精品亚洲一级av第二区| 国产视频一区二区在线看| 侵犯人妻中文字幕一二三四区| 妹子高潮喷水视频| 国产精品久久久久久精品古装| 久久久久网色| 成年人午夜在线观看视频| 又紧又爽又黄一区二区| 欧美日韩黄片免| 亚洲专区中文字幕在线| 欧美日韩亚洲高清精品| 大型黄色视频在线免费观看| 欧美+亚洲+日韩+国产| 色在线成人网| 丁香欧美五月| 欧美老熟妇乱子伦牲交| 18禁国产床啪视频网站| 男女床上黄色一级片免费看| 美女午夜性视频免费| 丁香六月天网| 中文欧美无线码| 国产一区二区三区综合在线观看| 精品少妇黑人巨大在线播放| 日本黄色日本黄色录像| 国产精品一区二区在线观看99| 国产av精品麻豆| 日本a在线网址| 久久天躁狠狠躁夜夜2o2o| 大香蕉久久成人网| 亚洲一卡2卡3卡4卡5卡精品中文| 99热国产这里只有精品6| 亚洲熟女精品中文字幕| 亚洲国产av新网站| 欧美黄色淫秽网站| 青青草视频在线视频观看| 一级毛片精品| 老司机深夜福利视频在线观看| 成年版毛片免费区| 婷婷成人精品国产| 男女午夜视频在线观看| 久热这里只有精品99| 午夜福利,免费看| 人人妻人人澡人人爽人人夜夜| 国内毛片毛片毛片毛片毛片| 人人妻人人澡人人爽人人夜夜| 免费女性裸体啪啪无遮挡网站| 国产精品久久久久久人妻精品电影 | 国产日韩欧美在线精品| 一区福利在线观看| 日韩欧美一区二区三区在线观看 | 狠狠精品人妻久久久久久综合| 国产高清视频在线播放一区| 国产一区有黄有色的免费视频| 国产不卡一卡二| 另类亚洲欧美激情| 五月开心婷婷网| 久久中文字幕人妻熟女| 国内毛片毛片毛片毛片毛片| 欧美亚洲 丝袜 人妻 在线| 亚洲国产av影院在线观看| 亚洲伊人久久精品综合| 一级片'在线观看视频| 男女高潮啪啪啪动态图| 久久久久久久久免费视频了| 中文字幕高清在线视频| 老司机亚洲免费影院| 天堂俺去俺来也www色官网| 国产在线一区二区三区精| 午夜福利视频在线观看免费| 国产日韩欧美在线精品| 国产一卡二卡三卡精品| 亚洲av电影在线进入| 51午夜福利影视在线观看| 中文欧美无线码| 久久这里只有精品19| 一级片'在线观看视频| 淫妇啪啪啪对白视频| 色综合婷婷激情| 少妇的丰满在线观看| 亚洲久久久国产精品| 大片电影免费在线观看免费| 热99国产精品久久久久久7| 亚洲色图av天堂| 最新美女视频免费是黄的| 日本av手机在线免费观看| 欧美精品啪啪一区二区三区| 热re99久久精品国产66热6| 欧美 日韩 精品 国产| tube8黄色片| kizo精华| 色94色欧美一区二区| 亚洲国产精品一区二区三区在线| bbb黄色大片| 亚洲国产欧美在线一区| 最新美女视频免费是黄的| 九色亚洲精品在线播放| 新久久久久国产一级毛片| 一个人免费看片子| 日韩欧美一区二区三区在线观看 | 久久国产精品影院| 亚洲 国产 在线| av在线播放免费不卡| 色94色欧美一区二区| 国产精品偷伦视频观看了| 黑人猛操日本美女一级片| 精品国产国语对白av| 久久久久久久大尺度免费视频| 久久毛片免费看一区二区三区| 欧美日韩亚洲综合一区二区三区_| 欧美日韩视频精品一区| 人妻久久中文字幕网| 男男h啪啪无遮挡| 妹子高潮喷水视频| 国产又爽黄色视频| 啦啦啦 在线观看视频| 99riav亚洲国产免费| 久久这里只有精品19| 国产在线精品亚洲第一网站| www日本在线高清视频| 999精品在线视频| 成人黄色视频免费在线看| 69精品国产乱码久久久| 自拍欧美九色日韩亚洲蝌蚪91| 久久国产亚洲av麻豆专区| 中文亚洲av片在线观看爽 | 99国产极品粉嫩在线观看| 人妻 亚洲 视频| 99国产综合亚洲精品| 麻豆av在线久日| 久久久久久亚洲精品国产蜜桃av| 色视频在线一区二区三区| 激情在线观看视频在线高清 | 男人舔女人的私密视频| 考比视频在线观看| 欧美在线一区亚洲| 在线av久久热| 99精品在免费线老司机午夜| 又紧又爽又黄一区二区| 波多野结衣av一区二区av| 精品乱码久久久久久99久播| 久久99一区二区三区| 久久久久久亚洲精品国产蜜桃av| 国产免费视频播放在线视频| 欧美成人午夜精品| 国产在线观看jvid| 日韩视频一区二区在线观看| 欧美日韩福利视频一区二区| 国产日韩一区二区三区精品不卡| 色婷婷久久久亚洲欧美| 久久人人97超碰香蕉20202| 日本精品一区二区三区蜜桃| 久久久久久久精品吃奶| 亚洲免费av在线视频| 多毛熟女@视频| 午夜福利视频精品| 在线十欧美十亚洲十日本专区| 黄色成人免费大全| 国产精品麻豆人妻色哟哟久久| 国产一区二区三区在线臀色熟女 | 国产精品亚洲一级av第二区| 精品视频人人做人人爽| 国产有黄有色有爽视频| 无人区码免费观看不卡 | 另类亚洲欧美激情| 国产精品 国内视频| av又黄又爽大尺度在线免费看| 国产成人精品久久二区二区免费| 久久久久久人人人人人| 欧美黑人欧美精品刺激| 亚洲av欧美aⅴ国产| 精品久久蜜臀av无| 国产成+人综合+亚洲专区| 亚洲精品中文字幕在线视频| 人妻 亚洲 视频| 国产精品久久久久久人妻精品电影 | 亚洲国产中文字幕在线视频| 亚洲欧美一区二区三区久久| 亚洲专区中文字幕在线| 国产在线免费精品| 99精品久久久久人妻精品| 精品一区二区三区av网在线观看 | 蜜桃在线观看..| 久久精品91无色码中文字幕| 欧美中文综合在线视频| 欧美日韩精品网址| 久久狼人影院| 亚洲国产精品一区二区三区在线| 国产97色在线日韩免费| 女人爽到高潮嗷嗷叫在线视频| 亚洲全国av大片| 日韩三级视频一区二区三区| 亚洲色图 男人天堂 中文字幕| 久久午夜综合久久蜜桃| 精品一区二区三区视频在线观看免费 | 午夜福利,免费看| 久久精品91无色码中文字幕| 搡老乐熟女国产| 不卡一级毛片| 亚洲精品一二三| 丝袜人妻中文字幕| 色婷婷av一区二区三区视频| www日本在线高清视频| 巨乳人妻的诱惑在线观看| 男女边摸边吃奶| 50天的宝宝边吃奶边哭怎么回事| 美女视频免费永久观看网站| 99re6热这里在线精品视频| 超色免费av| 精品国产乱子伦一区二区三区| 男人舔女人的私密视频| 亚洲熟妇熟女久久| 国产欧美日韩一区二区精品| 日本av手机在线免费观看| 久久婷婷成人综合色麻豆| 大香蕉久久网| 中文字幕人妻熟女乱码| 老司机深夜福利视频在线观看| 宅男免费午夜| 亚洲国产av新网站| 国产成人精品久久二区二区免费| 午夜福利欧美成人| 亚洲av日韩在线播放| 久久久久视频综合| 欧美日韩精品网址| h视频一区二区三区| 男女下面插进去视频免费观看| 自拍欧美九色日韩亚洲蝌蚪91| 国产精品一区二区精品视频观看| 精品视频人人做人人爽| cao死你这个sao货| 国产精品欧美亚洲77777| 欧美精品亚洲一区二区| 他把我摸到了高潮在线观看 | 一级a爱视频在线免费观看| 久久亚洲精品不卡| 欧美精品一区二区大全| 丰满迷人的少妇在线观看| 蜜桃国产av成人99| 欧美日韩亚洲国产一区二区在线观看 | 午夜福利免费观看在线| 亚洲成a人片在线一区二区| av不卡在线播放| 国产不卡一卡二| 亚洲人成电影观看| 青青草视频在线视频观看| 好男人电影高清在线观看| 亚洲av日韩精品久久久久久密| 国产极品粉嫩免费观看在线| 男男h啪啪无遮挡| 欧美精品高潮呻吟av久久| 亚洲色图av天堂| 国产主播在线观看一区二区| 一级片免费观看大全| 欧美大码av| 久久婷婷成人综合色麻豆| 成人免费观看视频高清| 国产欧美日韩综合在线一区二区| 黑人操中国人逼视频| tube8黄色片| 一级片免费观看大全| 9色porny在线观看| 国产高清视频在线播放一区| 久久国产精品大桥未久av| 人妻久久中文字幕网| 黑人巨大精品欧美一区二区mp4| 十八禁网站网址无遮挡| 国产国语露脸激情在线看| 搡老熟女国产l中国老女人| 在线观看一区二区三区激情| 亚洲精品中文字幕一二三四区 | 国产精品影院久久| 18禁美女被吸乳视频| 人妻久久中文字幕网| 热re99久久精品国产66热6| 国产av精品麻豆| 精品欧美一区二区三区在线| 亚洲欧美日韩另类电影网站| 美女高潮到喷水免费观看| 十分钟在线观看高清视频www| 激情在线观看视频在线高清 | 久久亚洲精品不卡| 日本av免费视频播放| 啦啦啦在线免费观看视频4| 一个人免费看片子| 国产精品久久久久久人妻精品电影 | 中文字幕高清在线视频| 天天躁狠狠躁夜夜躁狠狠躁| 亚洲精华国产精华精| 精品少妇久久久久久888优播| 老熟妇仑乱视频hdxx| 性色av乱码一区二区三区2| 欧美中文综合在线视频| 亚洲国产欧美日韩在线播放| 久久人妻福利社区极品人妻图片| 成人三级做爰电影| 日本黄色视频三级网站网址 | 日韩熟女老妇一区二区性免费视频| 亚洲国产欧美在线一区| 国产亚洲av高清不卡| 视频区欧美日本亚洲| 亚洲人成伊人成综合网2020| 欧美性长视频在线观看| 91麻豆精品激情在线观看国产 | 亚洲国产成人一精品久久久| 亚洲av电影在线进入| 国产单亲对白刺激| 侵犯人妻中文字幕一二三四区| 日韩熟女老妇一区二区性免费视频| 久久久久国产一级毛片高清牌| 侵犯人妻中文字幕一二三四区| 国产欧美亚洲国产| 国产区一区二久久| 亚洲精品一卡2卡三卡4卡5卡| 午夜福利免费观看在线| 国产精品 欧美亚洲| 亚洲国产欧美日韩在线播放| 香蕉丝袜av| 色尼玛亚洲综合影院| videos熟女内射| 天堂中文最新版在线下载| 国产人伦9x9x在线观看| 日韩免费av在线播放| 男女下面插进去视频免费观看| 久久午夜综合久久蜜桃| 午夜免费成人在线视频| 又大又爽又粗| 国产1区2区3区精品| 国产精品香港三级国产av潘金莲| 国产色视频综合| 一本久久精品| 一本一本久久a久久精品综合妖精| 亚洲免费av在线视频| 亚洲三区欧美一区| 欧美日韩亚洲综合一区二区三区_| 丰满人妻熟妇乱又伦精品不卡| 成人国产一区最新在线观看| 纵有疾风起免费观看全集完整版| 国产一区二区 视频在线| 免费看a级黄色片| 99riav亚洲国产免费| 欧美乱码精品一区二区三区| 1024香蕉在线观看| 亚洲人成伊人成综合网2020| www日本在线高清视频| 老汉色av国产亚洲站长工具| 精品久久蜜臀av无| 女人被躁到高潮嗷嗷叫费观| 少妇粗大呻吟视频| 波多野结衣av一区二区av| 这个男人来自地球电影免费观看| 男男h啪啪无遮挡| 国产精品久久久人人做人人爽| 人人妻人人澡人人看| 国产精品免费视频内射| 国产精品av久久久久免费| 精品第一国产精品| 久久精品熟女亚洲av麻豆精品| 免费在线观看黄色视频的| 亚洲精品一卡2卡三卡4卡5卡| 人人妻人人澡人人爽人人夜夜| 国产片内射在线| videos熟女内射| 久9热在线精品视频| 国产老妇伦熟女老妇高清| av有码第一页| 久久ye,这里只有精品| 丁香六月天网| 欧美日韩视频精品一区| 曰老女人黄片| 极品教师在线免费播放| 高潮久久久久久久久久久不卡| 丝袜喷水一区| 波多野结衣一区麻豆| 亚洲精品一卡2卡三卡4卡5卡| 叶爱在线成人免费视频播放| 人人妻人人添人人爽欧美一区卜| 一区福利在线观看| 狂野欧美激情性xxxx| 久久久久久久国产电影| 最近最新免费中文字幕在线| 欧美 亚洲 国产 日韩一| 十八禁网站网址无遮挡| 法律面前人人平等表现在哪些方面| 97在线人人人人妻| 亚洲五月婷婷丁香| 老司机深夜福利视频在线观看| 国产在线免费精品| 国产高清videossex| 久久久国产欧美日韩av| 日本一区二区免费在线视频| 一区二区三区激情视频| 大型av网站在线播放| 91九色精品人成在线观看| 老汉色av国产亚洲站长工具| 久久久精品94久久精品| 女人高潮潮喷娇喘18禁视频| 久久久久国内视频| 成年人午夜在线观看视频| 日本黄色日本黄色录像| 精品一区二区三区av网在线观看 | 俄罗斯特黄特色一大片| 久久青草综合色| 亚洲精品久久成人aⅴ小说| 亚洲免费av在线视频| 飞空精品影院首页| 免费看十八禁软件| 搡老熟女国产l中国老女人| 性高湖久久久久久久久免费观看| 婷婷丁香在线五月| 丁香六月天网| 一区二区av电影网| 婷婷成人精品国产| 黄片大片在线免费观看| 国产精品免费视频内射| 交换朋友夫妻互换小说| 一本综合久久免费| 91成年电影在线观看| 亚洲 欧美一区二区三区| 高清视频免费观看一区二区| 久久久久精品国产欧美久久久| 久久精品熟女亚洲av麻豆精品| 成人国语在线视频| 91字幕亚洲| 麻豆国产av国片精品| 久久久久久免费高清国产稀缺| 亚洲一区二区三区欧美精品| 水蜜桃什么品种好| 欧美精品人与动牲交sv欧美| 国产在视频线精品| 成在线人永久免费视频| 免费一级毛片在线播放高清视频 | 国产男女超爽视频在线观看| 啦啦啦视频在线资源免费观看| 欧美黄色片欧美黄色片| 正在播放国产对白刺激| 国产免费av片在线观看野外av| av片东京热男人的天堂| 狠狠精品人妻久久久久久综合| 如日韩欧美国产精品一区二区三区| 中文亚洲av片在线观看爽 | 母亲3免费完整高清在线观看| 五月天丁香电影| 少妇裸体淫交视频免费看高清 | 少妇猛男粗大的猛烈进出视频| 亚洲精品美女久久久久99蜜臀| 亚洲欧洲精品一区二区精品久久久| 国产不卡一卡二| 亚洲avbb在线观看| 成年动漫av网址| 国产单亲对白刺激| 涩涩av久久男人的天堂| 国产欧美日韩综合在线一区二区| 亚洲国产欧美日韩在线播放| 日本av免费视频播放| 不卡一级毛片| 成人手机av| 精品国产超薄肉色丝袜足j| 久久性视频一级片| 日韩中文字幕欧美一区二区| 亚洲精品自拍成人| 欧美日韩精品网址| 亚洲精华国产精华精| 一二三四在线观看免费中文在| 中亚洲国语对白在线视频| 国产精品国产高清国产av | 国产成人精品在线电影| 一级毛片女人18水好多| 老汉色av国产亚洲站长工具| 日本av手机在线免费观看| 亚洲精品美女久久av网站| 免费在线观看日本一区| 91精品三级在线观看| 午夜免费鲁丝| 91精品三级在线观看| 99精品在免费线老司机午夜| 啦啦啦 在线观看视频| 91精品国产国语对白视频| 日韩熟女老妇一区二区性免费视频| 男男h啪啪无遮挡| 日韩视频一区二区在线观看| 黑人操中国人逼视频| 中文字幕人妻丝袜制服| 欧美国产精品va在线观看不卡| 国产成人精品无人区| 国产精品免费视频内射| 国产精品成人在线| 婷婷成人精品国产| 女警被强在线播放| 99国产极品粉嫩在线观看| 十分钟在线观看高清视频www| 国产真人三级小视频在线观看| 国产激情久久老熟女| 国产免费av片在线观看野外av| 最新美女视频免费是黄的| 亚洲精品美女久久久久99蜜臀| 少妇猛男粗大的猛烈进出视频| 成在线人永久免费视频| 不卡av一区二区三区| 一级片免费观看大全| 日韩中文字幕欧美一区二区| 波多野结衣一区麻豆| 亚洲全国av大片| 在线av久久热| av欧美777| 亚洲欧洲精品一区二区精品久久久| 美女高潮喷水抽搐中文字幕| 少妇猛男粗大的猛烈进出视频| 大陆偷拍与自拍| 女性被躁到高潮视频| 视频在线观看一区二区三区| 亚洲av欧美aⅴ国产| 超碰成人久久| 久久免费观看电影| 国产精品久久久久久精品古装| 国产成人av教育| 国产精品久久久av美女十八| 国产精品 欧美亚洲| 黄片小视频在线播放| 日日爽夜夜爽网站| 久久中文字幕人妻熟女| 色婷婷久久久亚洲欧美| 男女边摸边吃奶| 亚洲精品在线美女| 99热国产这里只有精品6|