Dynamic routing network algorithm of low voltage
信息类专业英语翻译
Dynamic topology:As the channel of communicationchanges, some of the neighbors who were reachable on theprevious channel might not be reachable on the currentchannel and vice versa. As a result the topology of the network changes with the change in frequency of operation resulting in route failures and packet loss.Heterogeneity:Different channels may support differenttransmission ranges, data rates and delay characteristics.Spectrum-Handoff delay:For each transition from onechannel to another channel due to the PU’s activity, thereis a delay involved in the transition called Spectrum- Handoff delay.All these factors decrease the predictability of the cause oftransit-delay and subsequent packet loss on the network. Thetime latency during channel hand-off in cognitive networksmight cause the TCP round trip timer to time out. TCP willwrongly recognize the delays and losses due to the abovefactors as network congestion and immediately take steps toreduce the congestion window size knowing not the cause ofpacket delay. This reduces the efficiency of the protocol insuch environments.动态技术：随着信道通信的变化，一些邻进信道的用户在原信道没有发生变化而在新信道发生变化，或者相反。
网络异常检测与攻击隔离技术(看2.3)
关于论文使用授权的说明 学位论文作者完全了解北京邮电大学有关保留和使用学位论文的规定， 即：研究生在校攻读学位期间论文工作的知识产权单位属北京邮电大学。学 校有权保留并向国家有关部门或机构送交论文的复印件和磁盘，允许学位论 文被查阅和借阅；学校可以公布学位论文的全部或部分内容，可以允许采用 影印、缩印或其它复制手段保存、汇编学位论文。（保密的学位论文在解密 后遵守此规定）
２００４．１２
２００５．０６
２００５．１２
２００６．０６
２００６．１２
２００７．０６
２００７．１２
（抟耀（钠俐．ｃｍＩｃ．ｃｎ，２００７．１２
图１．１中国网民人数增长情况
北京１１１１ｊｔｌＬ大学Ｔ程硕Ｉ：学位论文
网络技术的快速发展与应用普及使得互联网空间的信息量急剧增加，随着 网民网络应用同趋丰富，尤其是在线视频业务飞速发展，网络带宽也呈现急剧 增长的趋势，截至２００７年１２月，中国互联网国际出口带宽数达到 ３６８，９２７Ｍｂｐｓ，年增长率为４３．７％。
ｉｎｔｅｒｖａｌｓ
ｆｉｎｄ
ｔｈｅ ａｎｏｍａｌｙ．Ｓｉｍｕｌａｔｉｏｎ
ｅｘｐｅｒｉｍｅｎｔｓ ｉｎｄｉｃａｔｅ ｔｈｅ
ｓｉｍｉｌａｒｉｔｙ ｓｔａｔｉｓｔｉｃ ｂａｓｅｄ ａｎｏｍａｌｙ ｄｅｔｅｃｔｉｏｎ ｍｅｔｈｏｄ ｉＳ ｅｆｆｅｃｔｉｖｅ．Ｔｈｅｎ ｔｈｅ ａｌｇｏｒｉｔｈｍ ｏｆ ａｎｏｍａｌｙ ｓｉｇｎａｔｕｒｅ ｅｘｔｒａｃｔｉｏｎ ｉｓ ｓｔｕｄｉｅｄ ａｆｔｅｒ ｔｈｅ ａｎｏｍａｌｙ ｉｓ
ｏｃｃｕｒ
ａｎｄ ｂｒｉｎｇ ｍｕｃｈ ｍｏｒｅ ｈａｒｍ．Ａｍｏｎｇ ａ１１ ｔｈｅ ｉｎｔｒｕｓｉｏｎｓ ｔｈｅ ａｎｏｍａｌｉｅｓ ｔｈａｔ ｔｅｎｄ ｔｏ ｅｘｈａｕｓｔ ｎｅｔｗｏｒｋ ｂａｎｄｗｉｄｔｈ ｓｕｃｈ
主动队列管理下大时滞网络路径拥塞控制算法
文章编号：1006 - 9348 (2021)03 - 0268 - 04主动队列管理下大时滞网络路径拥塞控制算法刘国芳，张炜(四川大学锦江学院，四川眉山620860)摘要:与传统的无线网络相比，大时滞网络对路径拥塞环境下的无线通道交换具有较高的要求。
为此提出主动队列管理下 大时滞网络路径拥塞控制算法。
首先利用主动队列管理算法对相邻路由节点网络路径的拥塞情况展开预测，进而分析网络 路由节点的队列状态;然后以优化后续节点队列、传输距离以及传输方向为目的，从路径概率选择、分组丢弃函数、WSN蚁 群路由选取三个角度优化网络路径，从而实现路径拥塞控制。
实验结果表明，上述算法能够有效缩短网络的传输时滞，且能 耗和丢包率较低，具有较高的应用价值。
关键词：主动队列管理;无线通道;交换网络;路由；拥塞控制中图分类号:TP399 文献标识码：BPath Congestion Control Algorithm for Large TimeDelay Networks under Active Queue Management 第38卷第3期__________________________计算机仿真____________________________2021年3月LIU Guo -fan g,Z H A N G W ei(Jinjiang College,Sichuan University,Meishan Sichuan620860，China)ABSTRACT：In the large time - delay network,there is a high demand for wireless channel switching in path congestion environment.In this regard,this paper puts forward a path congestion control algorithm with active queue management for large delay networks.Firstly,based on the active queue management algorithm,the congestion of the network path of the adjacent routing nodes was predicted,and the queue status of the network routing nodes was analyzed.Secondly,the optimization of subsequent node queue,transmission distance and transmission direction were taken as indicators to optimize the network path from path probability selection,packet drop function and WSN ant colony routing selection.Eventually,path congestion control was completed.The simulation results show that the algorithm has short transmission delay,low energy consumption and packet loss rate,and high practicability.KEYW ORDS：Active queue management；Wireless channel；Switching network；Routing；Congestion controli引言无线通道交换网络是设定在监测区域中的一些小型路 由节点，通过无线通信的方式衍生出的具有多跳性、自组织 性的网络系统[|]。
correlated equilibrium
Distributed Energy-Efficient Cooperative Routing in Wireless Networks Ahmed S.Ibrahim,Zhu Han†,and K.J.Ray LiuDepartment of Electrical and Computer Engineering,University of Maryland,College Park,MD20742,USA †Department of Electrical and Computer Engineering,Boise State University,Boise,ID83725,USAAbstract—Recently,cooperative routing in wireless networks has gained much interest due to its ability to exploit the broadcast nature of the wireless medium in designing power-efficient routing algorithms.Most of the existing cooperation-based routing algorithms are implemented byfinding a shortest-path routefirst.As such,these routing algorithms do not fully exploit the merits of cooperative communications at the physical layer.In this paper,we propose a cooperation-based routing algorithm,namely,Minimum Power Cooperative Routing (MPCR)algorithm,which makes full use of the cooperative communications while constructing the minimum-power route. The MPCR algorithm constructs the minimum-power route as a cascade of the minimum-power single-relay building blocks from the source to the destination.Hence,any distributed shortest-path algorithm can be utilized tofind the optimal route with polynomial complexity,while guaranteeing certain throughput. We show that the MPCR algorithm can achieve power saving of57.36%compared to the conventional shortest-path routing algorithms.Furthermore,the MPCR algorithm can achieve power saving of37.64%compared to the existing cooperative routing algorithms,in which the selected routes are constructed based on the noncooperative routes.I.I NTRODUCTIONIn wireless networks such as ad hoc networks,nodes spend most of their power in communication,either sending their own data or relaying other nodes’data[1].Therefore,de-signing power-efficient routing algorithms is one of the major concerns in wireless networks.Furthermore,the communi-cation power can be reduced by jointly considering other layers’protocols,which make use of the broadcast nature of the wireless medium.Moreover,these algorithms should be implemented in a distributed way.Therefore,the main goal of this paper is to design a distributed minimum-power routing algorithm for wireless networks,which exploits the broadcast nature of the wireless medium.Recently,cooperative communication for wireless networks has gained much interest due to its ability to mitigate fading through achieving spatial diversity,while resolving the diffi-culties of installing multiple antennas on small communication terminals.In cooperative communications,relays are assigned to help a sender in forwarding its information to its receiver. Thus,the receiver gets several replicas of the same informa-tion via independent channels.Various cooperative diversity protocols were proposed and analyzed in[2]-[10].The classical relay channel model based on additive white Gaussian noise(AWGN)channels was presented in[2].In[3], Laneman et al.described various techniques of cooperative communication,such as decode-and-forward,amplify-and-forward,selection relaying,and incremental relaying.In[4],a distributed space-time coded(STC)cooperative scheme was proposed by Laneman et al.In[5]and[6],Sendonaris et al.introduced user cooperation diversity.A two-user CDMA cooperative system,where both users are active and use orthogonal codes,was implemented in this two-part series. In[7],[8],relay-selection schemes for single-and multi-node decode-and-forward cooperative systems were proposed.In [9],the authors have provided SER performance analysis for the decode-and-forward multi-node scheme.Finally,a distrib-uted relay-assignment algorithm for wireless communications has been proposed in[10].The merits of the cooperative communications in the phys-ical layer have been explored.However,the impact of the co-operative communications on the design of the higher layers is not well-understood yet.Routing algorithms,which are based on the cooperative communications and known as cooperative routing[11],is an interesting research area and can lead to significant power savings.The cooperative routing proposed in [11]makes use of two facts:the Wireless Broadcast Advantage (WBA)in the broadcast mode and the Wireless Cooperative Advantage(WCA)in the cooperative mode.In the broadcast mode each node sends its data to more than one node,while in the cooperative mode many nodes send the same data to the same destination.The cooperative routing problem has been recently consid-ered in the literature[11]-[15].In[11],the optimum route is found through a dynamic programming algorithm.In[12], the minimum-power route is chosen while guaranteeingfixed transmission rate.In[13],Li et al.proposed the Cooperative Shortest Path(CSP)algorithm,which chooses the next node in the route that minimizes the power transmitted by the last L nodes added to the route.Sikora et al.presented in[14] an information-theoretic viewpoint of the cooperative routing in linear wireless network for both the power-limited and bandwidth-limited regimes.In addition,the authors in[14] analyzed the transmitted power,required to achieve a desired end-to-end rate.In[15],the authors proposed three cooperative routing algorithms,namely,relay-by-flooding,relay-assisted routing,and relay-enhanced routing.Most of the existing cooperation-based routing algorithms are implemented byfinding a shortest-path routefirst.Since the cooperative route is based on the shortest-path one,these routing algorithms do not fully exploit the merits of cooper-ative communications at the physical layer.This is our main motivation to propose a cooperation-based routing algorithm that takes into consideration the effect of the cooperative communications while constructing the minimum-power route. In this paper,we consider the minimum-power routing prob-lem with cooperation in wireless networks.The optimum route is defined as the route that requires the minimum transmitted power while guaranteeing certain Quality of Service(QoS).The QoS is characterized by the end-to-end throughput.We derive a cooperation-based link cost formula,which represents the minimum transmitted power that is required to guarantee the desired QoS over a particular link.The main contribu-tion of this paper is the proposed cooperation-based routing algorithm,namely the Minimum Power Cooperative Routing (MPCR)algorithm,which can choose the minimum-power route while guaranteeing the desired QoS.It will be shown that the MPCR algorithm can achieve power saving of57.36% compared to the conventional shortest-path routing algorithms. Furthermore it can achieve power saving of37.64%with re-spect to the Cooperation Along the Shortest Non-Cooperative Path(CASNCP)algorithm,whichfinds the shortest-path route first then it applies the cooperative communication upon the constructed route to reduce the transmitted power.The rest of the paper is organized as follows.In the next section,we formulate the minimum-power routing problem.In addition,we describe the network model and derive closed-form expressions for the minimum transmitted power per hop in Section II.We describe two cooperation-based routing algorithms in Section III.In Section IV,we show the numer-ical results for the power savings of the proposed algorithm. Finally,Section V concludes the paper.II.S YSTEM M ODEL AND L INK A NALYSISIn this section,we describe the network model and formulate the minimum-power routing problem.Then,we present the di-rect transmission and cooperative transmission modes.Finally, we derive the required power for these two transmission modes in order to achieve certain throughput.work ModelWe consider a graph G(N,E)with N nodes and E edges. Given any source-destination pair(S,D)∈{1,...,N},the goal is tofind the S−D route that minimizes the total transmitted power,while satisfying a specific throughput.For a given source-destination pair,denoteΩas the set of all possible routes,where each route is defined as a set consisting of its hops.For a routeω∈Ω,denoteωi as the i-th hop of this route.Thus,the problem can be formulated asmin ω∈Ωωi∈ωPωis.t.ηω≥ηo,(1)where Pωi denotes the transmitted power over the i-th hop,ηωis the end-to-end throughput,andηo represents the minimumdesired value of the end-to-end throughput.Letηωi denote thethroughput of the i-th hop,which is defined as the number of successfully transmitted bits per second per hertz(b/s/Hz)of a given hop.Furthermore,the end-to-end throughput of a certain routeωis defined as the minimum of the throughput values of the hops constituting this route,i.e.,ηω=minωi∈ωηωi.(2)It has been proven in[13]that the Minimum Energy Cooperative Path(MECP)routing problem,i.e.,tofind the minimum-energy route using cooperative radio transmission,isDTFig.1.Cooperative Transmission(CT)and Direct Transmission(DT)modes as building blocks for any route.NP-complete.This is due to the fact that the optimal path could be a combination of cooperative transmissions and broadcast transmissions.Therefore,we consider two types of building blocks:direct transmission(DT)and cooperative transmission (CT)building blocks.In Fig.1the DT block is represented by the link(i,j),where node i is the sender and node j is the receiver.In addition,the CT block is represented by the links (x,y),(x,z),and(y,z),where node x is the sender,node y is a relay,and node z is the receiver.The route can be considered as a cascade of any number of these two building blocks,and the total power of the route is the summation of the transmitted powers along the route.Thus,the minimization problem in(1) can be solved by applying any distributed shortest-path routing algorithm such as the Bellman-Ford algorithm[16].B.Direct and Cooperative Transmission ModesLet h u,v,d u,v,and n u,v represent the channel coefficient, length,and additive noise of the link(u,v),respectively.For the direct transmission between node i and node j,the received symbol can be modeled asr D i,j=P D d−αi,jh i,j s+n i,j,(3) where P D is the transmitted power in the direct transmission mode,αis the path loss exponent,and s is the transmitted symbol.For the cooperative transmission,we consider a modified version of the decode-and-forward incremental relaying coop-erative scheme,proposed in[3].The transmission scheme for a sender x,a relay y,and a receiver z,can be described as follows.The sender sends its symbol in the current time slot. Due to the broadcast nature of the wireless medium,both the receiver and the relay receive noisy versions of the transmitted symbol.The received symbols at the receiver and the relay can be modeled asr C x,z=P C d−αx,zh x,z s+n x,z,(4) andr C x,y=P C d−αx,yh x,y s+n x,y,(5) respectively,where P C is the source transmitted power in the cooperative transmission mode.Once the symbol is received,the receiver and the relay decode it.We assume that the relay and the receiver decide that the received symbol is correctly received if the received signal-to-noise ratio(SNR)is greater than a certain threshold, which depends on the transmitter and the receiver structures. Such system suffers from error propagation but its effect can be neglected.The rationale behind this is that when the relays operate in a high SNR regime,the dominant source of error isthe channel being in outage,i.e.,deep fade,which corresponds to the SNR falling below some threshold.This result has been proven in [17].If the receiver decodes the symbol correctly,then it sends an acknowledgment (ACK)to the sender and the relay to confirm a correct reception.Otherwise,it sends a negative acknowledgment (NACK)that allows the relay,if it received the symbol correctly,to transmit this symbol to the receiver in the next time slot.This model represents a modified form of the Automatic Repeat Request (ARQ),where the relay retransmits the data instead of the sender,if necessary.The received symbol at the receiver can be written asr Cy,z =P C d −αy,z h y,z s +n y,z .(6)In general,the relay can transmit with a power that is differentfrom the sender power P C .However,this complicates the problem of finding the minimum-power formula,as will be derived later.For simplicity,we consider that both the sender and the relay send their data employing the same power P C .In this paper,flat quasi-static fading channels are con-sidered,hence,the channel coefficients are assumed to be constant during a complete frame,and may vary from a frame to another.We assume that all the channel terms are independent complex Gaussian random variables with zero mean and unit variance.Finally,the noise terms are modeled as zero-mean,complex Gaussian random variables with equal variance N 0.C.Link Cost FormulationSince the throughput is a continuous monotonously-increasing function of the transmission power,the optimization problem in (1)has the minimum when ηω=ηo ,∀ω∈Ω.Since the end-to-end throughput ηω=min ωi ∈ωηωi ,then the optimum power allocation,which achieves a desired throughput ηo along the route ω,forces the throughput at all the hops ηωi to be equal to the desired one,i.e.,ηωi =ηo ,∀ωi ∈ω.(7)Thisresult can be explained as follows.LetP ∗ω1,P ∗ω2,···,P ∗ωnrepresent the required powers on a route consisting of n hops,where P ∗ωiresults in ηωi =ηo for i =1,···,n .If we increase the power of the i-th blockto P ωi >P ∗ωithen the resulting throughput of the i-th block increases,i.e.ηωi >ηo ,while the end-to-end throughput does not change as min ωi ∈ωηωi =ηo .Therefore,no need to increase the throughput of any hop over ηo ,which is indicated in (7).Since the throughput of a given link ωi is defined as the number of successfully transmitted bits per second per hertz,thus it can be calculated asηωi =p S ωi ×R ωi ,(8)where p S ωi and R ωi denote the per-link probability of success and transmission rate,respectively.We assume that the desired throughput can be factorized asηo =p S ×R o ,(9)where p S o and R o denote the desired per-link probability ofsuccess and transmission rate,respectively.In the sequel,we calculate the required transmitted power in order to achieve the desired per-link probability of success and transmission rate for both the direct and cooperative transmission modes.We note that the channel gain |h u,v |2between any two nodes u and v ,is exponentially distributed with parameter one [18].For the direct transmission mode in (3),the mutual infor-mation between sender i and receiver j can be given byI i,j =log1+P D d −αi,j |h i,j |2N 0.(10)Without loss of generality,we have assumed unit bandwidth in (10).The outage probability is defined as the probability that the mutual information is less than the required transmission rate R o .Thus,the outage probability of the link (i,j )is calculated asp Oi,j =Pr(I i,j ≤R o )=1−exp −(2R o −1)N 0d αi,j P o.(11)If an outage occurs,the data is considered lost.The probabilityof success is calculated as p S i,j =1−p Oi,j .Thus,to achieve the desired p S o and R o for direct transmission mode,the required transmitted power isP D=(2R o −1)N 0d αi,j−log(p o ).(12)For the cooperative transmission mode,the total outage probability is given byp O x,y,z=Pr(I x,z ≤R C )·Pr(I x,y ≤R C )+Pr(I x,z ≤R C )×1−Pr(I x,y ≤R C ) ×Pr(I y,z ≤R C ),(13)where R C denotes the transmission rate for each time slot.In (13),the first term corresponds to the event when both the sender-receiver and the sender-relay channels are in outage,and the second term corresponds to the event when both the sender-receiver and relay-receiver channels are in outage but the sender-relay is not.Consequently,the probability of success of the cooperative transmission mode can be calculated asp S =exp −g d αx,z +exp −g (d αx,y +d αy,z) −exp −g (d αx,y +d αy,z +d αx,z ) ,(14)whereg =(2R C−1)N 0P .(15)In (13)and (14),we assume that the receiver decodes the signals received from the relay either at the first time slot or at the second time slot,instead of combining the received signals together.In general,Maximum Ratio Combining (MRC)[19]at the receiver gives a better result.However,it requires the receiver to store an analog version of the received data from the sender,which is not practical.The probability that the source transmits only,denoted by Pr(φ),is calculated as Pr(φ)=1−Pr(I x,z ≤R C )+Pr(I x,z ≤R C )Pr(I x,y ≤R C )=1−exp −g d αx,y +exp −g (d αx,y +d αx,z ) ,(16)where the term 1−Pr(I x,z ≤R C )corresponds to the event when the sender-receiver channel is not in outage,while the other term corresponds to the event when both the sender-receiver and the sender-relay channels are in outage.The probability that the relay cooperates with the source is calculated asPr(φ)=1−Pr(φ).(17)Thus,the average transmission rate of the cooperative trans-mission mode can be calculated asR =R C·Pr(φ)+R C 2·Pr(φ)=R C 21+Pr(φ) ,(18)where R C corresponds to the transmission rate if the sender is sending alone in one time slot and R C /2corresponds to the transmission rate if the relay cooperates with the sender in the consecutive time slot.We set the probability of success in (14)as p S =p S o and the average transmission rate in (18)as R =R o .By approximating the exponential functions in (14)as exp(−x )≈1−x +x 2/2,we obtaing ≈1−p S od eq,(19)where d eq =d αx,z (d αx,y +d αy,z).Thus,R C can be obtained using (18)asR C =2R o 1+Pr(φ)≈2R o 2−exp − 1−p S o d eq d αx,y +exp − 1−p S o d eq (d αx,y +d αx,z ),(20)where we substituted (19)in (16).In addition,the required power per link can be calculated using (15)and (19)as P C ≈(2R C −1)N 0d eq1−p S o .(21)Finally,the average transmitted power of the cooperative transmission can be calculated asP C avg =P C ·Pr(φ)+2P C ·Pr(φ)=P C2−Pr(φ) ,(22)where Pr(φ)and P C are given in (16)and (21),respectively.III.C OOPERATION -B ASED R OUTING A LGORITHMS In this section,we propose two cooperation-based routing algorithms,which require polynomial complexity to find the minimum-power route.We assume that each node broad-casts periodically HELLO packet to its neighbors to update the topology information.In addition,we consider a simple Medium Access Control (MAC)protocol,which is the conven-tional Time Division Multiple Access (TDMA)scheme with equal time slots.First,we describe the proposed MPCR algorithm for a wireless network of N nodes.The MPCR algorithm can be distributively implemented by the Bellman-Ford shortest path TABLE I MPCR Algorithm.Step 1Each node x ∈{1,...,N }behaving as a sender calcu-lates the cost of the its outgoing link (x,z ),where z ∈N (x )isthe receiver as follows.For each other node y ∈N (x ),y =z ,node x calculates the cost of the cooperative transmission in (22)employing node y as a relay.Step 2The cost of the (x,z )-th link is the minimum cost among all the costs obtained in Step 1.Step 3If the minimum cost corresponds to a certain relay y ∗,node x employs this relay to help the transmission over that hop.Otherwise,it uses the direct transmission over this hop.algorithm [16].The derived power formulas for direct trans-mission and cooperative transmission are utilized to construct the minimum-power route.In the Bellman-Ford shortest path algorithm,each node i ∈{1,...,N }executes the iteration D i =min j ∈N (i )(d αi,j +D j ),where N (i )denotes the set of neighboring nodes of node i and D j represents the latest estimate of the shortest path from node j to the destination [16],which is included in the HELLO packet.Therefore,the MPCR algorithm is implemented by letting each node calcu-late the costs of its outgoing links then apply the Bellman-Ford algorithm.Table I describes the MPCR algorithm in details.The worst-case computational complexity of calculating thecosts at each node is O (N 2)since it requires two nested loops,and each has the maximum length of N to calculate all the possible cooperative transmission blocks.Second,we propose a cooperation-based routing algorithm,namely,Cooperation Along the Shortest Non-Cooperative Path(CASNCP)algorithm.The CASNCP algorithm is similar to the heuristic algorithms proposed by Khandani et al.in [11]and Yang et al.in [12]as it applies cooperative communica-tions upon the shortest-path route.However,it is implementedin a different way using the proposed cooperation-based link cost formula.First,it chooses the shortest-path route then itapplies the cooperative transmission mode upon each threeconsecutive nodes in the chosen route;first node as the sender,second node as the relay,and third node as the receiver.Table II describes the CASNCP algorithm.IV.N UMERICAL R ESULTS In this section,we present some computer simulations to illustrate the power savings of our proposed MPCR algorithm.We consider a 200×200grid,where N nodes are uniformlydistributed.The additive white Gaussian noise has varianceN 0=−70dBm.Given a certain network topology,we randomly choose a source-destination pair and apply the various routing algorithms,discussed in Section III,to choose the corresponding route.For each algorithm,we calculate the total transmitted power per route.Finally,these quantities are averaged over 1000different network topologies.First,we illustrate the effect of varying the desired through-put on the required transmitted power per route.Fig.2depicts the transmitted power per route,required by the different routing algorithms for path loss α=2and α=4.As shown,the transmitted power increases with α,which is obvious in (12),and can be shown in (22),that the transmitted power isTABLE II CASNCP Algorithm.Step 1Implement the Shortest Non-Cooperative Path (SNCP)algorithm using the distributed Bellman-Ford algorithm to choose the conventional shortest-path route ωS as follows.Each node i ∈{1,...,N }executes the iteration D i =min j ∈N (i )(d αi,j +D j ),where N (i )denotes the set of neigh-boring nodes of node i and D j represents the latest estimate of the shortest path from node j to the destination.Step 2For each three consecutive nodes on ωS ,the first,second,and third nodesbehave as the sender,relay,and receiver,respectively,i.e.,the first node sends its data to the third node with the help of the second node as discussed in the cooperative transmission mode.Fig.2.Required power per route versus the desired throughput for N =20nodes,N 0=−70dBm,and R d =2b/s/Hz in a 200×200grid.proportional to the distance to the power α.Since,both cases look similar with a shift in the transmitted power values,we will consider only α=4in the rest of this section as it is more appropriate to represent the wireless medium.It is shown that the SNCP algorithm,which applies the Bellman-Ford shortest-path algorithm,requires the most transmitted power per route.Applying the cooperative communication mode on each three consecutive nodes in the SNCP route results in reduction in the required transmitted power as shown in the CASNCP algorithm’s curve.Moreover,the MPCR algorithm requires the least transmitted power among the other routing algorithms.One of the major results of this paper is that the MPCR algorithm requires less transmitted power than the CASNCP algorithm.Intuitively,this result is because the MPCR ap-plies the cooperation-based link cost formula to construct the minimum-power route.On the contrary,the CASNCP algorithm first constructs shortest-path route then it applies the cooperative communication protocol on the established route.Therefore,the CASNCP algorithm is limited to applying the cooperative-communication protocol on certain number of nodes,while the MPCR algorithm can consider any node in the network to be in the CT blocks,which constitute the route.Thus,the MPCR algorithm reduces the required transmitted power more than the CASNCP algorithm.Fig.3depicts the required transmitted power per route by the different routing algorithms for different number of nodes at p S o =0.95and ηo =1.9b/s/Hz.As shown,the required transmitted power by any routing algorithm decreases with theFig.3.Required transmitted power per route versus the number of nodes for ηo =1.9b/s/Hz and α=4in a 200×200grid.Fig.4.Power savings per route versus the number of nodes for ηo =1.9b/s/Hz and α=4in a 200×200grid.number of nodes.Intuitively,the higher the number of nodes in a fixed area,the closer the nodes to each other,the lower the required transmitted power between these nodes,which results in lower required end-to-end transmitted power.We also calculate the power saving ratio as a measure of the improvement of the MPCR algorithm.The power saving of scheme 2with respect to scheme 1is defined asP ower Saving =P T (Scheme 1)−P T (Scheme 2)P T (Scheme 1)%,(23)where P T (.)denotes the total transmitted power for certain scheme.Fig.4depicts the power saving of the different routing algorithms with respect to each other.The shown curves are obtained through direct substitutions of the required transmitted power by each algorithm in (23).At N =100nodes,p S o =0.95,and ηo =1.9b/s/Hz,the power savings of MPCR algorithm with respect to the SNCP and CASNCP algorithms are 57.36%and 37.64%,respectively.In addition,the power saving of the CASNCP algorithm with respect to the SNCP algorithm is 31.62%.Fig.5depicts the required transmitted power per route of the different routing algorithms with respect to the desired bandwidth efficiency for N =20and N =100nodes.As mentioned with respect to Fig.2,the proposed MPCR algorithm requires the least transmitted power per route.In addition,we calculate the power saving of the MPCR algo-rithm as in (23).At R o =6b/s/Hz and N =100nodes,the MPCR algorithm reduces the transmitted power by 50.22%Fig.5.Required power per route versus the desired bandwidth efficiency for N0=−70dBm,p Sd=0.95b/s/Hz,andα=4in a200×200grid. and41.79%with respect to the SNCP and the CASNCP algorithms,respectively.In Fig.6,the average number of hops in each route, constructed by the different routing algorithms,is shown versus the number of nodes in the network.For the cooperative transmission mode,the average number of hops is defined ash C=1·Pr(φ)+2·Pr(φ)=2−Pr(φ),(24) and the average number of hops for the direct transmission mode is one.As shown,the routes constructed by either the CASNCP or the MPCR algorithms consist of number of hops that is less than the routes constructed by the SNCP algorithm. Moreover,the average number of hops increases with N as there are more available nodes in the network,which can be employed to reduce the transmitted power.In this section,we have illustrated using the numerical results the power savings of our proposed MPCR algorithm with respect to the SNCP and CASNCP algorithms.V.C ONCLUSIONSIn this paper,we have investigated the impacts of the co-operative communications on the minimum-power routing problem in wireless networks.For a given source-destination pair,the optimum route requires the minimum end-to-end transmitted power while guaranteeing certain throughput.We have proposed the MPCR algorithm,which applies the co-operative communication while constructing the route.The MPCR algorithm constructs the minimum-power route us-ing any number of the proposed cooperation-based building blocks,which require the least possible transmitted power. For comparison issues,we have also presented the CASNCP algorithm,which is similar to most of the existing cooperative routing algorithms.The CASNCP algorithmfirst constructs the conventional shortest-path route then applies a cooperative-communication protocol upon the established route.From the simulation results,the power savings of the MPCR algorithm with respect to the shortest-path and CASNCP algorithms are 57.36%and37.64%,respectively.R EFERENCES[1]L.M.Feeney and M.Nilsson,“Investigating the energy consumption ofa wireless network interface in an ad hoc networking environment,”inProc.of IEEE INFOCOM,Anchorage,AK,Apr.2001.Fig.6.Average number of hops per route versus the number of nodes for ηo=1.9b/s/Hz andα=4in a200×200grid.[2]T.M.Cover and A.El Gamal,“Capacity theorems for the relay channel,”IEEE .Theory,vol.25,no.5,pp.572-584,Sept.1979. [3]neman,D.N.C.Tse,and G.W.Wornell,“Cooperative diversityin wireless networks:efficient protocols and outage behaviour,”IEEE rm.Theory,vol.50,pp.3062-3080,Dec.2004.[4]neman and G.W.Wornell,“Distributed space-time coded pro-tocols for exploiting cooperative diversity in wireless networks,”IEEE rm.Theory,vol.49,pp.2415-2525,Oct.2003.[5] A.Sendonaris,E.Erkip,and B.Aazhang,“User cooperation diversity-Part I:system description,”IEEE m.,vol.51,pp.1927-1938, Nov.2003.[6] A.Sendonaris,E.Erkip,and B.Aazhang,“User cooperation diversity-Part II:implementation aspects and performance analysis,”IEEE Trans.Comm.,vol.51,pp.1939-1948,Nov.2003.[7] A.S.Ibrahim,A.K.Sadek,W.Su,and K.J.Ray Liu,“Cooperative com-munications with partial channel state information:when to cooperate?,”in Proc.of IEEE Global Telecommunications Conference(Globecom’05), pp.3068-3072,vol.5,Dallas,TX,28Nov.-2Dec.2005.[8] A.S.Ibrahim,A.K.Sadek,W.Su,and K.J.Ray Liu,“Relay Selec-tion in multi-node cooperative communications:When to cooperate and whom to cooperate with?,”in Proc.of IEEE Global Telecommunications Conference(Globecom’06),San Francisco,CA,Nov.2006.[9] A.K.Sadek,W.Su,and K.J.Ray Liu,“A class of cooperativecommunication protocols for multi-node wireless networks”,in Proc.of IEEE International Workshop on Signal Processing Advances in Wireless Communications(SPAWC’05),pp.560-564,New York,NJ,June2005.[10] A.K.Sadek,Z.Han,and K.J.Ray Liu,“A distributed relay-assignmentalgorithm for cooperative communications in wireless networks”,in Proc.of IEEE International Conference on Communications,Istanbul,Turkey, June2006.[11] A.E.Khandani,E.Modiano,L.Zheng,and J.Abounadi,“Cooperativerouting in wireless networks,”Chapter in Advances in Pervasive Comput-ing and Networking,Kluwer Academic Publishers,Eds.B.K.Szymanski and B.Yener,2004.[12]Z.Yang,J.Liu,and A.Host-Madsen,“Cooperative routing and powerallocation in ad-hoc networks,”in Proc.of IEEE Global Telecommunica-tion Conference,Globecom,Dallas,TX,Nov.2005.[13] F.Li,K.Wu,and A.Lippman,“Energy-efficient cooperative routingin multi-hop wireless ad hoc networks,”in Proc.of IEEE International Performance,Computing,and Communications Conference,pp.215-222,Phoenix,AZ,Apr.2006.[14]M.Sikora,neman,M.Haenggi,D.J.Costello,and T.E.Fuja,“Bandwidth-and power-efficient routing in linear wireless networks,”IEEE rm.Theory,vol.52,pp.2624-2633,Jun.2006.[15]J.Luo,R.S.Blum,L.J.Greenstein,L.J.Cimini,and A.M.Haimovich,“New approaches for cooperative use of multiple antennas in ad hoc wire-less networks,”in Proc.of IEEE60th Vehicular Technology Conference, vol.4,pp.2769-2773,Los Angels,CA,Sept.2004.[16] D.Bertsekas and R.Gallager,Data networks,2nd ed.,Prentice Hall,1991.[17]L.Zheng and D.N.C.Tse,“Diversity and multiplexing:a fundamentaltradeoff in multiple-antenna channels,”IEEE Trans.on Info.Theory,vol.49,pp.1073-1096,May2003.[18]J.G.Proakis,Digital communications,4th ed.,McGraw-Hill,2000.[19] D.G.Brennan,“Linear diversity combining techniques,”Proceedingsof the IEEE,vol.91,no.2,pp.331-356,Feb.2003.。
Ad hoc路由算法(文献综述)
Ad-hoc路由算法1 前言为满足信息社会对资源共享及信息传递的需求，计算机网络技术和无线通讯技术在近几十年得到了极大的发展。
20世纪50年代诞生的利用导线传输数据的有线网络经过几十年的发展，从双绞线、同轴电缆发展到如今的光纤通讯网络，网络的性能和覆盖范围虽然得到了很大的提升但是仍然无法满足人们在移动场景中对网络接入的需求。
在一些场合如抢险救灾、数字化战场、野外勘探及临时会议等场合无法快速高效的建立这常用无线网络，因此需要一种新型网络满足这些应用需求。
为满足在上述场合下快速高效组网的需求，无需基础设施的移动自组织网络（Mobile Ad-hoc Network）应用而生[2]。
Ad Hoc网络是一种不需要任何基站或固定基础设施的多跳无线网络，具有独立组网、自组织、动态拓扑、无约束移动、多跳路由等众多特点，能够快速地布设局部通信网络。
近年来，Ad Hoc网络研究得到了很大的发展，尤其是对网络路由协议的研究已经逐步成熟。
自二十世纪七十年代开始，由美国国防部所属的国防先进研究项目局推动了移动自组织网络方面最初的研究项目“战场环境中的数据包无线网络”（Packet Radio Networking），并在此后对多项相关研究进行支持。
最初由军方推动的移动自组织网主要应用在军事领域，然而随着微电子技术、嵌入式系统技术、无线通信技术等的发展和相关硬件成本的降低，移动自组织网络技术开始在民用领域推广[1]。
尤其在近十年，一些新技术与移动自组织网络的结合产生了许多新的研究热点如无线传感网络（Wirless Sensor Network）、车载自组网（Vehicular Ad hoc Network）、无线个人局域网（Wireless Personal Area Networks）以及无线Mesh网（Wireless Mesh Network）等[5]。
2 正文2.1、Ad-hoc网络的优点（1）无中心Ad hoc网络没有严格的控制中心。
基于蚁群算法的网络路由最优路径判断模块设计与实现
基于蚁群算法的网络路由最优路径判断模块设计与实现徐虹;杨雅志;赵明【摘要】网络中节点的能量是有限的，网络拓扑结构具有波动性，导致传统网络路由算法不能有效适应这些变化，自组织性较差，无法及时获取最优路径，大大降低网络性能。
因此，设计基于蚁群算法的网络路由最优路径判断模块。
其以FPGA 为控制核心实现硬件设计，具体包括控制模块、存储器模块、寻求后续节点集模块、采集后续节点模块、状态调整模块、信息素调整模块和最优路径判断模块。
模块实现部分给出了蚁群算法的核心代码。
实验结果表明，所设计的最优路径判断模块具有较高的收敛速率，获取的路径更短，能够延长网络的运行周期。
%Since energy in the network node is limited,and the network topology has volatility,which cause that the tradi⁃tional network routing algorithm can not effectively adapt to these changes,the self⁃organizing is poor,the optimal path can not be got timely,and the network performance is reduced greatly,the optimal path judgment module based on ant colony algorithm for network routing is designed. The FPGA as the control core is used to realize the hardware design,including the control module, memory module,subsequent nodes set seeking module,subsequent node acquisition module,state adjustment module,informa⁃tion adjustment module,optimal path judgment module and multiplex selection module. The core code of ant colony algorithm is presented in the process of module implementation. The experimental result shows that the designed optimal path judgment module has high⁃speed convergence and shorter access path,and can lengthen the operation cycle of the network.【期刊名称】《现代电子技术》【年(卷),期】2017(040)004【总页数】4页(P36-38,43)【关键词】蚁群算法;网络路由;最优路径;FPGA【作者】徐虹;杨雅志;赵明【作者单位】成都工业学院信息与计算科学系，四川成都 611730;成都工业学院信息与计算科学系，四川成都 611730;成都工业学院信息与计算科学系，四川成都 611730【正文语种】中文【中图分类】TN711-34;TP393无线传感器网络（WSN）通常是由传感器节点构成的自组织网络，在军事、医疗、工业等领域应用广泛。
《计算机网络》中英词汇对照表
《计算机网络》中英文对照表Chapter 11.1Internet：因特网Computer network ：计算机网络Host: 主机End system: 终端系统Packet switching: 分组交换Route: 路径Internet service provider (ISP): 因特网服务提供商Protocol: 协议Transmission Control Protocol (TCP)：传输控制协议1.2Client: 客户端Server: 服务器Peer: 对等机Reliable data transfer: 可靠数据传输Flow control: 流量控制Congestion-control: 拥塞控制User Datagram Protocol (UDP): 用户数据报协议1.3Circuit switching: 电路交换/线路交换Packet switching: 分组交换Frequency-division multiplexing (FDM): 频分多路复用Time-division multiplexing (TDM): 时分多路复用Bandwidth: 带宽Time slot: 时隙Frame: 帧Message: 报文:Packet: 分组Store-and-forward: 存储转发Datagram network: 数据报网络Virtual-circuit network: 虚电路网络1.4Router: 路由器Modem: 调制解调器Local area network (LAN): 局域网Ethernet: 以太网Wireless LAN: 无线局域网Guided media: 导向型介质Unguided media: 非导向型介质Twisted-pair copper wire: 双绞线Unshielded twisted pair(UTP): 非屏蔽双绞线Coaxial cable: 同轴电缆Fiber optics: 光线/光缆1.6Nodal processing delay: 结点处理延迟Queuing delay: 排队延迟Transmission delay: 发送延迟Propagation delay: 传播延迟Traffic intensity: 流通强度End-to-end delay: 端到端延迟1.7Layer: 层次Protocol stack: 协议栈Application layer: 应用层Transport layer: 传输层Network layer: 网络层Link layer: 链路层Physical layer: 物理层Encapsulation: 封装Message: 报文Segment: 报文段Datagram: 数据报Frame: 帧Chapter 22.1Client-server architecture: 客户端－服务器体系结构；C/S结构P2P architecture: 对等结构Processes: 进程Socket: 套接字Application programming interface (API): 应用程序编程接口IP address: IP地址Prot number: 端口号Syntax: 语法Semantics: 语义Full-duplex: 全双工Handshaking: 握手Real-time application: 实时应用2.2The World Wide Web: 万维网HyperText Transfer Protocol (HTTP): 超文本传输协议Web page: 网页Object: 对象HyperText Markup Language (HTML): 超文本标记语言URL：统一资源定位符Browser: 浏览器Persistent connection: 持久连接Non-persistent connection: 非持久连接Round-trip time (RTT): 往返时间Without pipelining: 非流水线方式With pipelining: 流水线方式Web cache: web 缓存Proxy server: 代理服务器2.3File Transfer Protocol (FTP): 文件传输协议Control connection: 控制连接Data connection: 数据连接Out-of-band: 带外In-band: 带内2.4Electronic Mail: 电子邮件User agent: 用户代理Mail server: 邮件服务器Simple Mail Transfer Protocol (SMTP): 简单邮件传输协议Mailbox: 邮箱Multipurpose Internet Mail Extensions (MIME): 多用途因特网邮件扩展协议Post Office Protocol (POP): 邮局协议Internet Mail Access Protocol (IMAP): Internet 邮件访问协议2.5Domain Name System (DNS): 域名系统Hostname: 主机名Host aliasing: 主机别名Mail server aliasing: 邮件服务器别名Load distribution: 负载分配Root DNS server: 根DNS服务器Top-Level Domain (TLD) servers: 顶级域DNS服务器Authoritative DNS servers: 授权DNS服务器；权威DNS服务器Local DNS server: 本地DNS服务器Database: 数据库Chapter 33.1Logical communication: 逻辑通讯3.2Multiplexing: 多路复用Demultiplexing: 多路分解Well-known port number: 众所周知的端口号3.3UDP segment: UDP报文段Checksum: 校验和；检查和Wrapped around: 回卷3.4Channel: 通道；信道Positive acknowledgement : 肯定应答Negative acknowledgement: 否定应答ARQ （automatic repeat request）: 自动重传请求Feedback: 反馈Retransmission: 重传Stop-and-wait protocol: 停止－等待协议Duplicate packets: 冗余分组Sequence number: 顺序号Timer: 定时器Alternating-bit protocol: 比特交替协议Utilization: 利用率Go-back-N (GBN): 回退N步Window size: 窗口大小Sliding-window protocol: 滑动窗口协议Cumulative acknowledgement: 累积确认Timeout: 超时Selective Repeat (SR): 选择重传3.5Connection-oriented: 面向连接Point-to-point: 点到点Three-way handshake: 三次握手Maximum segment size (MSS): 最大报文段大小Maximum transmission unit (MTU): 最大传输单元Piggybacked: 捎带Sample RTT: 样本RTTFast retransmit: 快速重传Selective acknowledgement: 选择确认Flow-control: 流量控制Receive window: 接收窗口3.7Congestion control: 拥塞窗口Self-clocking: 自定时的Additive-increase, multiplicative-decrease: 加性增，乘性减Slow star: 慢启动Congestion avoidance: 拥塞避免Threshold: 阈值Fast recovery: 快速恢复Bottleneck: 瓶颈Latency: 延迟Chapter 44.1Forwarding: 转发Routing: 路由Routing algorithm: 路由算法Forwarding table: 转发表Router: 路由器Jitter: 抖动Best-effort service: 尽力而为的服务4.2Virtual-circuit (VC) network: 虚电路网络Datagram network: 数据报网络Prefix: 前缀Longest prefix matching rule: 最长前缀匹配规则4.3Input port: 输入端口Switching fabric: 交换结构Routing processor: 路由处理器Crossbar: 交叉结构4.4Time-to-live (TTL) :生存时间Fragmentation: 分片；片段Dotted-decimal notation: 点分十进制表示法Subnet: 子网Subnet mask: 子网掩码Classless Interdomain Routing (CIDR): 无类别域际路由选择Dynamic Host Configuration Protocol（DHCP）：动态主机配置协议Plug-and-play: 即插即用Network address translation (NA T): 网络地址转换Internet Control Message Protocol (ICMP): 因特网控制报文协议Dual-stack: 双栈Tunneling: 隧道4.5Default router: 默认路由器Graph: 图A global routing algorithm : 全局路由算法A decentralized routing algorithm : 分布式路由算法Static routing algorithm: 静态路由算法Dynamic routing algorithm : 动态路由算法Link-State (LS): 链路状态Distance-Vector(DV): 距离向量Routing table: 路由表Autonomous system (AS): 自治系统Intra-autonomous system routing protocol: 自治系统内路由协议Inter-AS routing protocol: 自治系统间路由协议4.6Interior gateway protocol: 内部网关协议Routing Information Protocol (RIP): 路由信息协议Open Shortest Path First (OSPF): 开放最短路径优先协议Advertisement: 公告Hop: 跳Border Gateway Protocol (BGP): 边界网关协议4.7Broadcast: 广播Multicast: 多播Chapter 55.1Node: 结点Link: 链路Frame: 帧Medium access control (MAC): 介质访问控制Full-duplex: 全双工Half-duplex: 半双工Adapter: 适配器Network interface card (NIC): 网卡Interface: 接口5.2Parity check: 奇偶校验Odd: 奇数Even: 偶数Cyclic redundancy check (CRC): 循环冗余校验Polynomial: 多项式5.3Collide: 冲突Multiple access protocol: 多路访问协议Channel partitioning protocol: 信道划分协议Random access protocol: 随机访问协议Taking-turns protocol: 轮转协议Code division multiple access (CDMA): 码分多址访问Carrier sensing: 载波侦听Collision detection: 冲突检测Polling protocol: 轮询协议Token-passing protocol: 令牌传递协议Token: 令牌Local Area Network (LAN): 局域网Token-ring: 令牌环Fiber distributed data interface (FDDI): 光纤分布式数据接口Metropolitan Area Network (MAN): 城域网5.4Address Resolution Protocol (ARP): 地址解析协议Dynamic Host Configuration Protocol (DHCP): 动态主机配置协议5.5Ethernet: 以太网Preamble: 前导码Manchester encoding: 曼彻斯特编码5.6Hub: 集线器Collision domain: 冲突域Switch: 交换机Filtering: 过滤Forwarding: 转发Switch table: 交换表Self-learning: 自学习Plug-and-play devices: 即插即用设备Cut-through switching: 直通式交换5.7Point-to-point: (PPP): 点到点。
ERouting Final Exam CCNA Exploration路由协议和概念答案
ERouting Final Exam-CCNA Exploration:路由协议和概念(Version4.0)1.Which of the following are required when adding a network to the OSPF routing process configuration?network addressloopback addressa utonomous system numbersubnet maskwildcard maskarea ID2.Which of the following are primary functions of a router?(Choose two.)packet switchingmicrosegmentationdomain name resolutionpath selectionflow control3.Refer to the exhibit.When troubleshooting a network,it is important to interpret the output of various router commands.On the basis of the exhibit,which three statements are true?(Choose three.)The missing information for Blank1is the command show ip route.The missing information for Blank 1 is the command debug ip route.The missing information for Blank 2 is the number 100.The missing information for Blank2is the number120.The missing information for Blank 3 is the letter R.The missing information for Blank3is the letter C.4.Refer to the exhibit.Packets destined to which two networks will require the router to perform a recursive lookup?(Choose two.)10.0.0.0/864.100.0.0/16128.107.0.0/16172.16.40.0/24192.168.1.0/24192.168.2.0/245.When would the network administrator use the ip bandwidth-percent eigrp as-number percent command?when there is a low bandwidth connectionwhen the connection is on a shared mediumwhen the connection is serial instead of Ethernetwhen the link is always busy6.Refer to the exhibit.Cost for each path are shown.If all routers are configured to use OSPF,what would be the path of a packet sent from Router C to Router D if Router A was down?C-B-E-DC-B-A-D C-F-E-DC-F-B-A-D C-F-E-A-D7.What OSPF packet type is used to elect the designated router(DR)and backup designated router(BDR)on multiaccess networks?helloLSULSRDBDLSAck8.Refer to the exhibit.The hosts on the R1LAN are unable to access the Internet.What is incorrectly configured?the IP address of the Fa0/0 interface at R1the IP address of the S0/0/1 interface at R2the IP address of the S0/0/0interface at R1the subnet mask of the S0/0/1 interface at R29.Refer to the exhibit.Which summarization should R1use to advertise its networks to R2? 192.168.1.0/24192.168.0.0/24192.168.0.0/22192.168.1.0/2210.Refer to the exhibit.What are two of the routes added to the routing table of R1? (Choose two.)R 172.16.1.0/24 [120/1] via 192.168.3.0, 00:00:24, Serial0/0/0R192.168.1.0/24[120/1]via172.16.2.1,00:00:24,Serial0/0/1R 192.168.9.0/24 [120/1] via 172.16.1.2, 00:00:24, Serial0/0/0R192.168.100.0/24[120/1]via172.16.1.1,00:00:24,Serial0/0/0R 192.168.2.0/24 [120/1] via 172.16.1.2, 00:00:24, Serial0/0/011.A router boots and enters setup mode.What is the reason for this?The IOS image is corrupt.Cisco IOS is missing from flash memory.The configuration file is missing from NVRAM.The POST process has detected hardware failure.12.Refer to the exhibit.A router learns a route to the192.168.6.0network,as shown in the output of the show ip rip database command.However,upon running the show ip routecommand,the network administrator sees that the router has installed a different route to the192.168.6.0network learned via EIGRP.What could be the reason for the missing RIP route?Compared to RIP,EIGRP has a lower administrative distance.Compared to EIGRP, RIP has a higher metric value for the route.Compared to RIP, the EIGRP route has fewer hops.Compared to RIP, EIGRP has a faster update timer.13.All routers in a network are configured in a single OSPF area with the same priority value.No loopback interface has been set on any of the routers.Which secondary value will the routers use to determine the router ID?The highest MAC address among the active interfaces of the network will be used.There will be no router ID until a loopback interface is configured.The highest IP address among the active FastEthernet interfaces that are running OSPF will be used.The highest IP address among the active interfaces will be used.14.Refer to the exhibit.Routers R1and R2are directly connected via their serial interfaces and are both running the EIGRP routing protocol.R1and R2can ping the directly connected serial interface of their neighbor,but they cannot form an EIGRP neighbor adjacency.What action should be taken to solve this problem?Enable the serial interfaces of both routers.Configure EIGRP to send periodic updates.Configure the same hello interval between the routers. Configure both routers with the same EIGRP process ID.15.Refer to the exhibit.The interfaces of all routers are configured for OSPF area0.R3can ping R1,but the two routers are unable to establish a neighbor adjacency.What should the network administrator do to troubleshoot this problem?Check if the interfaces of the routers are enabled.Check the hello and dead intervals between the routers.Check the process ID of both routers.Check if CDP is enabled on all the routers.16.Refer to the exhibit.The hosts that are connected to R2are unable to ping the hosts that are connected to R1.How can this problem be resolved?Configure the router ID on both routers.Configure the R2router interfaces for area0.Configure a loopback interface on both routers.Configure the proper subnet masks on the router interfaces.17.Refer to the exhibit.The command ip route0.0.0.00.0.0.0S0/0/0is run on router R2. What are the two results of this command?(Choose two.)A static route will be updated in the routing table.The traffic from the Internet will be directed to R2.The traffic from the source network 172.16.0.0/22 will be blocked.The route will be specified as the default route for all networks not defined in the routing table.All the broadcasts will be forwarded via the S0/0/0 interface of R2.18.Refer to the exhibit.All routers are properly configured with default configurations and are running the OSPF routing protocol.The network is fully converged.A host on the 192.168.3.0/24network is communicating with a host on the192.168.2.0/24network.Which path will be used to transmit the data?The data will be transmitted via R3-R2.The data will be transmitted via R3-R1-R2.The traffic will be load-balanced between two paths — one via R3-R2, and the other via R3-R1-R2.The data will be transmitted via R3-R2, and the other path via R3-R1-R2 will be retained as the backup path.19.Refer to the exhibit.What is the meaning of the highlighted value120?It is the metric that is calculated by the routing protocol.It is the value that is used by the DUAL algorithm to determine the bandwidth for the link.It is the administrative distance of the routing protocol.It is the hold-down time, measured in seconds, before the next update.20.In a complex lab test environment,a router has discovered four paths to192.168.1.0/24 via the use of the RIP routing process.Which route will be installed in the routing table after the discovery of all four paths?R 192.168.1.0/24 [120/3] via 192.168.110.1, 00:00:17, Serial0/1/0R 192.168.1.0/24 [120/2] via 192.168.200.1, 00:00:17, Serial0/0/0R192.168.1.0/24[120/1]via192.168.100.1,00:00:17,Serial0/0/1R 192.168.1.0/24 [120/4] via 192.168.101.1, 00:00:17, Serial0/1/121.Refer to the exhibit.PC1is unable to access the Internet.What is the cause of the problem?An incorrect IP address is configured between the two routers. No static route is configured on Router2.A routing loop has occurred.No routing protocol is configured on either of the two routers.22.How does route poisoning prevent routing loops?New routing updates are ignored until the network has converged.Failed routes are advertised with a metric of infinity.A route is marked as unavailable when its Time to Live is exceeded.The unreachable route is cleared from the routing table after the invalid timer expires.23.Which statement is true about the metrics used by routing protocols?A metric is a value used by a particular routing protocol to compare paths to remote networks.A common metric is used by all routing protocols.The metric with the highest value is installed in the routing table.The router may use only one parameter at a time to calculate the metric.24.Which statement correctly describes a feature of RIP?RIP is a link-state routing protocol.RIP uses only one metric—hop count—for path selection.Advertised routes with hop counts greater than 10 are unreachable.Messages are broadcast every 10 seconds.25.Refer to the exhibit.OSPF is used for the routing protocol and all interfaces are configured with the correct IP addresses and subnet masks.During testing,it is found that router R1is unable to form an adjacency with R2.What is the cause of this problem?Both routers have been configured with incorrect router IDs.Both routers have been configured in different OSPF areas.Both routers have been configured with an incorrect network type.Both routers have been configured with different hello and dead intervals.26.A network administrator is in charge of two separate networks that share a single building.What device will be required to connect the two networks and add a common connection to the Internet that can be shared?hubrouteraccess pointEthernet switch27.Which network and mask combination requires the use of a classless addressing solution?10.32.0.0/11172.16.0.0/12192.168.0.0/24192.168.128.32/2728.A company is using static routes that are configured with an administrative distance of “1”on all routers in the network.The network administrator decides to introduce a dynamic routing protocol to reduce the manual configurations for the static routes.Which optionidentifies the correct procedure for the dynamic routing to take place in the network?The static routes and the dynamic routes will have the traffic alternate between them.The static routes will be automatically removed once the dynamic routing is configured.The static routes will be automatically updated with the next hop IP address once the dynamic routing is configured.The static routes must be manually removed from all routers in order for the dynamic routes to be installed in the routing table.29.Refer to the exhibit.Based on the partial output in the exhibit,why can users establish a console connection to this router without entering a password?The login command was not entered on the console line.The enable password should be an enable secret password.No username and password combination has been configured.Console connections cannot be configured to require users to provide passwords.30.Refer to the exhibit.When a static IP address is being configured on the host,what address should be used for the default gateway?10.1.1.110.1.1.2172.16.1.1192.168.1.131.Refer to the exhibit.The entire192.168.1.0network has been allocated to address hosts in the diagram.Utilizing VLSM with contiguous address blocks,which set of addresses andprefixes could be used to create an addressing solution with a minimum waste of IP addresses?Correct answer is image4.32.Refer to the exhibit.The network is configured for OSPF routing with default settings. The bandwidths have been configured correctly for each link.If the T1link between router A and router E fails,what path will a packet from router A take to reach the LAN attached to router F when the network has converged?A, B, C, FA, B, C, E, FA, D, G, E, FA,D,G,H,F33.Which candidate route has the longest match for a packet with a destination address of10.30.16.48?10.30.0.0/1610.30.15.0/2310.30.16.0/2410.30.16.32/2710.30.16.32/3034.Refer to the exhibit.The network is configured with RIPv2.However,network administrators notice that communication cannot be successfully completed from one LAN to another.A network administrator issues the show ip route command on the HQ router. Based on the output,what should be done to correct the problem?Disable the load balancing feature of RIPv2.Issue the no auto-summary command for RIPv2.Replace RIPv2 with EIGRP which supports VLSM.Make sure that the network statements include the correct subnet mask.35.Which multicast address does EIGRP use to send hello and updates packets?224.0.0.5224.0.0.6224.0.0.9224.0.0.1036.Refer to the exhibit.Why is the state of the serial0/0/0interface administratively down?An IP address has not been configured on the interface.The WIC was installed into the incorrect slot on the router.The default encapsulation on the interface has been modified.The no shutdown command has not been executed on the interface.37.Refer to the exhibit.How was the OSPF default gateway entry for R2determined? Default routes are automatically injected by OSPF into all advertisements.A static default gateway route is defined in the configuration of R2.The default-information originate command is applied on R1.The ISP defines the gateway of last resort and automatically passes it to R1 and R2.The ip default-gateway command is applied on R2.38.Refer to the exhibit.RIPv1has been properly configured on all routers in the network. However,users on LAN2have intermittent connectivity with the users on LAN1and LAN3. What is the cause of the problem?Both LAN networks are separated from router R2 with a variably subnetted Class C network 209.165.200.0/30.Neither router R1 nor router R3 has a static route configured that points to the variably subnetted 172.16.0.0/24 networks.Both routers R1and R3are sending the summarized172.16.0.0/16network to R2in their RIPv1routing updates.Both networks 172.16.1.0/24 and 172.16.100.0/24 are configured with a subnet mask different from the default classful mask.39.Which default EIGRP configuration must be modified to allow an EIGRP router to advertise subnets that are configured with VLSM?split horizonmetric K valuesautosummarizationhello and dead intervals40.What is a successor for a destination network in an EIGRP network?the next hop on the primary route with the largest feasible distance to the destinationthe next hop on the primary route with the smallest feasible distance to the destination41.Refer to the exhibit.Which route will be removed from the routing table if manual EIGRP summarization is disabled on the Serial0/0/0interface of Router3?0.0.0.0/0172.16.0.0/16172.16.1.0/24172.16.3.0/3042.Which port can be used for initial router configuration?AUXvty 0s0/0/0console43.Which two link-state routing protocol challenges does OSPF resolve through the election of a DR?(Choose two.)the extensive flooding of LSAs throughout the OSPF areathe excessive adjacencies when the number of routers increasesthe requirement for link-state database updates to be propagated between OSPF areasthe heavy CPU load that is imposed because each router must compute shortest paths by using the SPF algorithmthe requirement for each router to build a topological database of the internetwork to determinet he shortest paths between networks44.A routing table shows an EIGRP route to192.168.1.0/24with a metric of301440.What other term also describes this EIGRP metric value?feasible distancereported distancefeasible successorfeasibility condition45.Refer to the exhibit.The network administrator has run the following command on R1.R1(config)#ip route192.168.2.0255.255.255.0172.16.1.2What is the result of running this command?Traffic for network192.168.2.0is forwarded to172.16.1.2.This route is automatically propagated throughout the entire network.Traffic for all networks is forwarded to 172.16.1.2.The command invokes a dynamic routing protocol for 192.168.2.0.46.Refer to the exhibit.What will happen if interface Serial0/0/1goes down on Router1? The Dijkstra algorithm will calculate the feasible successor.DUAL will query neighbors for a route to network192.168.1.0.Neighbor 172.16.3.2 will be promoted to the feasible successor.Traffic destined to the 192.168.1.0 network will be dropped immediately due to lack of a feasible successor.47.Refer to the exhibit.A network administrator is accessing router R1from the console port.Once the administrator is connected to the router,which password should the administrator enter at the R1>prompt to access the privileged EXEC mode?Cisco001Cisco123Cisco789Cisco90148.Refer to the exhibit.Which option will provide the configuration that is needed for router R1to dynamically learn routes to the192.168.100.16/28,192.168.100.32/28,and 192.168.100.48/28subnetworks?with static routeswith a routed protocolwith a routing protocolwith directly connected routes49.Refer to the exhibit.What will happen when the router reloads?It will boot into ROMMON mode.It will ignore the start-up configuration file.It will look for the start-up configuration file on the TFTP server.It will attempt to load the start-up configuration file that is stored in NVRAM.50.On a router,which actions can be performed in user mode?perform password recoverymake global configuration changesview status of various router functionsmake changes to a specified interface。
2022年初级网络管理员考试题库(完整版)
2022年初级网络管理员考试题库（完整版）单选题1.在HTML语言中，（）可用来为图像定义一串预备的可替换的文本。
A、AltB、ReplaceC、TextD、Title答案：A解析：在HTML语言中，可以通过＜img>标签中的Alt属性为不能看到HTML文档中图像的浏览者提供文字说明。
2.在TCP/IP协议体系结构中，不可靠的传输层协议为（）。
A、UDPB、TCPC、ICMPD、SMTP答案：A解析：在TCP/IP协议栈中传输层有TCP协议和UDP协议2种，UDP协议是不可靠的协议。
3.下面关于HTTPS的描述中，错误的是（）。
A、HTTPS是安全的超文本传输协议B、HTTPS是HTTP和SSL/TLS的组合C、HTTPS和SHTTP是同一个协议的不同简称D、HTTPS服务器端使用的缺省TCP端口是443答案：C解析：超文本传输安全协议（HypertextTransferProtocolSecure，HTTPS）是超文本传输协议和SSL/TLS的组合，用以提供加密通讯及对网络服务器身份的鉴定。
HTTPS连接经常被用于万维网上的交易支付和企业信息系统中敏感信息的传输, HTTPS服务器端使用默认的TCP443端口。
HTTPS不应与在RFC2660中定义的安全超文本传输协议（S-HTTP）相混。
而SHTP则是HTTP协议的的扩展，目的是保证商业贸易的传输安全，只工作在应用层，仅限于web应用，因此并未获得广泛使用。
4.BGPrunsoverareliabletransport（）.Thiseliminatestheneedtoimplemente xplicitupdatefragmentation,retransmission,acknowledgement,and（请作答此空）.Anyauthenticationschemeusedbythetransportprotocolmaybeusedinad ditiontoBGP'sown（）mechanisms.TheerrornotificationmechanismusedinBGP （）thatthetransportprotocolsupportsa‚graceful‛close,i.e.,thatallou tstandingdatawillbedelivered（）theconnectionisclosed.A、synchronizationB、conflictC、transportD、sequencing答案：D解析：bgp通过可靠的传输协议运行。
强化学习框架下移动自组织网络分步路由算法
第４２卷第４期国 防 科 技 大 学 学 报Ｖｏｌ．４２Ｎｏ．４２０２０年８月ＪＯＵＲＮＡＬＯＦＮＡＴＩＯＮＡＬＵＮＩＶＥＲＳＩＴＹＯＦＤＥＦＥＮＳＥＴＥＣＨＮＯＬＯＧＹＡｕｇ．２０２０ｄｏｉ：１０．１１８８７／ｊ．ｃｎ．２０２００４００１ｈｔｔｐ：／／ｊｏｕｒｎａｌ．ｎｕｄｔ．ｅｄｕ．ｃｎ强化学习框架下移动自组织网络分步路由算法蒯振然，王少尉（南京大学电子科学与工程学院，江苏南京 ２１００２３）摘 要：移动自组织网络是一种无基础设施、由移动通信节点组成的无线网络，具有高动态特性。
传统的路由协议并不能适应节点移动性带来的频繁拓扑变化，简单的洪泛路由也会因开销过大降低网络的性能。
针对如何在移动自组织网络中自适应地进行路由选择，提出强化学习框架下的分步路由选择算法。
该算法以最小链路总往返时延为目标，基于强化学习进行路由搜寻，在筛选出符合目标需求节点集合的基础上，结合置信度选择路由。
在链路变得不可靠时，数据包被广播给筛选出的邻居节点集以提升路由可靠性并降低开销。
对提出的算法在分组到达率和路由开销等主要性能指标进行数值仿真分析。
仿真结果表明，提出的分步路由算法相比于听语音 聊科研与作者互动基于强化学习的智能鲁棒路由，在降低开销的同时，保持着相当的吞吐率。
关键词：移动自组织网络；强化学习；路由算法中图分类号：ＴＮ９２ 文献标志码：Ａ 开放科学（资源服务）标识码（ＯＳＩＤ）：文章编号：１００１－２４８６（２０２０）０４－００１－０６ＳｔｅｐｗｉｓｅｒｏｕｔｉｎｇａｌｇｏｒｉｔｈｍｉｎｍｏｂｉｌｅａｄｈｏｃｎｅｔｗｏｒｋｕｎｄｅｒｒｅｉｎｆｏｒｃｅｍｅｎｔｌｅａｒｎｉｎｇｆｒａｍｅｗｏｒｋＫＵＡＩＺｈｅｎｒａｎ，ＷＡＮＧＳｈａｏｗｅｉ（ＳｃｈｏｏｌｏｆＥｌｅｃｔｒｏｎｉｃＳｃｉｅｎｃｅａｎｄＥｎｇｉｎｅｅｒｉｎｇ，ＮａｎｊｉｎｇＵｎｉｖｅｒｓｉｔｙ，Ｎａｎｊｉｎｇ２１００２３，Ｃｈｉｎａ）Ａｂｓｔｒａｃｔ：Ｍｏｂｉｌｅａｄｈｏｃｎｅｔｗｏｒｋｉｓａｃｏｍｍｕｎｉｃａｔｉｏｎｎｅｔｗｏｒｋｆｏｒｍｅｄｂｙｍｏｂｉｌｅｎｏｄｅｓｗｉｔｈｎｏｎ ｉｎｆｒａｓｔｒｕｃｔｕｒｅ，ｗｈｉｃｈｈａｓｈｉｇｈｌｙｄｙｎａｍｉｃｃｈａｒａｃｔｅｒｉｓｔｉｃｓ．Ｃｏｎｖｅｎｔｉｏｎａｌｒｏｕｔｉｎｇｐｒｏｔｏｃｏｌｓｃａｎｎｏｔａｄａｐｔｔｏｔｈｅｆｒｅｑｕｅｎｔｔｏｐｏｌｏｇｙｃｈａｎｇｅｓｂｒｏｕｇｈｔｂｙｎｏｄｅｍｏｂｉｌｉｔｙ，ａｎｄｔｈｅｆｌｏｏｄｉｎｇｒｏｕｔｉｎｇａｌｓｏｃａｕｓｅｓｔｈｅｎｅｔｗｏｒｋｐｅｒｆｏｒｍａｎｃｅｄｅｇｒａｄａｔｉｏｎｄｕｅｔｏｔｈｅｅｘｃｅｓｓｉｖｅｒｏｕｔｉｎｇｏｖｅｒｈｅａｄ．Ａｓｔｅｐｗｉｓｅｒｏｕｔｉｎｇａｌｇｏｒｉｔｈｍｂａｓｅｄｏｎｒｅｉｎｆｏｒｃｅｍｅｎｔｌｅａｒｎｉｎｇｗａｓｐｒｏｐｏｓｅｄｆｏｒａｄａｐｔｉｖｅｒｏｕｔｉｎｇｉｎｍｏｂｉｌｅａｄｈｏｃｎｅｔｗｏｒｋｓ．Ｔｈｉｓａｌｇｏｒｉｔｈｍａｉｍｓａｔｔｏｔａｌｒｏｕｎｄｔｒｉｐｔｉｍｅｍｉｎｉｍｉｚａｔｉｏｎａｎｄｕｓｅｓｔｈｅｒｅｉｎｆｏｒｃｅｍｅｎｔｌｅａｒｎｉｎｇａｌｇｏｒｉｔｈｍｔｏｓｅｌｅｃｔｔｈｅｎｅｘｔｈｏｐ．Ａｆｔｅｒｓｅｌｅｃｔｉｎｇｔｈｅｓｅｔｏｆｎｏｄｅｓｔｈａｔｍｅｅｔｔｈｅｒｅｑｕｉｒｅｍｅｎｔｓｏｆｔｈｅｔａｒｇｅｔ，ｉｔｃｏｍｂｉｎｅｓｔｈｅｃｏｎｆｉｄｅｎｃｅｐａｒａｍｅｔｅｒｓｔｏｓｅｌｅｃｔｔｈｅｒｏｕｔｅ．Ｗｈｅｎｔｈｅｌｉｎｋｂｅｃｏｍｅｓｕｎｒｅｌｉａｂｌｅ，ｐａｃｋｅｔｓａｒｅｂｒｏａｄｃａｓｔｅｄｔｏｆｉｌｔｅｒｅｄｎｅｉｇｈｂｏｒｎｏｄｅｓｔｏｉｍｐｒｏｖｅｔｈｅｒｅｌｉａｂｉｌｉｔｙａｎｄｒｅｄｕｃｅｔｈｅｒｏｕｔｉｎｇｏｖｅｒｈｅａｄ．Ｔｈｅｍａｉｎｐｒｏｐｅｒｔｙｉｎｄｉｃａｔｉｏｎｏｆｔｈｅｐｒｏｐｏｓｅｄａｌｇｏｒｉｔｈｍ，ｓｕｃｈａｓｔｈｒｏｕｇｈｐｕｔａｎｄｒｏｕｔｉｎｇｏｖｅｒｈｅａｄ，ｗｅｒｅａｎａｌｙｚｅｄｔｈｅｏｒｅｔｉｃａｌｌｙ．Ｔｈｅｓｉｍｕｌａｔｉｏｎｒｅｓｕｌｔｓｓｈｏｗｔｈａｔ，ｃｏｍｐａｒｅｄｗｉｔｈｔｈｅｒｅｉｎｆｏｒｃｅｍｅｎｔｌｅａｒｎｉｎｇｂａｓｅｄｓｍａｒｔｒｏｂｕｓｔｒｏｕｔｉｎｇ，ｔｈｅｐｒｏｐｏｓｅｄｒｏｕｔｉｎｇａｌｇｏｒｉｔｈｍｒｅｄｕｃｅｓｔｈｅｏｖｅｒｈｅａｄａｎｄｍａｉｎｔａｉｎｓａｃｏｍｐｅｔｉｔｉｖｅｔｈｒｏｕｇｈｐｕｔ．Ｋｅｙｗｏｒｄｓ：ｍｏｂｉｌｅａｄｈｏｃｎｅｔｗｏｒｋ；ｒｅｉｎｆｏｒｃｅｍｅｎｔｌｅａｒｎｉｎｇ；ｒｏｕｔｉｎｇａｌｇｏｒｉｔｈｍ 移动自组织网络（ＭｏｂｉｌｅＡｄｈｏｃＮＥＴｗｏｒｋ，ＭＡＮＥＴ）是一种无基础设施、由移动通信节点组成的无线网络，具有组网灵活、配置方便和抗毁性强等特点［１］。
计算机英语单词词
计算机英语单词词部门： xxx时间： xxx整理范文，仅供参考，可下载自行编辑计算机英语词汇<１）1． artificial intelligence 人工智能2． paper-tape reader 纸空阅读机3． optical computer 光学计算机4． neural network 神经网络5． instruction set 指令集6． parallel processing 平行处理7． difference engine 差分机8． versatile logical element 通用逻辑器件9． silicon substrate 硅基10．vacuum tube 真空管<电子管）11. the storage and handling of data 数据的存储与处理12．very large-scale integrated circuit 超大规模集成电路13． central processing unit 中央处理器14．personal computer 个人计算机15． analogue computer 模拟计算机16．digital computer 数字计算机17． general-purpose computer 通用计算机18． processor chip 处理器芯片19． operating instructions 操作指令20． input device 输入设备21． circuit board 电路板22． beta testing β测试23． thin-client computer 瘦客户机电脑24． cell phone 蜂窝电话<移动电话）25． digital video 数码摄像机，数码影视26． Pentium processor 奔腾处理器27． virtual screen 虚拟屏幕28． desktop computer specifications 台式计算机规格29． radio frequency 射频30． wearable computer 可佩带式计算机31． Windows Registry 视窗注册表32． swap file 交换文件33． TMP file 临时文件34． power plug 电源插头35． free disk space 可用磁盘空间36． Control Panel 控制面板37． Start Menu 开始菜单38． Add/Remove Programs option 添加∕删除程序选项1． information retrieval 信息检索2． voice recognition module 语音识别模块3． touch-sensitive region 触感区，触摸区4． address bus 地址总线5． flatbed scanner 平板扫描仪6． dot-matrix printer 点阵打印机<针式打印机）7． parallel connection 并行连接8． cathode ray tube 阴极射线管9． video game 电子游戏<港台亦称电玩）10． audio signal 音频信号11． operating system 操作系统12． LCD (liquid crystal display>液晶显示<器）b5E2RGbCAP13． inkjet printer 喷墨打印机14． data bus 数据总线15． serial connection 串行连接16． volatile memory 易失性存储器17． laser printer 激光打印机18． disk drive 磁盘驱动器19． BIOS (Basic Input Output Sys tem> 基本输入输出系统p1EanqFDPw20． video display 视频显示器21． ISA slot ISA总线槽22． configuration register 配置寄存器23． still camera 静物照相机24． token packet 令牌包25． expansion hub 扩展集线器26． USB<Universal Serial Bus）通用串行总线27． root hub 根集线器28． I/O device 输入输出设备29． control frame 控制帧30． PCI (Peripheral Component In terconnect> 外部设备互连DXDiTa9E3d31． video tape 录像带32． aspect ratio <电视、电影图像的）高宽比，纵横比33． CD-RW 可擦写光驱34． laser diode 激光二极管35． reflective layer反射层36． optical disk光盘37． high resolution高分辨率38． floppy disk 软盘1． data set 数据集2． pointing device 指点设备3． graphical user interface 图形化用户界面4． time-slice multitasking 分时多任务处理5． object-oriented programming 面向对象编程6． click on an icon 点击图标7． context switching 上下文转换8． distributed system 分布式系统9． pull-down lists of commands 命令的下拉列表10． simultaneous access 同时访问11． command-line interface 命令行界面12． multitasking environment 多任务化环境13． spreadsheet program 电子制表程序14．main memory 主存15． storage media 存储介质16． disk file 磁盘文件17． command interpreter 命令解释器18． network connection 网络连接19．DOS (disk operating system> 磁盘操作系统20． copy a data file 拷贝数据文件21． serial port 串行端口22． configuration utility 配置工具23． ISDN 综合业务数字网24． token ring 令牌环25． fast Ethernet 快速以太网26． virtual memory 虚拟内存27． source code 源代码28． swap space 交换空间29． Internet protocol 因特网协议30．SVGA (Super Video Graphics Array> 超级视频图形阵列31． network throughput 网络吞吐量32． registry access 注册表存取33． scalable file server 规模可变的文件服务34． static Web page 静态网页35． physical memory 物理内存36． Plug and Play 即插即用37． network adapter 网络适配器38．SMP (symmetric multiprocessing> 对称多任务处理1．storage register 存储寄存器2．function statement 函数语句3．program statement 程序语句4．object-oriented language 面向对象语言5．assembly language 汇编语言6．intermediate language 中间语言，中级语言7．relational language 关系<型）语言8．artificial language 人造语言9．data declaration 数据声明10． SQL 结构化查询语言11． executable program 可执行程序12． program module 程序模块13．conditional statement 条件语句14． assignment statemen t赋值语句15．logic language 逻辑语言16． machine language 机器语言17．procedural language 过程语言18． programming language 程序设计语言19． run a computer program 运行计算机程序20． computer programme r 计算机程序设计员1．function call 函数调用2．event-driven programming 事件驱动编程3．click on a push button 点击按钮4．application window 应用程序窗口5．class hierarchy 类继承6．child window 子窗口7．application development environment 应用程序开发环境8．pull-down menu 下拉菜单9．dialog box 对话框10．scroll bar 滚动条1．native code 本机代码2．header file 头文件3．multithreaded program 多线程编程4．Java-enabled browser 支持Java的浏览器5．machine code 机器码6．assembly code 汇编码7．Trojan horse 特洛伊木马程序8．software package 软件包1．inference engine 推理机2．system call 系统调用3．compiled language 编译语言4．parallel computing 平行计算5．pattern matching 模式匹配6．free memory 空闲内存7．interpreter program 解释程序8．library routine 库程序9．intermediate program 中间程序，过渡程序10．source file 源文件11．interpreted language 解释<性）语言12．device driver 设备驱动程序13．source program 源程序14．debugging program 调试程序15．object code 目标代码16．application program 应用程序17．utility program 实用程序18．logic program 逻辑程序19． ink cartridge 墨盒20．program storage and execution 程序的存储与执行1．Windows socket Windows套接字接口2．Winsock interface Winsock接口3．file repository 文件属性4．client-side application 客户端应用程序5．HTML tag HTML标记6．Web browser 万维网浏览器7．hardware platform 硬件平台8．custom control 定制控件9．OLE (object linking and embedding> 对象链接和嵌入10．WAN (wide area network> 广域网1．search path 搜索路径2．dynamic library 动态链接库3．code set 代码集4．ancestor menu 祖辈菜单5．end user 最终用户6．menu item 菜单项7．cross-platform application 跨平台应用程序8．character set 字符集1．procedure call 过程调用2．structured message protocol 结构化消息协议3．secure protocol 安全协议4．networking protocol 网络协议5．processing node 处理节点6．homogeneous system 同构系统7．cost effectiveness 成本效益8．message encryption 信息加密<术）9．message format 信息格式10．component code 组件编码11．sequential program 顺序程序12．multicast protocol 多址通信协议13．routing algorithm 路由算法14．open system 开放式系统15．heterogeneous environment 异构型环境16．distributed processing 分布式处理17．resource sharing 资源共享18．structured message passing 结构化信息传送19．communication(s> link 通信链路20．development tool 开发工具1．logical entity 逻辑实体2．client-server architecture 客户机-服务器结构3．CPU cycle CPU周期4．graphics acceleration 图形加速5．software licensing 软件许可6．word-processing application 字处理应用程序7．load balancing 负载平衡8．remote procedure call 远程过程调用9．hardware configuration 硬件配置10．peer-to-peer network 对等网络1．font server 字体服务器2．data management logic 数据管理逻辑规则3．disk space 磁盘空间4．conceptual model 概念模型5．client-server model 客户–服务器模型6．graphics display 图形显示7．general-purpose hardware 通用硬件8．system expandability 系统可扩展性(3>RTCrpUDGiT1． language precompiler 程序语言预编译器2． business logic implementation 业务逻辑实现3． query processor 查询处理器4． data modeling 数据建模5． storage engine 存储引擎6． tiered architecture 分层结构7． database manager 数据库管理员8． data presentation layer 数据表现层9． logical database design 逻辑上的数据库设计10． entity relationship diagram 实体关系图11． query language 查询语言12． host language 主机语言13．Data Modification Language (DML> 数据修改语言14． data redundancy 数据冗余15． relational database 关系数据库16． relational data model 关系数据模型17． database management system (DBMS> 数据库管理系统18．data element 数据元素19． data access 数据存取20． query optimization 查询优化1． global temporary table 全局临时表2． partitioned data 分区的数据3． virtual table 虚拟<临时）表4． permanent table 永久<固定）表5． log out of a system 退出登录的系统6． primary key 主键7． foreign key 外键8． database object 数据库对象9． clustered index 簇索引10． local temporary table 本地临时表1． data module 数据模块2． object repository 对象库3． local database 本地<机）数据库4． client dataset 客户端数据集5． remote database server 远程数据库服务器6． flat file 平面文件7． data source 数据源8． Distributed Component Object Model (DCOM> 分布式组件对象模型5PCzVD7HxA1． microwave radio 微波无线电2． digital television 数字电视3． DSL 数字用户线路4． analog transmission 模拟传输5． on-screen pointer 屏幕<触摸屏）上的指示<器）6． computer terminal 计算机终端7． radio telephone 无线电话8． cellular telephone 蜂窝电话<移动电话）9． decentralized network 分散的网络10． wire-based internal network 基于普通网线的内部网络11．fiber-optic cable 光缆12． fax machine 传真机13． wireless communications 无线通信14． point-to-point communications 点对点通信15． modulated electrical impulse 调制电脉冲16． communication(s> satellite 通信卫星17． telegraph key 电报电键17． transmission medium 传输媒体19． cordless telephone 无绳电话20．metal conductor 金属导体1． error recovery 错误恢复2． parity function 奇偶函数3． video on demand 视频点播4． collision detection 冲突检测5． protocol layering 协议层6． architectural model 体系结构模型7． packet switching 包交换8． enterprise network 企业网9． protocol suite 协议组commercial backbone 商用骨干网1． high-definition TV 高清晰度电视2． frame relay 帧中继3． data rate 数据传输率4． metropolitan area network 城域网5． set-top box 机顶盒6． multi-mode fiber 多模光纤7． protocol stack 协议堆栈8． VPI (virtual path identifier> 虚拟路径标识符1． coaxial cable 同轴电缆2． computer networking 计算机网络3． multiple-access network 多路访问网络4． management software 管理软件5． broadband connection 宽带连接6． confidential information 机密信息7． monolithic system 单片机系统8． star network 星型网络9． bus network 总线型网络10． ring network 环形网络11． network resources 网络资源12．public key system 公钥体制13．public telephone network 公用电话网14． data encryption system 数据加密系统15． information superhighway 信息高速公路16． information age 信息时代17． computer security 计算机安全18． data network 数据网19． data link 数据链路20． access protocol 存取协议1． switched internetwork 交换式内部网2． routing protocol 路由协议3． carrier sense 载波侦听4． spanning tree 生成树5． hierarchical network 分层网络6． dynamic routing 动态路由选择7． VLAN (virtual local area network> 虚拟局域网8． UNI (user network interface> 用户网络接口9． campus network 校园网10． modular model 模块模型1． diskless workstation 无盘工作站2． group scheduling 成组调度3． remote node 远程节点4． printer port 打印口5． remote access 远程访问6． DUN (Dial-Up Networking> 拨号联网7． parallel port 并行端口NOS (network operating system> 网络操作系统(4>jLBHrnAILg1． network layout 网络布局xHAQX7 4J0X2． physical topology 物理拓扑结构3． logical topology 逻辑拓扑结构4． star configuration 星型结构5． physical network connection 物理网络连接6． high-end active hub 高端主动式集线器7． passive hub 被动式集线器8． network node 网络节点9． electrical ground 电气接地10． data flow 数据流11．wiring closet 布线室12． multistation access unit 多站访问单元13．star topology 星形拓扑结构14．bus topology 总线拓扑结构15． ring topology 环形拓扑结构16． network topology 网络拓扑结构17． centralized network management 集中式网络管理18． intelligent hub 智能集线器19． network hub 网络集线器20．physical network 物理网络1． heterogeneous network 异构网络2． packet delivery 包发送3． IBM compatible IBM兼容的4． IP datagram IP数据报5． DOS box DOS箱<机）6． HTTP (Hypertext Transfer Protocol> 超文本传送协议7． NNTP (Network News Transfer Protocol> 网络新闻传送协议8． SMTP (Simple Mail Transfer Protocol> 简单邮件传送协议9． security hole 安全漏洞10． system crash 系统崩溃1． physical address 物理地址2． data transfer 数据迁移3． header checksum 报头校验4． stream delivery <数据）流发送5． virtual circuit 虚电路6． network layer 网络层7． full-duplex transmission 全双工传输ARP (Address Resolution Protocol> 地址解释协议1． list server 列表服务器2． transmission scheme 传输模式3． data packet 数据包4． Mbps 每秒兆字节5． hypermedia document 超媒体文档6． FTP 文件传输协议7． host network 主机网络8． dedicated access 专线访问9． storage format 存储格式10． mail server 邮件服务器11． multimedia file 多媒体文件12．dial-up access 拨号访问13． LAN (local area network> 局域网14． retrieve files 检索文件15． ISP (Internet Service Provider> 因特网服务供应商16． WWW (World Wide Web> 万维网17． URL (Uniform Resource Locator> 统一资源定位符18．TCP (Transmission Control Protocol> 传输控制协议19． data stream 数据流20． log on 登录1． plain text 纯文本2． destination address3． mail-user agent 邮件用户代理4． message transfer agent 消息传送代理5． graphics-based file6． analog signal 模拟信号LDAYtRyK fE7． domain name 域名8． text file 文本文件9． text editor 文本编辑器10． e-mail address 电子邮件地址1． sound card 声卡2． Web page 网页3． video camera 摄像机，摄像头4． plug-in software 嵌入软件5． input/output port 输入∕输出端口6． home page 主页7． video capture card 视频捕获卡8． chat room 聊天室1． electric motor 电动机2． desktop publishing 桌面出版系统<台式出版系统）3． information-related services 信息相关服务4． information-based occupation 基于信息的职业5． information processor 信息处理6． textual data 文本的数据Zzz6ZB 2Ltk7． numerical data 数字的数据8． audio data 音频数据9． fibre optics 纤维光学10．digital thermometer 数字温度计11．information revolution 信息革命12．technological revolution 技术革命13．global market 全球市场dvzfvkw MI114． IT (information technology> 信息技术15． multimedia product 多媒体产品16．information specialist 信息专家17．database management 数据库管理18．video data 视频数据19． information-processing system 信息处理系统20．telephone helpline 电话服务热线1． tabular data 表格数据2． raster image 光栅图像3． vector model 矢量模型4． statistical analysis system 统计分析系统5． model atmospheric circulation 模拟大气循环6． computer-based tool 基于计算机的工具7． geographic information system 地理信息系统8． database operation 数据库操作9． grid cell 网格单元10．closed loop 闭环1． domain-specific tag 特定<指定）域标记2． handheld terminal 手持终端设备3． life cycle 生命周期<生存周期）4． mobile agent toolkit 移动代理工具包5． XML (eXtensible Markup Language> 扩展标签语言6． data mining 数据挖掘7． game theory 博弈论8． keyword-based text search(ing> 基于关键字的搜索(5>rqyn14ZNXI1．user authentication 用户认证2．electronic purse 电子钱包3．information filter 信息过滤4．data integrity 数据完整性5．smart card 智能卡6．HTML 超文本标记语言7．symmetric key cryptosystem 对称密钥密码系统8．message authentication code 信息鉴定码9．unauthorized access control 未授权访问控制10．electronic catalog 电子目录11．electronic money (或cash> 电子货币12．search engine 搜索引擎13．digital signature 数字签名14．user interface 用户界面15． EFT (Electronic Funds Transfer> 电子资金转帐16．public key cryptosystem 公钥密码系统17．PDA (personal digital assistant> 个人数字助理18．hypertext link 超文本链接19．3D image 三维图像20．credit card 信用卡1．vendor-centric model 客户中心模式2．Web site 网站3．Web surfing 网上冲浪4．middleware server 中间件服务5．back-end platform 后端平台6．e-Business strategy 电子商务策略7．binary format 二进制格式8．customer-oriented e-Business system 面向客户的电子商务系统9．ISV (independent software vendor> 独立软件推销商10．information infrastructure 信息基础结构设施1．Web storefront 网上店面2．electronic press kit 电子版发行包3．online retail 在线零售4．multimedia demo 多媒体演示5．online access 联机访问6．value-added services 增值业务7．product promotion 产品推销8．communication medium 通信媒体1．encryption program 加密程序2．deletion command 删除命令3．authorized user 授权的用户。
Routing algorithm
Routing AlgorithmAbstractRouting algorithm can be distinguished by many features depending on the designer’s specific objectives. There are many kinds of routing algorithms with different affections on the network and router resources.The purpose of a routing algorithm is to define a set of rules for transferring units of data, known as packets, from one node to another.Key Wordshop path length least cost update time time delay Dijkstra’s algorithm Bellman-Ford algorithm. IntroductionRouting algorithm is to improve the function of routing protocol with the least overhead.In our book,it only talks about routing in switched networks including circuit-switching network and packet-switching network.In a circuit-switching network,to cope with the growing demands on public telecommunication networks,virtually all providers have moved away from the static hierarchical approach to a dynamic approach.In a packet-switching network ,the selection of a route is generally based on some performance criterion asfollows:1.to choose the minimum-hop route(least-cost routing)2.decision time and placework information source and update timeHence,a large number of routing stragies have evolved for dealing with the routing requirements of packet-switching networks including fixed routing,flooding,random routing and adaptive routing.The original routing algorithm designed in 1969 was a distributed adaptive algorithm ,which is a version of the Bell-Ford algorithm.After some years of experience the original routing algorithm was replaced by a quite difference using delay as the performance criterion.The third generation damp routing oscillations and reduce routing overhead. Routing algorithm should be flexible. The key technological factors are as follows:1. the shorest route(least hop or shorest path length) or the best route2.the communication subnet should adopt virtual circuit or datagram3.general routing algorithm or distributed routing algorithm4. cosider about the network topology,traffic and time delay5.static routing or dynamic routingThe most common routing algorithm is least-cost algorithm which is the variation of Dijkstra’s algorithm and the Bellman-Ford algorithm.System ModelExamples of adaptive-routing algorithms are the Routing Information Protocol (RIP) and the Open-Shortest-Path-First protocol (OSPF). Adaptive routing dominates the Internet. However, the configuration of the routing protocols often requires a skilled touch; networking technology has not developed to the point of the complete automation of routing. In P2P logical network, there is a simple ROP route which is responding to a complex RON route in communication network.The key point of this routing algorithm is how to pass the data to the destination fastly and reliably.。
基于交通时空特征的车辆全局路径规划算法
ISSN 1674-8484汽车安全与节能学报, 第12卷第1期, 2021年J Automotive Safety and Energy, Vol. 12 No. 1, 2021基于交通时空特征的车辆全局路径规划算法杜 茂，杨 林*，金 悦，涂家毓（上海交通大学机械与动力工程学院，上海 200240，中国）摘要：为降低混合动力汽车（HEV）的出行时间和出行能耗，提出了一种基于时空动态交通信息的路径规划算法。
分析了影响车辆通行时间和全程最低能耗的因素。
一种基于广义回归网络（GRNN）模型，拟合计算了道路通行时间以及整体路径的全程能耗。
构建了基于并行A*算法的车辆路径规划算法，为确定起终点位置后的车辆，规划了一条耗时更短、更加节能的路径。
进行了仿真对比试验。
结果表明：相比于依据平均车速与道路功率的计算方法，该算法能够获得更优的出行路径，可降低车辆能耗11%以上，缩短行车时间13%以上。
因而，该算法可为车辆规划更优的路径。
关键词：混合动力汽车(HEV)；城市交通；路径规划；时空搜索中图分类号： U 469.7 文献标识码： A DOI： 10.3969/j.issn.1674-8484.2021.01.005 Vehicle global path planning algorithm based on spatio-temporal characteristics of trafficDU Mao, YANG Lin*, JIN Yue, TU Jiayu(School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China) Abstract: A path planning algorithm based on spatiotemporal dynamic traffic information was proposed toreduce the travel time and energy consumption of hybrid electric vehicles (HEV). The factors that affect thevehicle travel time and the minimum energy consumption in the whole path were analyzed. The travel timeand the path energy consumption are calculated based on the generalized regression network (GRNN) model.A vehicle path planning algorithm based on parallel A * algorithm was constructed to plan a shorter time-consuming or more energy-saving path for vehicles after determining the starting and ending positions. Thevirtual simulation test was implemented. The results show that the proposed algorithm can obtain a better travel path and reduce vehicle energy consumption by more than 11% or driving time by more than 13% comparedwith the calculation methods through average speed or power parameters. Therefore, the proposed algorithm can plan a better route for vehicles.Key words: hybrid electric vehicle (HEV); urban traffic; path planning; space-time searching收稿日期 / Received ：2020-11-11。
2en
Below you will find the assessment items as presented on the exam as well as the scoring rules associated with the item.Cisco Networking Academy content is copyrighted and the unauthorized posting, distribution or sharing of this exam content is prohibited.Close WindowAssessment SystemExam Viewer - ERouting Practice Final Exam - CCNA Exploration: 路由协议和概念 (Version 4.0)1What are two functions of a router? (Choose two.) gf e d c It connects multiple IP networks. gf e d c It controls the flow of data via the use of Layer 2 addresses.g fe d c It determines the best path to send packets. g fe d c It manages the VLAN database. g fe d c It increases the size of the broadcast domain.ObservableDescriptionMax Value1correctness of responseOption 1 and Option 3 are correct.1 point for each correct option.0 points if more options are selected than required.22When a router boots, what is the default order to locate the Cisco IOS if there is no boot system command? nm l k j ROM, TFTP server, flash n ml k j flash, TFTP server, ROM n ml k j flash, NVRAM, TFTP server nm l k j NVRAM, TFTP server, flashObservableDescriptionMax Value1correctness of response2 points for Option 20 points for any other option23Which router component is used to store the routing table? nm l k j Flash n m l k j NVRAM n ml k j ROM n ml k j SDRAMObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option24Refer to the exhibit. How many routes are child routes? nm l k j 1 n m l k j 3 n ml k j 4 n ml k j 6ObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option25Refer to the exhibit. Which statement is true concerning the routing configuration?nm l k j Using dynamic routing instead of static routing would have required fewer configuration steps. n ml k j The 10.1.1.0/24 and 10.1.2.0/24 routes have adjacent boundaries and should be summarized. n ml k j Packets routed to the R2 Fast Ethernet interface require two routing table lookups. nm l k j The static route will not work correctly.ObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option26Refer to the exhibit. The network administrator issues the command no ip classless on Router1. What forward on a packet that is received by Router1 and is destined for host 192.168.0.26? nm l k j The packet will be dropped. n ml k j The packet will be forwarded to the gateway of last resort. n ml k j The packet will match the 192.168.0.0 network and be forwarded out Serial 0/0. n ml k j The packet will most closely match the 192.168.0.8 subnet and be forwarded out Serial 0/1.ObservableDescriptionMax Value1correctness of response2 points for Option 10 points for any other option27Refer to the exhibit. Routers R1 and R3 use different routing protocols with default administrative distance valu properly configured and the destination network is advertised by both protocols.Which path will be used to transmit the data packets between PC1 and PC2? nm l k j The packets will travel via R2-R1. n ml k j The packets will travel via R2-R3. nm l k j The traffic will be load-balanced between two paths — via R2-R1 and via R2-R3. n ml k j The packets will travel via R2-R3, and the other path via R2-R1 will be retained as the backup path.ObservableDescriptionMax Value1correctness of response2 points for Option 10 points for any other option28Refer to the exhibit. Router R1 is configured as shown in the exhibit. PC1 on 172.16.1.0/24 network can reach R1. The rest of the routers are configured with the correct IP addresses on the interfaces. Routers R2 and R3 d dynamic routing enabled. How far will PC1 be able to successfully ping? nml k j router R1 Fa0/0 interface n ml k j router R1 S0/0/0 interface n ml k j router R2 S0/0/0 interface n ml k j router R2 Fa0/0 and S0/0/1 interfaces n ml k j router R3 Fa0/0 and S0/0/0 interfacesObservableDescriptionMax Value1correctness of response2 points for Option 20 points for any other option29Refer to the exhibit. All routers are properly configured to use the EIGRP routing protocol with default settings, converged. Which statement correctly describes the path that the traffic will use from the 10.1.1.0/24 network to network?nm l k j It will use the A-D path only. n ml k j It will use the path A-D, and the paths A-C-D and A-B-D will be retained as the backup paths. It will use all the paths equally in a round-robin fashion.ml j The traffic will be load-balanced between A-B-D and A-C-D.ObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option210Which two statements are true regarding link-state routing protocols? (Choose two.)f e c They are aware of the complete network topology. fe c They offer rapid convergence times in large networks. fe c They do not include subnet masks in their routing updates. fe c They rely on decreasing hop counts to determine the best path. fe c They do not work well in networks that require special hierarchical designs. fe c They pass their entire routing tables to their directly connected neighbors only.ObservableDescriptionMax Value1correctness of responseOption 1 and Option 2 are correct.1 point for each correct option.0 points if more options are selected than required.211Refer to the exhibit. R1 knows two routes, Path A and Path B, to the Ethernet network attached to R3. R1 learn 10.2.0.0/16 from a static route and Path B to network 10.2.0.0/16 from EIGRP. Which route will R1 install in its nm l k j Both routes are installed and load balancing occurs across both paths. nm l k j The route via Path B is installed because the EIGRP route has the best metric to network 10.2.0.0/16. n ml k j The route via Path A is installed because the static route has the best metric to network 10.2.0.0/16. n ml k j The route via Path B is installed because the EIGRP route has the lowest administrative distance to networ n ml k j The route via Path A is installed because the static route has the lowest administrative distance to networkObservableDescriptionMax Value1correctness of response2 points for Option 50 points for any other option212What two routing protocols use a hierarchal network topology? (Choose two.)gf e d c IS-ISg f e d c EIGRP g f e d c OSPF g fe d c RIPv1 g fe d c RIPv2ObservableDescriptionMax Value1correctness of responseOption 1 and Option 3 are correct. 1 point for each correct option.0 points if more options are selected than required.213Refer to the exhibit. Based on the output from the show running-config and debug ip rip commands, what a are added to the routing table of R1? (Choose two.)gf e d c R 192.168.1.0/24 [120/1] via 172.16.2.1, 00:00:24, Serial0/0/1g fe d c R 192.168.100.0/24 [120/1] via 172.16.1.1, 00:00:24, Serial0/0/0 g fe d c S 192.168.1.0/24 [1/0] via FastEthernet0/0 gf e d c R 192.168.9.0/24 [120/1] via 172.16.2.1, 00:00:24, Serial0/0/0g fe d c R 192.168.2.0/24 [120/1] via 172.16.1.2, 00:00:24, Serial0/0/0ObservableDescriptionMax Value1correctness of responseOption 2 and Option 3 are correct.1 point for each correct option.0 points if more options are selected than required.214Refer to the exhibit. The network has three connected routers: R1, R2 and R3. The routes of all three routers a be verified from the output?m l j ml j The IP address of the S0/0/0 interface of R1 is 10.1.1.2. ml j The IP address of the S0/0/1 interface of R2 is 10.3.3.2. m l j R2 is connected to the S0/0/1 interface of R3.ObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option215Refer to the exhibit. All router interfaces are configured with an IP address and are operational. If no routing pro configured, what information will be included in the show ip route command output for router A? nm l k j All of the 192.168.x.0 networks will be in the routing table. n ml k j Routes to networks 192.168.1.0/24, 192.168.2.0/24, and 192.168.3.0/24 will be in the routing table. n ml k j The routing table will be empty because routes and dynamic routes have not been configured. n ml k j A default route is automatically installed in the routing table to allow connectivity between the networks.ObservableDescriptionMax Value1correctness of response2 points for Option 20 points for any other option216Refer to the exhibit. A network administrator is accessing router R1 from the console port. Once the administra router, which password should the administrator enter at the R1> prompt to access the privileged EXEC mode?nm l k j Cisco001 n ml k j Cisco123 n ml k j Cisco789 n ml k j Cisco901ObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option217Which of the following could describe the devices labeled "?" in the graphic? (Choose three.) gf ed c DCE g fe d c CSU/DSU gf e d c LAN switchg fe d c modem g fe d c hubObservableDescriptionMax Value1correctness of responseOption 1, Option 2, and Option 4 are correct.1 point for each correct option.0 points if more options are selected than required.318Refer to the exhibit. Which router is advertising subnet 172.16.1.32/28? nm l k j Router1 n m l k j Router2 n ml k j Router3 n ml k j Router4ObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option219Refer to the exhibit. The show cdp neighbors command was run at R1. Which two facts about the newly dete determined from the output? (Choose two.) gf e d c ABCD is a router that is connected to R1.f e c The device is connected at the Serial0/0/1 interface of R1.f e c R1 is connected at the S0/0/1 interface of device ABCD.f e c ABCD does not support switching capability.Observable Description Max Value1correctness of response Option 1 and Option 3 are correct.1 point for each correct option.0 points if more options are selected than required.220 A static route has been configured on a router. However, the destination network no longer exists. What shouldremove the static route from the routing table?m l j Change the routing metric for that route.m l j Nothing. The static route will go away on its own.m l j Change the administrative distance for that route.m l j Remove the route using the no ip route command.Observable Description Max Value1correctness of response 2 points for Option 40 points for any other option221Refer to the exhibit. A ping between host A and host B is successful, but pings from host A to operational hosts is the reason for this problem?n m l k j The FastEthernet interface of R1 is disabled.n m l k j One of the default routes is configured incorrectly.n m l k j A routing protocol is not configured on both routers.n m l k j The default gateway has not been configured on host A.Observable Description Max Value1correctness of response 2 points for Option 20 points for any other option222Refer to the exhibit. The network has three connected routers: R1, R2, and R3. The routes of all three routers a are operational and pings are not blocked on this network.Which ping will fail?n m l k j from R1 to 172.16.1.1n m l k j from R1 to 192.168.3.1n m l k j from R2 to 192.168.1.1n m l k j from R2 to 192.168.3.1Observable Description Max Value1correctness of response 2 points for Option 20 points for any other option223Refer to the exhibit. What action will R2 take for a packet that is destined for 192.168.2.0?n m l k j It will drop the packet.n m l k j It will forward the packet via the S0/0/0 interface.n m l k j It will forward the packet via the Fa0/0 interface.n m l k j It will forward the packet to R1.Observable Description Max Value1correctness of response2 points for Option 40 points for any other option224Refer to the exhibit. The users on the local network 172.16.1.0/24 complain that they are unable to connect to t should be taken to remedy the problem?nm l k j A new static route must be configured on R1 with the R3 serial interface as the next hop. n ml k j A new default route must be configured on R1 with the R3 serial interface as the next hop. nm l k j The default route on R2 should be configured with the R3 serial interface as the next hop. n ml k j The default route on R2 must be replaced with a new static route and the next hop should be the R1 FastEObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option225Refer to the exhibit. What summary address can Router2 advertise to Router1 to reach the three networks on R without advertising any public address space or overlapping the networks on Router1? nm l k j 172.16.0.0/8 nm l k j 172.16.0.0/10 n ml k j 172.16.0.0/13 n ml k j 172.16.0.0/20ObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option226Refer to the exhibit. Host A is unable to access the Internet, and troubleshooting has revealed that this is due t What is incorrectly configured in this network? nm l k j the IP address of the Fa0/0 interface of R1 n ml k j the subnet mask of the S0/0/0 interface of R1 n ml k j the IP address of the S0/0/0 interface of R1 nm l k j the subnet mask of the S0/0/0 interface of R2ObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option227Refer to the exhibit. A new PC was deployed in the Sales network. It was given the host address of 192.168.10gateway of 192.168.10.17. The PC is not communicating with the network properly. What is the cause? nm l k j The default gateway is incorrect. n ml k j The address is in the wrong subnet. nm l k j The host address and default gateway are swapped. n ml k j 192.168.10.31 is the broadcast address for this subnet.ObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option228Refer to the exhibit. The network administrator is planning IP addressing of a new network. What part of this ad be changed to allow communication between host A and the server?n m l k j the IP address of the servern m l k j the default gateway of host An m l k j the IP address of host An m l k j the default gateway of the serverObservable Description Max Value1correctness of response 2 points for Option 10 points for any other option229Which network design feature requires the deployment of a classless routing protocol?n m l k j private IP addressingn m l k j advertising default routesn m l k j variable length subnet masksn m l k j summarization on major network boundariesObservable Description Max Value1correctness of response 2 points for Option 30 points for any other option230 A network administrator needs to assign the very last usable IP address in the 172.24.64.0/18 network range toserves this LAN. Which IP address should the administrator configure on the interface?n m l k j172.16.128.154/18n m l k j172.16.255.254/18n m l k j172.24.64.254/18n m l k j172.24.127.254/18Observable Description Max Value1correctness of response 2 points for Option 40 points for any other option231Refer to the exhibit. All routers are running RIPv1. The two networks 10.1.1.0/29 and 10.1.1.16/29 are unable t What can be the cause of this problem?nm l k j Because RIPv1 is a classless protocol, it does not support this access. n ml k j RIPv1 does not support discontiguous networks. nm l k j RIPv1 does not support load balancing. n ml k j RIPv1 does not support automatic summarization.ObservableDescriptionMax Value1correctness of response2 points for Option 20 points for any other option232Refer to the exhibit. What information can be determined from the highlighted output? nm l k j R1 is originating the route 172.30.200.32/28. n ml k j Automatic summarization is disabled. nm l k j The 172.30.200.16/28 network is one hop away from R1. n ml k j A classful routing protocol is being used.ObservableDescriptionMax Value1correctness of response2 points for Option 20 points for any other option233What does RIP use to reduce convergence time in a larger network?ml j It reduces the update timer to 15 seconds if there are more than 10 routes. ml j It uses triggered updates to announce network changes if they happen in between the periodic updates. ml j It uses random pings to detect if a pathway is down and therefore is preemptive on finding networks that arObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option234A network administrator has enabled RIP on routersB andC in the network diagram. Which of the following co updates from being sent to Router A? nm l k j A(config)# router rip A(config-router)# passive-interface S0/0nm l k j B(config)# router rip B(config-router)# network 192.168.25.48 B(config-router)# network 192.168.25.64 nm l k j A(config)# router rip A(config-router)# no network 192.168.25.32nm l k j B(config)# router rip B(config-router)# passive-interface S0/0nm l k j A(config)# no router ripObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option235Refer to the exhibit. Both routers are using the RIP protocol. Devices on the 192.168.2.0 network can ping the cannot ping devices on the 192.168.1.0 network. What is a possible cause of this problem?n m l k j The routers are configured with different versions of RIP.n m l k j R2 is not forwarding the routing updates.n m l k j The R1 configuration should include the no auto-summary command.n m l k j The maximum path number has been exceeded.Observable Description Max Value1correctness of response 2 points for Option 10 points for any other option236Which two statements are correct about the split horizon with poison reverse method of routing loop preventiong f e d c It is enabled by default on all Cisco IOS implementations.g f e d c It assigns a value that represents an infinite metric to the poisoned route.g f e d c It sends back the poisoned route update to the same interface from where it was received.g f e d c It instructs routers to hold all changes that might affect routes, for a specified period of time.g f e d c It limits the number of hops a packet can traverse through the network before it is discarded.Observable Description Max Value1correctness of response Option 2 and Option 3 are correct.1 point for each correct option.0 points if more options are selected than required.237Refer to exhibit. Given the topology shown in the exhibit, what three commands are needed to configure EIGR (Choose three.)g f e d c Paris(config)# router eigrp 100g f e d c Paris(config)# router eigrpg f e d c Paris(config-router)# network 192.168.6.0g f e d c Paris(config-router)# network 192.168.7.0g f e d c Paris(config-router)# network 192.168.8.0g f e d c Paris(config-router)# network 192.168.9.0Observable Description Max Value1correctness of response Option 1, Option 4, and Option 5 are correct.1 point for each correct option.0 points if more options are selected than required.338 A router has EIGRP configured as the only routing protocol. In what way might EIGRP respond if there is no fea destination network and the successor route fails?n m l k j It broadcasts hello packets to all routers in the network to re-establish neighbor adjacencies.n m l k j It sends queries to adjacent neighbors until a new successor route is found.n m l k j It immediately sends its entire routing table to its neighbors.n m l k j It will set the metric for the failed route to infinity.Observable Description Max Value1correctness of response 2 points for Option 20 points for any other option239Refer to the exhibit. Hosts on the BOS Fa0/0 LAN are able to ping the Fa0/1 interface on the JAX router and a and ORL routers. Why would hosts from the 10.0.0.0/24 network not be able to ping hosts on the Fa0/0 LAN of n m l k j The JAX router has the wrong process ID.n m l k j The JAX router needs the network 10.0.0.0 0.0.0.255 area 0 command.n m l k j The JAX router needs the network 192.168.3.0 0.0.0.255 area 0 command.n m l k j The BOS router needs the network 192.168.3.0 0.0.0.255 area 0 command.Observable Description Max Value1correctness of response 2 points for Option 30 points for any other option240Which three statements describe the operation of routing with EIGRP? (Choose three.)g f e d c As new neighbors are discovered, entries are placed in a neighbor table.g f e d c If the feasible successor has a higher advertised cost than the current successor route, then it becomes theg f e d c If hello packets are not received within the hold time, DUAL must recalculate the topology.g f e d c The reported distance is the distance to a destination as advertised by a neighbor.g f e d c EIGRP maintains full knowledge of the network topology in the topology table and exchanges full routing inneighboring routers in every update.g f e d c EIGRP builds one routing table that contains routes for all configured routed protocols.Observable Description Max Value1correctness of response Option 1, Option 3, and Option 4 are correct.1 point for each correct option.0 points if more options are selected than required.341Refer to the exhibit. What happens to a packet that has 172.16.0.0/16 as the best match in the routing table tha nml k j The packet is discarded. nm l k j The packet is flooded out all interfaces. n ml k j The packet is forwarded via Serial0/0/0. n ml k j The packet is forwarded via FastEthernet0/0.ObservableDescriptionMax Value1correctness of response2 points for Option 10 points for any other option242 A network is configured with the IP, IPX, and AppleTalk protocols. Which routing protocol is recommended fornm l k j RIPv1 n m l k j RIPv2 nm l k j EIGRP n ml k j OSPFObservableDescriptionMax Value1correctness of response2 points for Option 30 points for any other option243Refer to the exhibit. Which two statements are true based on the exhibited output? (Choose two.) gf e d c Automatic summarization is disabled.g fe d c The EIGRP routing protocol is being used. g fe d c There is one feasible successor in the routing table. gf e d c The serial interface S0/0/0 is administratively down.g fe d c The router is originating the route to 172.16.1.0/24 via the S0/0/0 interface.Observable DescriptionMax Value1correctness of responseOption 1 and Option 2 are correct. 1 point for each correct option.0 points if more options are selected than required.244Refer to the exhibit. Two routers are unable to establish an adjacency. What is the possible cause for this? nm l k j The two routers are connected on a multiaccess network. n ml k j The hello and dead intervals are different on the two routers. n ml k j They have different OSPF router IDs. nm l k j They have different process IDs.ObservableDescriptionMax Value1correctness of response2 points for Option 20 points for any other option245What command would the network administrator apply to a router that is running OSPF to advertise the entireincluded in 172.16.0.0/19 in area 0?nm l k j R1(config-router)# network 172.16.0.0 0.0.0.255 area 0 nm l k j R1(config-router)# network 172.16.0.0 0.0.3.255 area 0 n ml k j R1(config-router)# network 172.16.0.0 0.0.15.255 area 0 n ml k j R1(config-router)# network 172.16.0.0 0.0.31.255 area 0ObservableDescriptionMax Value1correctness of response2 points for Option 40 points for any other option246What should be considered when troubleshooting a problem with the establishment of neighbor relationships b(Choose two.)gf e d c OSPF interval timers mismatchg fe d c administrative distance mismatch g fe d c interface network type mismatch g fe d c no loopback interface configured gf e d c gateway of last resort not redistributedObservableDescriptionMax Value1correctness of response1 point for each correct option. 0 points if more options are selected than required. 247Which two components are used to determine the router ID in the configuration of the OSPF routing process?gf e d c the IP address of the first FastEthernet interfaceg fe d c the highest IP address of any logical interface gf e d c the highest IP address of any physical interfaceg fe d c the default gateway IP address gf e d cthe priority value of 1 on any physical interfaceObservable DescriptionMax Value 1correctness of responseOption 2 and Option 3 are correct.1 point for each correct option.0 points if more options are selected than required. 248What is the function of the OSPF LSR packet?nm l k j It is used to confirm the receipt of LSUs. n ml k j It is used to establish and maintain adjacency with other OSPF routers. n ml k j It is used by the receiving routers to request more information about any entry in the DBD. n m l k j It is used to check the database synchronization between routers.Observable DescriptionMax Value 1correctness of response2 points for Option 30 points for any other option 249Refer to the exhibit. All interfaces are configured with the correct IP addresses and subnet masks. OSPF has brouting protocol. During troubleshooting, it is determined that hosts on network B can ping the Lo0 interface on reach hosts on network A. What is the cause of the problem?nm l k j Routers R1 and R2 have incorrect router IDs configured. nm l k j Router R1 is unable to form a neighbor relationship with router R2. n ml k j Routers R1 and R2 have been configured in different OSPF areas. n ml k j The configuration of router R1 fails to include network A in the OSPF routing process.Observable Description Max Value 1correctness of response2 points for Option 40 points for any other option 250Refer to the exhibit. The interface addresses and OSPF priorities are configured as shown. Because of the boo router A is currently the DR and router B is the BDR. If router A fails and is replaced the next day by a new rou protocol action or actions will happen when router D is connected to the network?nm l k j Router B will remain the DR, and router C will remain the BDR. n ml k j Router D will be elected DR, and router B will remain the BDR. nm l k j Router C will become the DR, and router B will become the BDR. n m l k j Router B will remain the BDR, and OSPF will function on the segment via the use of only the BDR.Observable DescriptionMax Value 1correctness of response2 points for Option 10 points for any other option 2Reset ViewShowing 1 of 1Prev Page:1NextClose WindowAll contents copyright ©2001-2010 Cisco Systems, Inc. All rights reserved. Privacy Statement and Trademarks.。
分光器稀疏配置约束下动态多播路由算法设计与仿真
密级分类号ＵＤＣ学位论文分光器稀疏配置约束下动态多播路由算法设计与仿真者姓名：王晓静导教师：杜荔副教授东北大学信息科学与工程学院请学位级别：硕士学科类别：工学科专业名称：通信与信息系统文提交日期：２０１０年６月论文答辩日期：２０１０年６月位授予日期：答辩委员会主席：荡汐民阅人：茸逮竖、砉ｒ磊尔北人学２０１０年６月ＡＴｈｅｓｉｓｉｎＴｅｌｅｏｍｍｕｎｉｃａｔｉｏｎｓａｎｄＩｎｆｏｒｍａｔｉｏｎＳｙｓｔｅｍｓＭｕｌｔｉｃａｓｔＤｅｓｉｇｎａｎｄＳｉｍｕｌａｔｉｏｎｏｆＤｙｎａｍｉｃＲｏｕｔｉｎｇＡｌｇｏｒｉｔｈｍｗｉｔｈＳｐａｒｓｅＬｉｇｈｔＳｐｌｉｔｔｅｒＣｏｎｆｉｇｕｒａｔｉｏｎＣｏｎｓｔｒａｉｎｔｂｙＷａｎｇＸｉａｏｊｉｎｇＳｕｐｅｒｖｉｓｏｒ：ＡｓｓｏｃｉａｔｅＰｒｏｆｅｓｓｏｒＤｕＬｉＮｏｒｔｈｅａｓｔｅｒｎＵｎｉｖｅｒｓｉｔｙＪｕｎｅ２０１０）独创性声明本人声明，所呈交的学位论文是在导师的指导下完成的。
论文中取得的研究成果除加以标注和致谢的地方外，不包含其他人己经发表或撰写过的研究成果，也不包括本人为获得其他学位而使用过的材料。
与我一同工作的同志对本研究所做的任何贡献均己在论文中作了明确的说明并表示谢＝艺思０学位论文作者签名：王眈穆日期：２汐，汐．６、瑚学位论文版权使用授权书本学位论文作者和指导教师完全了解东北大学有关保留、使用学位论文的规定：即学校有权保留并向国家有关部门或机构送交论文的复印件和磁盘，允许论文被查阅和借阅。
本人同意东北大学可以将学位论文的全部或部分内容编入有关数据库进行检索、交流。
作者和导师同意网上交流的时间为作者获得学位后：半年口一年口一年半口学位论文作者签名：王吼穆签字日期：２．０１０、６、２８／’两年西导师签名：签字日期：杠荔ＺＯｌ口．‘、２８东北大学硕士学位论文摘要分光器稀疏配置约束下动态多播路由算法设计与仿真摘要多播路由和波长分配是光网络多播研究的一个重要方面，与ＩＰ层多播相比，光层多播有一些特殊约束，包括波长连续性约束、分光节点稀疏配置约束、能量损伤约束。
Google Cloud VPN 配置指南说明书
2. On the VPC network page, click on Create VPC Network. 3. Fill in a Name and, optionally, a Description.
ISAKMP peer. 14. Copy this and save it somewhere secure 15. Under Routing options, choose Policy-based. 16. Under Remote network IP ranges, enter the internal IP address range of your router. 17. Under Local IP ranges, enter the local-address-selector that you created earlier.
5. Click on Create to create the VPC network.
6. Once the VPC network has been created, you will be returned to the Home Page. On the Home Page, from the menu on the left under Networking, select Hybrid Connectivity, then VPN.
C613-02084-00 REV A
Introduction | Page 3
Google Cloud VPN
Static routing
To configure static routing to a Google Cloud VPN, use the following steps: 1. Log in to your Google Cloud Account. Navigate to the Home Page. From the menu on the left,
高效间隙路由
ORIGINAL ARTICLEEnergy efficient clearance routing in WSNDevesh Pratap Singh•R.H.GoudarReceived:11January2014/Revised:19April2014ÓThe Society for Reliability Engineering,Quality and Operations Management(SREQOM),India and The Division of Operation and Maintenance,Lulea University of Technology,Sweden2014Abstract Wireless Sensor network consists of large number of sensor nodes,which are deployed densely.A key challenge of wireless sensor network is that their sensor nodes contain less amount of battery power.Con-sequently,efficient utilization of battery power inside the network is a main design consideration.In the querying or monitoring application like critical condition in a particular region,we need the network that respond in few seconds. For the purpose of response in less time,we require energy efficient routing of information inside the network.In this paper,we are proposing an energy efficient clearance routing method to collect/route the query towards a par-ticular region inside the network.Keywords WSNÁLocalizationÁRoutingÁEnergyefficiencyÁCoverage holes1IntroductionSensor networks are composed of huge number of sensor nodes;this forms the sensor network as a dense network. The sensor nodes in the network contains scarce amount of resources.Due to the untethered and unattended nature of sensor network,it is impossible to increase the power of these resources.The Sensor network contains less amount of power supply,so it is required to design an energy efficient routing protocol.It is required in WSN to disseminate some query to a par-ticular region.For example:We want to collect,‘‘What is the temperature at a region in a particular time range’’(Intan-agonwiwat et al.2000).An efficient way is needed to route the query towards a particular geographic region,Geographic routing(Finn1987;Bose et al.2001;Karp and Kung2000)are used to route the query in multi-hop network,in which an individual node not reliable and the topology of the network frequently changes.By using the location information the node gets the next hop for transferring the query.These nodes get the location information by the application of GPS and localizing systems(Li et al.2000;Das et al.2005).Many routing protocols are suggested for wireless net-work architecture are ad hoc on demand distance vector routing(AODV)(Perkins and Royer1999)and dynamic source routing(DSR)(Johnson et al.2003)are well known protocol for routing.Greedy perimeter stateless routing (GPSR)(Karp and Kung2000)protocol uses802.11mac layer,it is good for symmetric ad hoc networks Location Aided Routing(Ko and Vaidya1998)uses geographical location of nodes for routing the data in ad-hoc network. However none of these protocols are well suited for dis-seminating the queries in WSN.2Related works and motivationFinn(Finn1987)has given a geographical routing,in which he applied a restrictiveflooding to overcome the hole inside the network.One problem with this method is,D.P.SinghDepartment of Computer Science&Engineering(CSE),Graphic Era University,Dehradun,Indiae-mail:devesh.geu@R.H.Goudar(&)Department of Computer Network Engineering(CNE), Visvesvaraya Technological University(VTU),Belgaum590018,Indiae-mail:rhgoudar@DOI10.1007/s13198-014-0263-0it is difficult to decide the scope of the search.Karp and Kung (2000)have proposed GPSR,in GPSR the network is converted into a planer graph.The planer graph is much sparser than the original network.The nodes in this method are used in idle listening mode;it results in more utilization of battery power (Stemm and Katz 1997).Woo and Singh (2001)has given Scalable routing pro-tocol for ad hoc network,in this the location information of the nodes is maintained inside the network and the message is forwarded towards the destination region according to the location of the region.When the message arrived at thedestination region,source routing is used to send the message to a particular node.In this protocol,request/reply messages are used,that makes this protocol unsuitable for wireless sensor network.Imielinski and Goel (2000)have suggested a method for querying and monitoring in a physical space.In this,firstFig.1Construction of shortest pathtree Fig.4Euclidean space with threeJOINsFig.5Calculation of rotationalangleFig.6Euclidean space with exteriorboundaryFig.2MessageformatFig.3Formation of JOINthe query is forwarded towards the geonode,which behaves like the head of datacube.The geonode maintains a indexed structure and on the basis of this indexed structure it multicast the query inside the data cube.It is difficult to maintain the indexed structure inside the WSN.Chang and Tassiulas (2000)have given a type of flow augmentation and redirection algorithm,which balance the energy among the nodes,the problem with this approach is it needs the prior knowledge of message generation rates,consequently it assumes the stable topology in wireless ad hoc networks.Gao (2000),Gao et al.(2000)have proposed a table driven technique for the communication between the sen-sor nodes and the central information collecting entity,Fig.7Query propagation inside thenetworkFig.8Finding a point in the circumference ofholeFig.9Transmission and receivingmodelFig.10Formation of JOIN for oneholeFig.11Formation of JOINs for two holeswhich is known as USER.The work done by Gao et al.was good but it is not feasible to scale the fixed routing table and multi-path structure for a huge number of sensor nodes in sensor network.Sa´nchez et al.(2009)have analyzed the suppression of beacon messages in routing.The reactive information of location is exchanged inside the network,when a node need to route the data,consequently it supports the bea-conless protocols.They have analyzed many protocols such as:geographic random forwarding (GeRaF),implicit geographic forwarding (IGF)BOSS and BLR.Barbancho et al.(2008)have proposed a routing pro-tocols based on artificial intelligence techniques that guarantees some QOS requirements.This protocol uses a self-organized map for routing.The simulation results shows that this protocol reduces the end to end delay compared to directed diffusion.LGossiping protocol (Kheiri et al.2009),the nodes forward the message to its neighbors that are in its cover-age range,in this way the message reaches to the sink.This protocol uses the GPS to determine the location of each node.The delay problem was solved in some extent bythisFig.12Energy required when message size =1,024bits for different range of sensornodesFig.13Energy required when message size =5,120bits for different range of sensor nodesprotocol but it contains a problem of reachability of many messages to the sink.Norouzi et al.(2010)have given an ELGossiping pro-tocol for data distribution,the nodes in this protocol for-ward the data to its neighbor on the basis of coverage and lowest distance to the sink.Energy and distance are the important metrics that are exploited in this protocol.Many problem with this protocol are solved in its extensions but it still contains a problem of reachability of many messages to the sink.Now a days location aware sensor networks are used to remove the coverage holes and replacing the dead sensor nodes with active sensor nodes.Localization problem is discussed in (Sugihara and Gupta 2011;Jin et al.2011).Fig.14Traffic generated for different message lengthTable 1Simulation parameters Parameter Value Area 191km (fix)Network size 3routers 1coordinator Topologies Tree Simulation time 180min Packet size1,500bits Packet inter-arrival time (s)Constant (1)CSMA/CA parameters Default Sensing duration (s)0.1Physical layer parametersDefaultFig.15Throughput (number of sensor nodes are 45)The analysis and modeling of coverage holes in different routing protocols in wireless sensor network are elaborated in (Jia et al.2012;Liu et al.2011).3Problem formulationBefore describing energy efficient clearance routing (EECR)method,we want to state the assumptions that are taken,in description of the algorithm.•The base station (BS)node contains the GPS system by some localization algorithm we can get the location of other nodes inside the network.•The link between the sensor nodes is bidirectional.•Each query flow towards a region of interest inside the network.3.1Network modelOur network contains many sensor nodes n1,n2,n3,n4……..,generally pronounced as nodes in thefollowingFig.16End to end delay(number of sensor nodes are45)Fig.17Throughput (number of sensor nodes are60)Fig.18End to end delay(number of sensor nodes are 60)literature.Each sensor node senses its sensing area inside the sensing range.The sensor node can forward the sensed data in the form of message to other nodes in its sensing range.The network contains a region which is not covered by any node inside the network;these uncovered regions are known as hole.In a network there can be more than one hole.The network can be represented by Euclidean space and the nodes in the network can be known as Euclidean point.Each node has a little bit of memory to store the value of hop count.Now we want to describe our algorithm,after this algorithm we will elaborate all the step of algorithm then we will describe our query dissemination/data collection in the network.3.2EECR Algorithm(1)Step1.Select any exterior node p in the network. Step2.Perform theflooding from an exterior node p in the network and each node records the minimumhop count from p.This can provide the shortestpath tree from the node p.Step3.Determine the nodes that form the JOIN.The JOIN is a point where different homotopy typeshortest path tree meets.If there are multipleJOIN inside a network then there are more thanone hole inside the network.It can be found bythe method given by Alstrup et al.(2002)forcalculating the LCA.If there are two points(s,t)on the JOIN with same LCA(s,t).If we movefrom s or t(consider the edge between s and t)and go through LCA(s,t)with following thespanning tree,we will get a cycle.Step4.Determine the two nodes in the JOIN with minimum Euclidean distance from the node Step5.Determine the third node present in the inner boundary of the hole.It can be found by using thegeometric rules and Euclidean distance.Step6.Determine the nodes present in the inner bound-ary of the hole.Step7.Determine the nodes present in the outer bound-ary of the region.Step8.Now we have the set of cycles that provide the inner and outer boundary.Now we can get themedial axis between inner and outer boundary.3.2.1Construction of shortest path treeFirst initialize the hop countfield in each node as?except the exterior node p.The nodeflood the message,which contains its ID and the hop countfield(for p it’s0),the nodes that receive the message from p update its hop count by incre-menting the value of hop countfield in p’s message by1.Then the nodes that receive the message from the pflood the mes-sage with its ID and its hop count to its predecessor.In this way the predecessor nodes update its hop count by incrementing the value of its ancestor node hop count value by1.If any node inside the network get the message from its neighbor,first it compare the hop count value stored with the currently calculated value if the hop count value stored is less than the hop count value currently calculated it maintains the stored value otherwise it change the value by the currently calculated value and change the ID of the ancestor node accordingly.Figures1and2illustrate the method of construction of shortest path tree and message format respectively.For example,in the Fig.1node p isflooding the message (1||0)to its neighbors(r,s,t)which are in its sensing range. After receiving the message from p the neighbor nodes r,s and t calculates the hop count value which becomes 0?1=1,since1is less than?,the updated value for hop count in nodes r,s and t is stored as1.Now the node rfloods the message(2||1)to its neighbor’s x,y,p and s the neighbor nodes calculates the hop count values as followsp’s hop count value=r’s hop count value(from received message)?1=1?1=2s’s hop count value=r’s hop count value(from received message)?1=1?1=2x’s hop count value=r’s hop count value(from received message)?1=1?1=2y’s hop count value=r’s hop count value(from received message)?1=1?1=2Since the calculated hop count value for p is greater than the stored old value,so the hop count value for p will not be change,the calculated hop count value for s is also greater than the stored old value so it will also be not change.The calculated value of hop count for nodes x and y will change because it is less than the stored old value. We can say by this method we can get a spanning tree.Table2Comparison resultsParameters3600s7200sAODV EECR AODV EECR Retransmissionattempts0.02830.02580.0587440.039735 Network load16,965.4316,109.2416,580.5415,832.77 Data dropped10.1386110.1386138.019820.27723 Throughput29,363.7416,099.129,378.9715,812.5Lemma 1The method for shortest path tree will provide thedistinct spanning tree branches of depths d 1þd 22ÂÃand d 1þd 22ÂÃ,Where d 1and d 2are the depth of spanning tree branches.Proof Suppose the network’s Euclidean space contains one hole then there will be two distinct spanning tree branches in this Euclidean space that grow towards the sides of the hole.LetFig.19Throughput(bits/s)Fig.20Network load (bits/s)the depth of left spanning tree branch (T L )is d 1and the depth of right spanning tree branch (T R )is d 2.The nodes (n 1,n 2,n 3…..)at depth d 1contain the hop count value as h 1and the nodes (n 10,n 20,n 30……..)at depth d 2contain the hop count value h 2.Let the nodes (n 1,n 2,n 3…..)at depth d 1get the message fromnodes (n 10,n 20,n 30……..)at depth d 2and vice versa.IftheFig.21Retransmission attempts(packets)Fig.22Data dropped (bits/s)h 1[h 2?1then the nodes n 10,n 20,n 30……..will be added to T L and depth of T L becomes d 1?1.If h 2[h 1?1then the nodes n 1,n 2,n 3…will be added to T R and depth of T R becomesd 2?1.This process repeats until the depth becomes d 1þd 22ÂÃor d 1þd 2ÂÃ.After this either h 1will be one greater than h 2or h 2will be one greater than h 1or both becomesequal.Fig.23Data dropped(bits/s)Fig.24Network load (bits/s)3.2.2Determine the node that form the JOINWith the shortest path tree detected,we would like tofind the JOIN corresponding each hole present in the Euclidean space.Definition1The nodes where two branches of shortest path tree meets are known as JOIN nodes and the space that comes under these nodes is known as JOIN.The depth of the left shortest path tree branch(T L)is d1 and right shortest path tree branch(T R)is d2.Suppose there are n1,n2,n3….nodes are present at depth d1, n01;n02;n03......nodes are present at depth d2.The nodes that comes under each other sensing range are comes in the JOIN nodes.If we add the edge among these nodes we will get a graph,because these nodes are present at the fork of either tree T L or T R.The Euclidean distance between p and n1,n2,n3…..or n01;n02;n03......is b,then b will always be greater than zero.Generally we take the value of b in some constant fraction of the diameter of the sensing range. Alstrup et al.(2002)have given the method for getting the least common ancestor(LCA).The LCA of the two noden1;n01in the JOINðn1;n01Þ(will be far from the nodesn1and n01)will be the path from JOINðn1;n01Þto n1andpath from JOINðn1;n01Þto n10are well separated.Figure3 illustrates the formation of JOIN.When there is more than one hole in the Euclidean space there will be more than one JOIN spaces.By the above process we canfind the node in the different JOINs.3.2.3Determine the nodes present in the inner boundaryof the holeThe nodes in sensor network can be placed in a Euclidean space.Let the nodes that are present in the JOIN are n1,n2, n3,n4……….Find the Euclidean distance between p and n1,p and n2,p and n3,p and other nodes present in JOIN. Find the two lowest Euclidean distance nodes in JOIN. These nodes become the nodes in the circumference of the hole.Nowfind the node with highest depth that present in the line between p and a with lowest Euclidean distant node infirst JOIN in the shortest path tree which becomes the third node on the circumference of the hole.If there are more than one hole in the Euclidean space then there will be more than one JOIN,if the Euclidean space contains two holes then there will be two JOINs.We can get the two nodes on the circumference of the second hole by getting the two lowest Euclidean distance node from p in second JOIN.We can get the third node in the circumference of the second hole byfinding the node with highest Euclidean distance infirst JOIN.By repeating this process we can get three nodes on other JOINs present in the Euclideanspace. Fig.25Retransmission attempts(packets)For example,as we can see in the above Fig.4there are three holes.The nodes in the first JOIN are a(10,20,40),b(30,32,60),c(20,40,21),d(10,50,21),e(15,20,30),f(12,40,52),nodes in the second JOIN are g(5,60,30),h(10,45,15),i(20,25,22),j(15,30,40),k(20,30,40)and nodes in third JOIN are x(20,30,50),y(30,40,60),z(30,20,10),w(10,20,40).The node p is present at (0,0,0).The Euclidean distance (D)is givenasFig.26Throughput(bits/s)Fig.27Throughput (bits/s)Fig.28Retransmission attempts(packets)Fig.29Network load(bits/s)Euclidean Distance for nodes at first JOIN:D p ;a ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið10À0Þ2þð20À0Þ2þð40À0Þ2q ¼45:82D p ;b ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið30À0Þ2þð32À0Þ2þð60À0Þ2q ¼74:32D p ;c ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið20À0Þ2þð40À0Þ2þð21À0Þ2q ¼49:41D p ;d ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið10À0Þ2þð50À0Þ2þð21À0Þ2q ¼55:14D p ;e ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið15À0Þ2þð20À0Þ2þð30À0Þ2q ¼39:05D p ;f ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið12À0Þ2þð40À0Þ2þð52À0Þ2q ¼66:69Here the lowest two values are 45.82and 39.05,so a and e are on the circumference of the first hole.Euclidean distance for nodes at second JOIN:D p ;g ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið5À0Þ2þð60À0Þ2þð30À0Þ2q ¼67:27D p ;h ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið10À0Þ2þð45À0Þ2þð15À0Þ2q ¼48:47D p ;i ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið20À0Þ2þð25À0Þ2þð22À0Þ2q ¼38:84D p ;j ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið15À0Þ2þð30À0Þ2þð40À0Þ2q ¼52:20D p ;k ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið20À0Þ2þð30À0Þ2þð40À0Þ2q ¼53:85Here the lowest two values are 48.47and 38.84,so h and i are on the circumference of the second hole.Euclidean distance for nodes at third JOIN:D p ;x ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið20À0Þ2þð30À0Þ2þð50À0Þ2q ¼61:64D p ;y ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið30À0Þ2þð40À0Þ2þð60À0Þ2q ¼78:10D p ;z ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið30À0Þ2þð20À0Þ2þð10À0Þ2q ¼37:42D p ;w ðÞ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið10À0Þ2þð20À0Þ2þð40À0Þ2q ¼45:83Here,the lowest two values are 37.42and 45.83,so z and w are on the circumference of the third hole.3.2.4Algorithm (2)for determining the position of thirdnode on the JOIN(first)1.Start from the root node p,take the depth value d =02.Find the equation of line (L)between p and a.3.Find the next node (L n )in the shortest path tree that belongs in line L as follows (a)Put the coordinates value of L n in L and get the value(V)Fig.30Data dropped (bits/s)(b)if the value V becomes zero after±C where C isa threshold(c)then increment d and mark L n as the next root(d)repeat step3until we get the child node4.Mark the node L n as the third node on the circumfer-ence of thefirst hole.From the Fig.4we can say that we will get node L n as m (5,10,20)at depth d=10.3.2.5If the hole is present in the form of circle Suppose the equation of the circle isðxÀhÞ2þðyÀkÞ2¼r2ð1Þwhere the centre of the circle is(h,k)Calculating the centre of the circle and radius given the three point on the circumference as x1;y1ðÞ;x2;y2ðÞand x3;y3ðÞ:put x=x1and y=y1in Eq.(1)we can getðx1ÀhÞ2þðy1ÀkÞ2¼r2ð2ÞPut x=x2and y=y2in Eq.(1)ðx2ÀhÞ2þðy2ÀkÞ2¼r2ð3ÞPut x=x3and y=y3in Eq.(1)ðx3ÀhÞ2þðy3ÀkÞ2¼r2ð4ÞFrom the Eqs.(2),(3)and(4)We can get,ðx1ÀhÞ2þðy1ÀkÞ2¼ðx2ÀhÞ2þðy2ÀkÞ2ð5Þðx2ÀhÞ2þðy2ÀkÞ2¼ðx3ÀhÞ2þðy3ÀkÞ2ð6ÞSolve the Eq.(5)2h x1Àx2ðÞþ2k y1Ày2ðÞ¼x21þy21ÀÁÀx22þy22ÀÁLet x1Àx2¼A;y1Ày2¼B and x21Àþy21ÞÀx22þÀy22Þ¼2DSo,2Ahþ2Bk¼2D or AhþBk¼Dð7ÞSolve the Eq.(6)2h x2Àx3ðÞþ2k y2Ày3ðÞ¼x22þy22ÀÁÀx23þy23ÀÁLet x2Àx3¼A0;y2Ày3¼B0and x22þy22ÀÁÀx23þÀy23Þ¼2D0So,2A0hþ2B0k¼2D0or A0hþB0k¼D0ð8ÞSolve the Eqs.(7)and(8)for the values of h and kh¼2DBA0AÀðDB0þBD0ÞBA0ÀAB0ð9Þk¼DA0ÀAD0BA0ÀAB0ð10ÞFig.31Throughput(bits/s)Put the value of h and k from Eqs.(9),(10)and x =x 1,y =y 1in the circle Eq.(1).We can get the radius of the circle.So the we are known with the centre of the circle as (h,k)and radius as r.Suppose the coverage radius of the sensor node is given as rIn Fig.5,suppose the distance,L =r 0Then,h ¼r 0rÂ180p ð11ÞCalculation of the point d ðx 00;y 00Þfrom the point s(x 1,y 1)x 00y 001½ ¼x 1y 11½ Â100010Àh Àk 12435Âcos h sin h 0Àsin h cos h 00012435Â100010h k 12435same way,we can calculate the point z (x 000,y 000)asx 000y 0001½ ¼x 00y 001½ Â100010Àh Àk 12435Âcos h sin h 0Àsin h cos h 00012435Â100010h k12435 3.2.6If the hole is present in the form of other polygon The algorithm to find the inner boundary of the polygon asfollows:Consider the Fig.1with one hole.//Input:Take a buffer as AStep 1.Consider the left shortest path tree branch a.Find the value t 1=d 1-d -1.b.Put the coordinates of node n 1in A.c.Find the parent (P)of parent node of n 1.d.While t 1=0e.Find the right most child of P put the coordinate value of right most child in Af.t 1=t 1-1g.Assign node P as the parent of PStep 2.consider the right shortest path tree branch a.Find the value t 2=d 2-d -1.b.Put the coordinates of node n 10in A.c.Find the parent (P)of parent node of n 10.d.While t 2=0e.Find the left most child of P put the coordinate value of left most child in Af.t 2=t 2-1g.Assign node P as the parent ofPFig.32Retransmission attempts (packets)A contains the coordinates of the nodes present in cir-cumference of the hole.If there is more than one hole in the Euclidean space then apply the above algorithm for each pair of the lowest Euclidean distance at each hole.Lemma2The above algorithm will always provide the nodes on the circumference of the polygon.Proof If a tree T contains no any child then there will be no any left most child,in this case parent node becomes the left most child.But if we have a hole in the Euclidean space then there will be at least two subtrees,one on left side and another on right side.We know that these subtrees will meet at some JOIN.Then there will be some nodes will present on the circumference of the hole and these nodes will be the left most childs of the right subtree and right most child of the left subtree.As per the algorithm we are getting the left most or right most childs in right subtree or left subtree respectively at each level of the trees until we are not getting the node p in the Euclidean space.3.2.7Determine the nodes present in the exteriorboundary of the holeForfinding the exterior boundary of the hole we need to find the outer nodes.Definition2Outer node is a node which is far from the interior node in the Euclidean space.Following steps gives the Algorithm(3)forfinding the exterior boundary of the hole.//Input:take two buffer B and CStart from the root node p in the treeFind the val=max(d1,d2)For i=1to val doFind the left most child(l)and right most child(r)Put the Euclidean point of l in B and Euclidean point of r in CThe buffer B and C provides the nodes present in the exterior boundary of the Euclidean space.For example:Let the Euclidean space is given as fol-lows in Fig.6.3.2.8Determine the medial axis between inner and outerboundaryTo determine the medial axis we are using the Euclidean distance between the node and inner boundary,outer boundary.We are using the voronoi diagram to form the different region inside the Euclidean space on the basis of Euclidean distance from the inner or outer boundary.Let D i is the distance of a node from the inner boundary and D o is the distance of this node from the outer boundary.If D i is greater than D o then this node will belong to the region near to inner boundary.If D o is greater than the D i then this node will belong to the region near the outer boundary.The Fig.33Network load(bits/s)。
希尔斯通t系列智能下一代防火墙t3860 t5060 t5860用户指南说明书
Hillstone T-Series Intelligent Next-Generation FirewallT3860 / T5060 / T5860According to the latest research 66 percent of security breaches go undetected for 7-8 months. And, more than 85 percent of breaches originate from the web with drive-by downloads being the top web threat. This implies two things: First, a user does not have to click on anything to become infected with malware; and second, all organizations have infected hosts inside their network.Hillstone ,s T-Series intelligent Next-Generation Firewall (iNGFW) is an application-aware firewall that continuously monitors the network. It can identify attacks on all operating systems, applications, devices and browsers. It provides visibility into every stage of an attack and it can detect security breaches within minutes/seconds. It prioritizes hosts with the greatest security risks and provides contextual information about the threat. Security administrators can drill-down into the attack, including packet captures, to analyze all threat details.Hillstone ,s T-Series is designed for mid to large sized enterprises that need advanced levels of security, enhanced visibility, and continuous network uptime.TM- Outbound link load balancing includes policy based routing, ECMPand weighted, embedded ISP routing and dynamic detection- Inbound link load balancing supports SmartDNS and dynamicdetection- Automatic link switching based on bandwidth and latency- Link health inspection with ARP, PING, and DNSVPN• IPSec VPN:- IPSEC Phase 1 mode: aggressive and main ID protection mode- Peer acceptance options: any ID, specific ID, ID in dialup user group - Supports IKEv1 and IKEv2 (RFC 4306)- Authentication method: certificate and pre-shared key- IKE mode configuration support (as server or client)- DHCP over IPSEC- Configurable IKE encryption key expiry, NAT traversal keep alivefrequency- Phase 1/Phase 2 Proposal encryption: DES, 3DES, AES128, AES192,AES256- Phase 1/Phase 2 Proposal authentication: MD5, SHA1, SHA256,SHA384, SHA512- Phase 1/Phase 2 Diffie-Hellman support: 1,2,5- XAuth as server mode and for dialup users- Dead peer detection- Replay detection- Autokey keep-alive for Phase 2 SA• IPSEC VPN realm support: allows multiple custom SSL VPN logins associated with user groups (URL paths, design)• IPSEC VPN configuration options: route-based or policy based• IPSEC VPN deployment modes: gateway-to-gateway, full mesh,hub-and-spoke, redundant tunnel, VPN termination in transparent mode• One time login prevents concurrent logins with the same username • SSL portal concurrent users limiting• SSL VPN port forwarding module encrypts client data and sends the data to the application server• SSL VPN tunnel mode supports clients that run iOS, Android, and Windows XP/Vista including 64-bit Windows OS’• Host integrity checking and OS checking prior to SSL tunnel connections • MAC host check per portal• Cache cleaning option prior to ending SSL VPN session• L2TP client and server mode, L2TP over IPSEC, and GRE over IPSEC• View and manage IPSEC and SSL VPN connectionsUser and Device Identity• Local user database• Remote user authentication: LDAP, Radius, Active Directory• Single-sign-on: Windows AD• 2-factor authentication: 3rd party support, integrated token server with physical and SMS• User and device-based policiesIPS• 7,000+ signatures, protocol anomaly detection, rate-based detection, custom signatures, manual, automatic push or pull signature updates, integrated threat encyclopedia• IPS Actions: default, monitor, block, reset (attackers IP or victim IP, incoming interface) with expiry time• Packet logging option• Filter Based Selection: severity, target, OS, application or protocol• IP exemption from specific IPS signatures• IDS sniffer mode• IPv4 and IPv6 rate based DOS protection with threshold settings against TCP Syn flood, TCP/UDP/SCTP port scan, ICMP sweep,TCP/UDP/SCIP/ICMP session flooding (source/destination)• Active bypass with bypass interfaces• Provides predefined template of defense configuration• Predefined prevention configurationThreat Protection• Breach Detection- Near real-time breach detection (seconds/minutes)- Detailed description and severity of malware closely resembling attack - Pcap files and log files provide corroborating evidence- Confidence level provides certainty of attack• Network Behavior Analysis- L3-L7 baseline traffic compared to real-time traffic to revealanomalous network behavior- Built-in mitigations technologies include: session limits, bandwidthlimits and blocking- Graphical depiction of anomalous behavior compared to baseline and upper and lower thresholds• Network Risk Index quantifies the threat level of the network based on the aggregate host index.• Host Risk Index quantifies the host threat level based on attack severity, detection method, and confidence level.• Over 1.3 million AV signatures• Botnet server IP blocking with global IP reputation database• Flow-based Antivirus: protocols include HTTP, SMTP, POP3, IMAP,FTP/SFTP• Flow-based web filtering inspection• Manually defined web filtering based on URL, web content and MIME header• Dynamic web filtering with cloud-based real-time categorization database: over 140 million URLs with 64 categories (8 of which are security related)• Additional web filtering features:- Filter Java Applet, ActiveX or cookie- Block HTTP Post- Log search keywords- Exempt scanning encrypted connections on certain categories forprivacy• Web filtering profile override: allows administrator to temporarily assign different profiles to user/group/IP• Web filter local categories and category rating override• Proxy avoidance prevention: proxy site category blocking, rate URLs by domain and IP address, block redirects from cache & translation sites, proxy avoidance application blocking, proxy behavior blocking (IPS)• Inspect SSL encrypted traffic.Application Control• Over 3,000 applications that can be filtered by name, category, subcategory, technology and risk• Each application contains a description, risk factors, dependencies, typical ports used, and URLs for additional reference• Actions: block, reset session, monitor, traffic shapingHigh Availability• Redundant heartbeat interfaces• Active/Passive• Standalone session synchronization• HA reserved management interface• Failover:- Port, local & remote link monitoring- Stateful failover- Sub-second failover- Failure notification• Deployment Options:- HA with link aggregation- Full mesh HA- Geographically dispersed HAAdministration• Management access: HTTP/HTTPS, SSH, telnet, console• Central Management: Hillstone Security Manager (HSM), web service APIs• System Integration: SNMP, syslog, alliance partnerships• Rapid deployment: USB auto-install, local and remote script execution • Dynamic real-time dashboard status and drill-in monitoring widgets • Language support: EnglishLogs & Reporting• Logging facilities: local memory and storage (if available), multiple syslog servers and multiple Hillstone Security Audit (HSA) platforms • Encrypted logging and log integrity with HSA scheduled batch log uploading• Reliable logging using TCP option (RFC 3195)• Detailed traffic logs: forwarded, violated sessions, local traffic, invalid packets• Comprehensive event logs: system and administrative activity audits, routing & networking, VPN, user authentications, WiFi related events • IP and service port name resolution option• Brief traffic log format optionProduct Specification4GE Bypass Extension ModuleIOC-4XFP8SFP+ Extension Module4SFP+ Extension Module4 x SFP+, SFP+ module not included(1)IPS Throughput data is obtained under 1M-byte-payload HTTP traffic with test of 32K-byte scanning.(2) AV Throughput data is obtained under 1M-byte-payload HTTP traffic with file attachment.(3) IPSec Throughput data is obtained under Preshare Key AES256+SHA-1 configuration and 1400-byte packet size packet .Unless specified otherwise, all performance, capacity and functionality are based on StoneOS 5.5R1. Results may vary based on StoneOS® version and deployment.。
基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法
江苏农业学报(ＪｉａｎｇｓｕＪ.ｏｆＡｇｒ.Ｓｃｉ.)ꎬ２０２３ꎬ３９(９):１８２７￣１８３３ｈｔｔｐ://ｊｓｎｙｘｂ.ｊａａｓ.ａｃ.ｃｎ张善文ꎬ许新华ꎬ齐国红.基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法[Ｊ].江苏农业学报ꎬ２０２３ꎬ３９(９):１８２７￣１８３３ｄｏｉ:１０.３９６９/ｊ.ｉｓｓｎ.１０００￣４４４０.２０２３.０９.００４基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法张善文ꎬ㊀许新华ꎬ㊀齐国红(郑州西亚斯学院电子信息工程学院ꎬ河南郑州４５１１５０)收稿日期:２０２２￣１１￣１２基金项目:国家自然科学基金项目(６２１７２３３８)ꎻ河南省科技厅科技攻关项目(２２２１０２１１０１３４)ꎻ河南省高等学校重点科研项目(２２Ｂ５２００４９)作者简介:张善文(１９６５－)ꎬ男ꎬ陕西西安人ꎬ博士ꎬ教授ꎬ主要从事人工智能在精准农业中的应用研究ꎮ(Ｅ￣ｍａｉｌ)ｗｊｄｗ７１６＠１６３.ｃｏｍ㊀㊀摘要:㊀在黄瓜叶部病害检测中ꎬ传统方法简单但检测正确率低ꎬ难以处理多种多样的病害叶片图像ꎬ深度卷积网络的检测正确率高ꎬ但依赖于大量训练样本ꎬ训练时间长ꎮ本研究提出一种基于孪生多尺度空洞胶囊网络(Ｓｉ￣ａｍｅｓｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃａｐｓｕｌｅｎｅｔｗｏｒｋꎬＳＭＳＤＣＮｅｔ)的黄瓜叶部病害检测方法ꎬ该方法整合了孪生网络㊁空洞卷积网络和胶囊网络的优势ꎬ将多尺度空洞卷积模块Ｉｎｃｅｐｔｉｏｎ引入胶囊网络ꎬ作为孪生网络的子网络ꎬ构建孪生多尺度空洞胶囊网络模型ꎬ提取多尺度判别特征ꎬ再进行矢量化处理ꎬ最后经动态路由算法得到具有空间位置信息的胶囊向量ꎬ进行病害检测与识别ꎮＳＭＳＤＣＮｅｔ克服了深度卷积网络需要大量训练样本㊁训练时间长以及对旋转和仿射变换敏感的问题ꎬ并且克服了多尺度卷积网络训练参数较多的问题ꎮ在一个较小的黄瓜病害叶片图像数据集上进行试验ꎬ病害检测精度达９０％以上ꎮ结果表明ꎬ该方法能够实现小训练样本集的黄瓜叶部病害检测ꎬ为训练样本有限情况下的作物病害检测提供了一种新方法ꎮ关键词:㊀黄瓜病害ꎻ孪生网络ꎻ多尺度空洞卷积ꎻ胶囊网络ꎻ孪生多尺度空洞胶囊网络中图分类号:㊀ＴＰ３９１.４１ꎻＳ６４２.２㊀㊀㊀文献标识码:㊀Ａ㊀㊀㊀文章编号:㊀１０００￣４４４０(２０２３)０９￣１８２７￣０７ＣｕｃｕｍｂｅｒｌｅａｆｄｉｓｅａｓｅｄｅｔｅｃｔｉｏｎｂａｓｅｄｏｎＳｉａｍｅｓｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃａｐ￣ｓｕｌｅｎｅｔｗｏｒｋＺＨＡＮＧＳｈａｎ￣ｗｅｎꎬ㊀ＸＵＸｉｎ￣ｈｕａꎬ㊀ＱＩＧｕｏ￣ｈｏｎｇ(ＳｃｈｏｏｌｏｆＥｌｅｃｔｒｏｎｉｃｓＩｎｆｏｒｍａｔｉｏｎＥｎｇｉｎｅｅｒｉｎｇꎬＺｈｅｎｇｚｈｏｕＳＩＡＳＵｎｉｖｅｒｓｉｔｙꎬＺｈｅｎｇｚｈｏｕ４５１１５０ꎬＣｈｉｎａ)㊀㊀Ａｂｓｔｒａｃｔ:㊀Ｉｎｃｕｃｕｍｂｅｒｌｅａｆｄｉｓｅａｓｅｄｅｔｅｃｔｉｏｎꎬｔｈｅｔｒａｄｉｔｉｏｎａｌｍｅｔｈｏｄｓａｒｅｓｉｍｐｌｅｂｕｔｌｏｗｄｅｔｅｃｔｉｏｎａｃｃｕｒａｃｙꎬａｎｄｔｈｅｙａｒｅｄｉｆｆｉｃｕｌｔｔｏｄｅａｌｗｉｔｈｔｈｅｖａｒｉｏｕｓｄｉｓｅａｓｅｄｌｅａｆｉｍａｇｅｓ.Ｄｅｅｐｃｏｎｖｏｌｕｔｉｏｎｎｅｕｒａｌｎｅｔｗｏｒｋｓ(ＣＮＮｓ)ｈａｖｅｈｉｇｈｄｅｔｅｃｔｉｏｎａｃ￣ｃｕｒａｃｙꎬｂｕｔｔｈｅｙｒｅｌｙｏｎａｌａｒｇｅｎｕｍｂｅｒｏｆｔｒａｉｎｉｎｇｓａｍｐｌｅｓꎬａｎｄｔｈｅｔｒａｉｎｉｎｇｔｉｍｅｉｓｌｏｎｇ.ＡｃｕｃｕｍｂｅｒｌｅａｆｄｉｓｅａｓｅｄｅｔｅｃｔｉｏｎｍｅｔｈｏｄｂａｓｅｄｏｎＳｉａｍｅｓｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃａｐｓｕｌｅｎｅｔｗｏｒｋ(ＳＭＳＤＣＮｅｔ)ｗａｓｐｒｏｐｏｓｅｄ.ＩｔｉｎｔｅｇｒａｔｅｄｔｈｅａｄｖａｎｔａｇｅｓｏｆＳｉａ￣ｍｅｓｅｎｅｔｗｏｒｋꎬｄｉｌａｔｅｄｃｏｎｖｏｌｕｔｉｏｎｎｅｔｗｏｒｋａｎｄｃａｐｓｕｌｅｎｅｔｗｏｒｋ(ＣａｐｓＮｅｔ).ＩｎＳＭＳＤＣＮｅｔꎬｔｈｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃｏｎｖｏｌｕ￣ｔｉｏｎｉｎｃｅｐｔｉｏｎｍｏｄｕｌｅｗａｓｉｎｔｒｏｄｕｃｅｄｉｎｔｏＣａｐｓＮｅｔｔｏｃｏｎｓｔｒｕｃｔｔｈｅｔｗｏｓｕｂ￣ｎｅｔｗｏｒｋｓｆｏｒＳｉａｍｅｓｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃａｐｓｕｌｅｎｅｔｗｏｒｋｍｏｄｅｌꎬｔｈｅｎｔｈｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｓｃｒｉｍｉｎａｎｔｆｅａｔｕｒｅｓｗｅｒｅｅｘｔｒａｃｔｅｄａｎｄｖｅｃｔｏｒｉｚｅｄ.Ｆｉｎａｌｌｙꎬｔｈｅｃａｐｓｕｌｅｖｅｃｔｏｒｗｉｔｈｓｐａ￣ｔｉａｌｌｏｃａｔｉｏｎｉｎｆｏｒｍａｔｉｏｎｗａｓｏｂｔａｉｎｅｄｔｈｒｏｕｇｈｔｈｅｄｙｎａｍｉｃｒｏｕｔｉｎｇａｌｇｏｒｉｔｈｍｆｏｒｄｅｔｅｃｔｉｎｇａｎｄｒｅｃｏｇｎｉｚｉｎｇｃｕｃｕｍｂｅｒｌｅａｆｄｉｓｅａ￣ｓｅｓ.ＳＭＳＤＣＮｅｔｏｖｅｒｃａｍｅｔｈｅｐｒｏｂｌｅｍｓｏｆｄｅｅｐｃｏｎｖｏｌｕｔｉｏｎａｌｎｅｔｗｏｒｋｓｔｈａｔｒｅｑｕｉｒｅｄａｌａｒｇｅｎｕｍｂｅｒｏｆｔｒａｉｎｉｎｇｓａｍｐｌｅｓꎬｌｏｎｇｔｒａｉｎｉｎｇｔｉｍｅꎬａｎｄｓｅｎｓｉｔｉｖｉｔｙｔｏｒｏｔａｔｉｏｎａｎｄａｆｆｉｎｅｔｒａｎｓｆｏｒｍａｔｉｏｎꎬａｎｄｏｖｅｒｃａｍｅｔｈｅｐｒｏｂｌｅｍｔｈａｔｍｕｌｔｉ￣ｓｃａｌｅｃｏｎｖｏｌｕｔｉｏｎａｌｎｅｔｗｏｒｋｓｒｅｑｕｉｒｅｄｍｏｒｅｔｒａｉｎｉｎｇｐａｒａｍｅｔｅｒｓ.Ｄｉｓｅａｓｅｄｅｔｅｃｔｉｏｎｅｘｐｅｒｉｍｅｎｔｓｗｅｒｅｃｏｎｄｕｃｔｅｄｏｎａｓｍａｌｌｃｕｃｕｍｂｅｒｄｉｓｅａｓｅｌｅａｆｉｍａｇｅｄａｔａｓｅｔ.Ｔｈｅｄｅｔｅｃｔｉｏｎａｃｃｕｒａｃｙｗａｓｍｏｒｅｔｈａｎ９０％.Ｔｈｅｒｅｓｕｌｔｓｓｈｏｗｅｄｔｈａｔｔｈｅｐｒｏｐｏｓｅｄｍｅｔｈｏｄｃｏｕｌｄｄｅｔｅｃｔｃｕｃｕｍｂｅｒｌｅａｆｄｉｓｅａｓｅｗｉｔｈｓｍａｌｌｔｒａｉｎ￣７２８１ｉｎｇｓａｍｐｌｅｓｅｔꎬｗｈｉｃｈｐｒｏｖｉｄｅｄａｎｅｗｍｅｔｈｏｄｆｏｒｄｉｓｅａｓｅｄｅｔｅｃｔｉｏｎｕｎｄｅｒｔｈｅｃｏｎｄｉｔｉｏｎｏｆｌｉｍｉｔｅｄｔｒａｉｎｉｎｇｓａｍｐｌｅｓ.Ｋｅｙｗｏｒｄｓ:㊀ｃｕｃｕｍｂｅｒｄｉｓｅａｓｅꎻＳｉａｍｅｓｅｎｅｔｗｏｒｋꎻｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃｏｎｖｏｌｕｔｉｏｎꎻｃａｐｓｕｌｅｎｅｔｗｏｒｋꎻＳｉａｍｅｓｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃａｐｓｕｌｅｎｅｔｗｏｒｋ(ＳＭＳＤＣＮｅｔ)㊀㊀及时㊁准确地检测作物病害是病害防治的提前ꎮ传统的作物病害检测方法特别依赖于提取的分类特征ꎬ但很难从复杂㊁多变㊁多种多样的病害叶片图像中提取到通用的特征ꎬ所以这些方法的检测率不高且泛化性不强[１￣３]ꎮ卷积神经网络(ＣＮＮ)能够克服传统病害检测方法的不足ꎬ提取复杂图像的深层分类特征ꎬ在作物病害检测中取得了显著效果[４￣５]ꎮ邵明月等[６]总结了从早期的作物病害检测方法到基于ＣＮＮ检测方法的研究进展ꎬ阐述了近年来ＣＮＮ在作物病害检测中的研究成果以及各种方法的优缺点ꎮＮｇｕｇｉ等[７]回顾了近年来基于图像处理和机器学习的作物病虫害识别方法ꎬ分析了１０种ＣＮＮ架构在识别准确率㊁召回率㊁准确性㊁特异性㊁Ｆ１得分㊁训练持续时间和存储要求方面的作物病害识别性能ꎬ为该领域的深入研究提供了技术指导ꎮ深度ＣＮＮ的检测准确率比较高ꎬ但从多种多样的病害叶片图像中提取有效的分类特征ꎬ需要大量训练样本ꎬ模型的训练时间很长ꎬ而且对图像的旋转和仿射变换比较敏感[８]ꎮ为了克服深度ＣＮＮ的不足ꎬ出现了很多改进的ＣＮＮ模型ꎮ利用多个ＣＮＮ的组合能够提取互补的分类特征ꎬ增强模型的检测能力[９]ꎻ多尺度ＣＮＮ能够极大提高病害检测方法的特征提取能力[１０]ꎻ将注意力机制加入ＣＮＮ能够加快网络收敛ꎬ提高网络的抗噪能力[１１]ꎻ轻量级ＣＮＮ减少了训练样本和网络参数ꎬ加速网络收敛[１２￣１３]ꎻ迁移学习加快了模型训练ꎬ增强了模型的泛化能力[１４]ꎻ空洞卷积能够在不增加训练参数和不增加计算代价的情况下扩大卷积的感受野ꎬ从而提高多尺度特征提取性能[１５￣１６]ꎮ胶囊网络(ＣａｐｓＮｅｔ)利用动态路由算法提取胶囊特征向量ꎬ弥补了ＣＮＮ失去空间位置信息的不足ꎬ克服了深度ＣＮＮ对图像的旋转和仿射变换较为敏感的问题[１７]ꎮＣａｐｓＮｅｔ使用向量特征状态的重要信息ꎬ向量的模和方向分别表示特征存在的概率和姿态信息ꎬ移动特征可以改变胶囊向量ꎬ但不影响特征存在概率ꎬ特别适用于不规则的㊁多种多样的作物病害叶片图像识别[１８]ꎮ孪生网络(Ｓｉａｍｅｓｅ)能够克服ＣＮＮ和ＣａｐｓＮｅｔ依赖于大量训练样本的问题ꎮ孪生网络通过共享权值来实现其功能ꎬ能够通过已有类别中的特征相似度ꎬ准确判断２个相似的样本ꎬ还能利用先验知识识别新样本的类别[１９￣２０]ꎬ在样本类别多且每个类别的训练样本较少的情况下取得较好的效果[２１]ꎮ针对实际大田作物病害叶片图像较少ꎬ包含遮挡和复杂背景ꎬ病害叶片图像病斑的大小㊁形状和颜色等差异较大等问题ꎬ本研究拟提出一种基于孪生多尺度空洞胶囊网络(ＳＭＳＤＣＮｅｔ)的黄瓜叶部病害检测方法ꎬ减少对模型训练样本量的依赖ꎬ在有限训练样本中获取到多类别样本间的共同特征ꎬ以期提高黄瓜病害检测准确率ꎮ１㊀材料与方法１.１㊀图像数据集与图像预处理本研究构建了１个用于黄瓜叶部病害检测的病害叶片图像数据集ꎬ共有５００幅图像ꎬ包括黄瓜常见的白粉病㊁褐斑病㊁黑斑病㊁炭疽病和角斑病的叶片图像各１００幅ꎬ其中７０％的图像在陕西省杨凌农业示范区采集ꎬ３０％的图像来自于数据集ＰｌａｎｔＶｉｌｌａｇｅ(ｈｔｔｐｓ://ｗｗｗ.ｐｌａｎｔｖｉｌｌａｇｅ.ｏｒｇ/ｅｎ/ｔｏｓ)ꎮ不同病害的黄瓜叶片图像如图１所示ꎮ图１㊀感染不同病害的黄瓜叶片图像Ｆｉｇ.１㊀Ｉｍａｇｅｓｏｆｃｕｃｕｍｂｅｒｌｅａｖｅｓｉｎｆｅｃｔｅｄｗｉｔｈｄｉｆｆｅｒｅｎｔｄｉｓｅａｓｅｓ８２８１江苏农业学报㊀２０２３年第３９卷第９期㊀㊀数据集中的叶片图像大小不一ꎬ因此在试验过程中将图像均匀归一化为２２４ˑ２２４像素ꎬ以５折交叉验证策略将数据集随机划分为训练集和测试集ꎬ训练集用于模型训练ꎬ测试集用于测试模型的性能ꎬ用５次试验结果的平均值作为识别结果ꎮ１.２㊀相关方法１.２.１㊀多尺度空洞卷积模块Ｉｎｃｅｐｔｉｏｎ㊀Ｉｎｃｅｐｔｉｏｎ是一个多尺度卷积网络模块ꎬ能够并行组合不同的卷积层ꎬ由不同卷积层提取的特征在深度㊁维度上拼接以形成更深的矩阵ꎬ提取不同尺度特征ꎬ其结构见图２ꎮ１.２.２㊀胶囊网络㊀胶囊网络以胶囊层为数据处理单元ꎬ采用动态路由算法在胶囊层之间传输数据ꎬ具有比ＣＮＮ更好的特征表达能力ꎮ胶囊网络的基本结构如图３所示ꎬ卷积模块用于从原始图像中提取分类特征ꎬ主胶囊模块用于将提取的分类特征表示转换为特征矢量ꎬ数字胶囊模块使用动态路由算法更新网络参数ꎬ避免因池化操作而造成的损失ꎬ最终输出特征向量ꎬ其长度为测试样本属于某一类的概率ꎮ图２㊀Ｉｎｃｅｐｔｉｏｎ模块Ｆｉｇ.２㊀Ｉｎｃｅｐｔｉｏｎｍｏｄｕｌｅ１.２.３㊀孪生网络㊀孪生网络的结构比较简单ꎬ图４显示ꎬ孪生网络通过２个具有相同结构㊁共享权重的ＣＮＮ模型提取２个输入图像的特征向量ꎬ然后通过最小化相同类别之间的损失ꎬ同时最大化不同类别样本的损失训练模型参数ꎬ再通过迭代训练后判别２个样本是否相似ꎬ最后计算２组特征向量的相似度ꎬ进行样本分类识别ꎮＷｉｊ为模型的卷积核ꎮ图３㊀胶囊网络的基本结构Ｆｉｇ.３㊀Ｓｔｒｕｃｔｕｒｅｏｆｃａｐｓｕｌｅｎｅｔｗｏｒｋ图４㊀孪生网络结构Ｆｉｇ.４㊀ＳｔｒｕｃｔｕｒｅｏｆＳｉａｍｅｓｅｎｅｔｗｏｒｋ１.３㊀孪生多尺度空洞胶囊网络(ＳＭＳＤＣＮｅｔ)针对深度ＣＮＮ模型训练参数多ꎬ需要大量训练样本ꎬ但从大田黄瓜种植园难以采集到足够的病害叶片图像ꎬ导致基于ＣＮＮ及其改进模型的作物病害识别方法的训练性能较差ꎬ容易出现过拟合等问题ꎮ本研究提出一种孪生多尺度空洞胶囊网络ꎬ应用于黄瓜病害识别ꎮＳＭＳＤＣＮｅｔ的基本构架如图５Ａ所示ꎬ由２个结构相同的空洞Ｉｎｃｅｐｔｉｏｎ模块进行特征提取ꎬ再基于余弦距离的分类准则进行病害检测ꎮ由图５Ａ看出ꎬＳＭＳＤＣＮｅｔ为孪生胶囊网络的改进模型ꎬ由２个结构相同的卷积层和胶囊层构成ꎬ病害检９２８１张善文等:基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法测由相似度度量层和分类器完成ꎮ卷积层由３个空洞Ｉｎｃｅｐｔｉｏｎ模块构成(图５Ｂ)ꎬ其中批量归一化层加快模型的正向㊁反向传播速率ꎮ在ＳＭＳＤＣＮｅｔ中ꎬ利用空洞卷积代替Ｉｎｃｅｐｔｉｏｎ模块中卷积ꎬ构建一种空洞Ｉｎｃｅｐｔｉｏｎ模块ꎬ其结构见图５ＣꎮＷｉｊ为模型的卷积核ꎮ图５㊀多尺度空洞胶囊网络(ＳＭＳＤＣＮｅｔ)架构Ｆｉｇ.５㊀ＡｒｃｈｉｔｅｃｔｕｒｅｏｆＳｉａｍｅｓｅｍｕｌｔｉ￣ｓｃａｌｅｄｉｌａｔｅｄｃａｐｓｕｌｅｎｅｔｗｏｒｋ(ＳＭＳＤＣＮｅｔ)㊀㊀主胶囊层设置３２个主胶囊ꎬ数字胶囊层有１０个数字胶囊ꎬ每个胶囊输入一个６ˑ６ˑ８张量ꎬ输出１６ˑ１０的矩阵ꎮ由图像对的相似度判定待分类图像的类别ꎬ２个特征向量ｘ１㊁ｘ２之间的余弦距离[ｄｃｏｓ(ｘ１ꎬｘ２)]作为相似度ꎬ计算如下:ｄｃｏｓ(ｘ１ꎬｘ２)＝ｘ１ ｘ２/( ｘ１ ｘ２ )(１)通过对比误差损失函数Ｌｏｓｓ进行反向优化模型参数ꎬ表示如下:Ｌｏｓｓ(ｌꎬｘ１ꎬｘ２)＝１２Ｎ Ｎｎ＝１ｌｄ２ｃｏｓ(ｘ１ꎬｘ２)＋(１－ｌ)[λ－ｄ２ｃｏｓ(ｘ１ꎬｘ２)ꎬ０]ｍａｘ(２)式中ꎬｘ１㊁ｘ２为２个特征向量ꎻｄｃｏｓ(ｘ１ꎬｘ２)为特征向量ｘ１㊁ｘ２之间的相似度ꎻｌ为２个图像是否匹配的标签ꎻＮ为样本个数ꎻλ为设定的阈值ꎬ由试验确定ꎬ默０３８１江苏农业学报㊀２０２３年第３９卷第９期认值为０.５ꎮ设待检测图像映射的特征向量为ｓꎬ带标签图像的特征向量为ｔｉ(ｔｉɪ[１ꎬＣ])ꎬＣ为病害类别数ꎬ按相似度打分排序ꎬ用最大值判定待检测图像的病害类型(Ｌａｂｅｌ)ꎮＬａｂｅｌ＝[ｄｃｏｓ(ｓꎬｔｉ)]ｍａｘ(３)式中ꎬｄｃｏｓ(ｓꎬｔｉ)为待识别图像与带标签图像的相似度ꎮ１.４㊀基于ＳＭＳＤＣＮｅｔ的黄瓜病害识别过程图６显示ꎬ基于ＳＭＳＤＣＮｅｔ的黄瓜病害识别过程主要为:第一ꎬ对图像进行归一化处理ꎬ并构建图像对ꎻ第二ꎬ在ＳＭＳＤＣＮｅｔ的每个子网络中ꎬ首先由３个空洞Ｉｎｃｅｐｔｉｏｎ模块依次提取图像的多尺度卷积特征ꎬ然后由胶囊网络进一步提取特征向量ꎬ得到１６ˑ１０的向量矩阵ꎻ第三ꎬ构建图像的特征向量对ꎬ按照公式(１)计算每个向量对图像的相似度ꎻ第四ꎬ利用公式(２)得到的误差损失函数Ｌｏｓｓ优化模型参数ꎬ反复训练直至完成所设定的迭代次数ꎻ第五ꎬ计算待识别图像与带标签图像的相似度ｄｃｏｓ(ｓꎬｔｉ)ꎬ由ｄｃｏｓ(ｓꎬｔｉ)的最大值判断待识别图像的病害类型ꎮ图６㊀基于ＳＭＳＤＣＮｅｔ的黄瓜病害识别过程Ｆｉｇ.６㊀ＣｕｃｕｍｂｅｒｄｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎｂａｓｅｄｏｎＳＭＳＤＣＮｅｔ２㊀结果与分析为了证明利用ＳＭＳＤＣＮｅｔ进行黄瓜病害识别的有效性ꎬ进行了试验验证ꎮＳＭＳＤＣＮｅｔ训练的批处理大小设为２５ꎬ迭代次数为３０００ꎬ学习率设为１.５ˑ１０－４ꎬＡｄａｍ作为模型的优化器ꎮ所有方法均在Ｐｙ￣ｔｈｏｎ３.３环境下以Ｔｅｎｓｏｒｆｌｏｗ１.４编程实现ꎬＣＮＮ网络模型在深度学习的Ｋｅｒａｓ框架下搭建ꎬ系统为Ｕｂｕｎ￣ｔｕ１４.０４ꎬ核心硬件运算平台为Ｉｎｔｅｌｉ７ＣＰＵＴｉＧＰＵ１０８０ꎮ对３种孪生网络[孪生卷积神经网络(ＳＣ￣ＮＮ)[２０]㊁孪生胶囊网络(ＳＣａｐｓＮｅｔ)[２１]和ＳＭＳＤＣ￣Ｎｅｔ]进行５折交叉验证对比试验ꎬ分析识别率随迭代次数变化的结果ꎮ图７显示ꎬ当迭代次数达到１０００次后ꎬＳＭＳＤＣＮｅｔ趋于收敛ꎬ其收敛效果和病害检测精度明显高于其他２种模型ꎮ表明空洞Ｉｎｃｅｐ￣ｔｉｏｎ收敛最快㊁病害识别效果更好ꎮＳＭＳＤＣＮｅｔ和ＳＣａｐｓＮｅｔ的收敛效果优于ＳＣＮＮꎬ主要原因是ＳＭＳ￣ＤＣＮｅｔ和ＳＣａｐｓＮｅｔ中将多尺度卷积Ｉｎｃｅｐｔｉｏｎ模块加入ＣａｐｓＮｅｔ作为孪生网络的子网络来构建孪生网络ꎬＩｎｃｅｐｔｉｏｎ模块替代了ＳＣＮＮ中的卷积层ꎬ解决了ＣＮＮ中最大池化导致重要信息丢失的问题ꎮＳＭＳＤ￣ＣＮｅｔ与ＳＣａｐｓＮｅｔ的主要区别在于ꎬＳＣａｐｓＮｅｔ利用Ｉｎｃｅｐｔｉｏｎ模块ꎬ而ＳＭＳＤＣＮｅｔ利用空洞Ｉｎｃｅｐｔｉｏｎ模块ꎬ减少了网络训练参数ꎬ所以加速了网络收敛ꎮＳＣＮＮ:孪生卷积神经网络ꎻＳＣａｐｓＮｅｔ:孪生胶囊网络ꎻＳＭＳＤＣＮｅｔ:孪生多尺度空洞胶囊网络ꎮ图７㊀３个孪生网络的病害检测精度随迭代次数的变化Ｆｉｇ.７㊀ＤｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎｒａｔｅｓｏｆｔｈｒｅｅＳｉａｍｅｓｅｎｅｔｗｏｒｋｓｖｅｒ￣ｓｕｓｉｔｅｒａｔｉｏｎｔｉｍｅｓ㊀㊀为了证明ＳＭＳＤＣＮｅｔ在训练样本有限情况下的有效性ꎬ将ＳＭＳＤＣＮｅｔ与４种近期提出的作物病害识别方法[多尺度残差神经网络(ＭＳＲＮＮ)[１０]㊁改进卷积神经网络(ＩＣＮＮ)[２２]㊁ＶＧＧ轻量级卷积神经网络(ＶＧＧ￣ＩＣＮＮ)[１２]和轻量级卷积神经网络１３８１张善文等:基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法(ＬＷＣＮＮ)[１３]]进行对比ꎬ这５种方法在原始图像数据集上进行验证试验ꎬ训练样本数与测试样本数之比为ｍʒｎꎬ其中ｍ为每类病害的训练样本数ꎬｎ为每类病害的测试样本数ꎮ选择迭代次数为３０００次ꎬ重复试验５次ꎮ表１显示ꎬ当训练样本数越来越少时ꎬ５种方法的病害检测精度越来越低ꎬ训练样本数由９０降低为１０时ꎬＭＳＲＮＮ㊁ＩＣＮＮ㊁ＶＧＧ￣ＩＣＮＮ㊁ＬＷＣＮＮ和ＳＭＳＤＣＮｅｔ的病害检测精度分别降低２４ ９０个百分点㊁２１ ０４个百分点㊁３６ ５８个百分点㊁３８ ０６个百分点㊁８ ３５个百分点ꎬＳＭＳＤＣＮｅｔ的病害检测精度降低幅度最小ꎮ当ｍʒｎ＝１０ʒ９０ꎬ即训练样本数为１０且测试样本数为９０时ꎬＳＭＳＤＣＮｅｔ的病害检测精度为８２ ８９％ꎬ比ＭＳＲＮＮ㊁ＩＣＮＮ㊁ＶＧＧ￣ＩＣ￣ＮＮ和ＬＷＣＮＮ分别高４０ １０个百分点㊁３１ ２７个百分点㊁３６ ０１个百分点和３３ ８６个百分点ꎮ试验结果表明ꎬＳＭＳＤＣＮｅｔ能够在训练样本比较少的情况下达到较高的病害检测精度ꎬ这是因为ＳＭＳＤＣＮｅｔ通过计算２幅图像的相似度极大地减少网络参数的训练负担ꎬ对于训练样本数为ｍ的训练集可以组成[ｍ(ｍ－１)/２]个不同的训练样本对ꎬ由此增加训练样本的利用率ꎬ从而减少模型对训练样本量的依赖ꎮ表１㊀５种病害识别方法在原始数据集上的黄瓜病害检测精度Ｔａｂｌｅ１㊀Ｄｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎｒａｔｅｓｏｆｆｉｖｅｄｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎｍｅｔｈ￣ｏｄｓｏｎｔｈｅｏｒｉｇｉｎａｌｄａｔａｓｅｔｍʒｎ病害检测精度(％)ＭＳＲＮＮＩＣＮＮＶＧＧ￣ＩＣＮＮＬＷＣＮＮＳＭＳＤＣＮｅｔ９０ʒ１０６７.６９７２.６６８３.４６８７.０９９１.２４８０ʒ２０６７.１９７２.２２８２.１３８５.４１９０.５６７０ʒ３０６５.８３７１.６４８０.７８８４.０１９０.４２６０ʒ４０６１.９０６１.８３６１.８１７２.８３９０.３３５０ʒ５０５１.１６５６.９６５５.４８７１.７３８６.２０４０ʒ６０４９.９４５５.５１５４.６３７０.４３８５.７９３０ʒ７０４７.３４５３.４３５２.１７６８.５１８４.８９２０ʒ８０４３.４３５２.１３５１.５１６５.１６８３.４１１０ʒ９０４２.７９５１.６２４６.８８４９.０３８２.８９ｍ:训练样本数ꎻｎ:测试样本数ꎻＭＳＲＮＮ:多尺度残差神经网络ꎻＩＣ￣ＮＮ:改进卷积神经网络ꎻＶＧＧ￣ＩＣＮＮ:ＶＧＧ轻量级卷积神经网络ꎻＬＷＣＮＮ:轻量级卷积神经网络ꎻＳＭＳＤＣＮｅｔ:孪生多尺度空洞胶囊网络ꎮ㊀㊀为了验证ＳＭＳＤＣＮｅｔ的鲁棒性ꎬ通过改变ＳＭＳ￣ＤＣＮｅｔ中的不同模块ꎬ采用５折交叉验证方法进行消融试验ꎮ表２显示ꎬＩｎｃｅｐｔｉｏｎ模块明显优于传统胶囊网络中的卷积和ＶＧＧ１６的卷积ꎬ原因是Ｉｎｃｅｐ￣ｔｉｏｎ为多尺度卷积ꎻ空洞Ｉｎｃｅｐｔｉｏｎ的训练时间比Ｉｎ￣ｃｅｐｔｉｏｎ少ꎬ原因是它扩大了感受野但没有增加训练参数ꎻ余弦距离的效果略优于欧氏距离ꎬ其原因是余弦距离能够直接用于相似度度量ꎮ表２结果表明ꎬＳＭＳＤＣＮｅｔ具有一定的鲁棒性ꎬ原因是它利用了孪生网络㊁空洞卷积网络和胶囊网络的优势ꎮ表２㊀基于ＳＭＳＤＣＮｅｔ的消融试验结果Ｔａｂｌｅ２㊀ＲｅｓｕｌｔｓｏｆａｂｌａｔｉｏｎｅｘｐｅｒｉｍｅｎｔｂａｓｅｄｏｎＳＭＳＤＣＮｅｔ改变ＳＭＳＤＣＮｅｔ结构㊀㊀㊀㊀㊀检测精度(％)训练时间(ｈ)两个子网络都为传统的ＣＮｅｔ８１.２２１４.３１卷积模块的卷积子网络为ＶＧＧ１６８３.１６１３.１４卷积模块的卷积核为Ｉｎｃｅｐｔｉｏｎ模块８７.３５１１.８２相似度度量为欧氏距离８９.１０８.９２ＳＭＳＤＣＮｅｔ结构未改变对照９０.５６８.７７３㊀结论由于黄瓜病害叶片图像复杂多样ꎬ传统方法很难提取到鲁棒的图像分类特征ꎬ深度ＣＮＮ需要大量训练样本ꎬ本研究在空洞卷积网络㊁多尺度卷积网络㊁胶囊网络和孪生网络的基础上ꎬ提出了一种基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法ꎮ该方法通过对比样本相似度扩大训练次数ꎬ解决病害叶片图像不足的问题ꎬ采用胶囊网络提取特征的空间信息和位置信息ꎬ选用余弦距离相似度更好地区分２个胶囊向量的差异性ꎮ结果表明ꎬＳＭＳ￣ＤＣＮｅｔ具有良好的泛化性ꎬ该方法为黄瓜病害检测与识别的进一步研究提供了参考ꎮ下一步工作主要在于优化模型ꎬ在拥有更多病害类型的数据集上验证模型的有效性㊁可行性ꎮ参考文献:[１]㊀刁智华ꎬ袁万宾ꎬ刁春迎ꎬ等.病害特征在作物病害识别中的应用研究综述[Ｊ].江苏农业科学ꎬ２０１９ꎬ４７(５):７１￣７４. [２]㊀ＦＥＲＮÁＮＤＥＺＣＩꎬＬＥＢＬＯＮＢꎬＷＡＮＧＪꎬｅｔａｌ.Ｃｕｃｕｍｂｅｒｐｏｗ￣ｄｅｒｙｍｉｌｄｅｗｄｅｔｅｃｔｉｏｎｕｓｉｎｇｈｙｐｅｒｓｐｅｃｔｒａｌｄａｔａ[Ｊ].ＣａｎａｄｉａｎＪｏｕｒ￣ｎａｌｏｆＰｌａｎｔＳｃｉｅｎｃｅꎬ２０２２ꎬ１０２(１):２０￣３２.[３]㊀ＳＨＥＮＷＺꎬＧＵＡＮＹꎬＷＡＮＧＹꎬｅｔａｌ.Ｒｅｓｅａｒｃｈｏｎｒｉｃｅｌｅａｆｄｉｓ￣ｅａｓｅｒｅｃｏｇｎｉｔｉｏｎｂａｓｅｄｏｎＢＰｎｅｕｒａｌｎｅｔｗｏｒｋ[Ｊ].ＪｏｕｒｎａｌｏｆＮｏｒｔｈ￣ｅａｓｔＡｇｒｉｃｕｌｔｕｒａｌＵｎｉｖｅｒｓｉｔｙ(ＥｎｇｌｉｓｈＥｄｉｔｉｏｎ)ꎬ２０１９ꎬ２６(３):７５￣８６.２３８１江苏农业学报㊀２０２３年第３９卷第９期[４]㊀ＬＩＮＫꎬＧＯＮＧＬꎬＨＵＡＮＧＹꎬｅｔａｌ.Ｄｅｅｐｌｅａｒｎｉｎｇ￣ｂａｓｅｄｓｅｇｍｅｎ￣ｔａｔｉｏｎａｎｄｑｕａｎｔｉｆｉｃａｔｉｏｎｏｆｃｕｃｕｍｂｅｒｐｏｗｄｅｒｙｍｉｌｄｅｗｕｓｉｎｇｃｏｎｖｏ￣ｌｕｔｉｏｎａｌｎｅｕｒａｌｎｅｔｗｏｒｋ[Ｊ].ＦｒｏｎｔｉｅｒｓｉｎＰｌａｎｔＳｃｉｅｎｃｅꎬ２０１９.ＤＯＩ:１０.３３８９/ｆｐｌｓ.２０１９.００１５５.[５]㊀ＫＡＵＲＳꎬＰＡＮＤＥＹＳꎬＧＯＥＬＳ.Ｐｌａｎｔｓｄｉｓｅａｓｅｉｄｅｎｔｉｆｉｃａｔｉｏｎａｎｄｃｌａｓｓｉｆｉｃａｔｉｏｎｔｈｒｏｕｇｈｌｅａｆｉｍａｇｅｓ:ａｓｕｒｖｅｙ[Ｊ].ＡｒｃｈｉｖｅｓｏｆＣｏｍ￣ｐｕｔａｔｉｏｎａｌＭｅｔｈｏｄｓｉｎＥｎｇｉｎｅｅｒｉｎｇꎬ２０１９ꎬ２６(２):５０７￣５３０. [６]㊀邵明月ꎬ张建华ꎬ冯㊀全ꎬ等.深度学习在植物叶部病害检测与识别的研究进展[Ｊ].智慧农业(中英文)ꎬ２０２２ꎬ４(１):２９￣４６. [７]㊀ＮＧＵＧＩＬＣꎬＡＢＥＬＷＡＨＡＢＭꎬＡＢＯ￣ＺＡＨＨＡＤＭ.Ｒｅｃｅｎｔａｄ￣ｖａｎｃｅｓｉｎｉｍａｇｅｐｒｏｃｅｓｓｉｎｇｔｅｃｈｎｉｑｕｅｓｆｏｒａｕｔｏｍａｔｅｄｌｅａｆｐｅｓｔａｎｄｄｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎ￣ａｒｅｖｉｅｗ[Ｊ].ＩｎｆｏｒｍａｔｉｏｎＰｒｏｃｅｓｓｉｎｇｉｎＡｇｒｉ￣ｃｕｌｔｕｒｅꎬ２０２１ꎬ８(１):２７￣５１.[８]㊀岳有军ꎬ李雪松ꎬ赵㊀辉ꎬ等.基于改进ＶＧＧ网络的农作物病害图像识别[Ｊ].农机化研究ꎬ２０２２ꎬ４４(６):１８￣２４. [９]㊀ＹＵＨꎬＣＨＥＮＧＸꎬＬＩＺꎬｅｔａｌ.Ｄｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎｏｆａｐｐｌｅｌｅａｆｕｓｉｎｇｌｉｇｈｔｗｅｉｇｈｔｍｕｌｔｉ￣ｓｃａｌｅｎｅｔｗｏｒｋｗｉｔｈＥＣＡＮｅｔ[Ｊ].ＣｏｍｐｕｔｅｒＭｏｄｅｌｉｎｇｉｎＥｎｇｉｎｅｅｒｉｎｇ＆Ｓｃｉｅｎｃｅｓꎬ２０２２ꎬ１３２(３):７１１￣７３８. [１０]何㊀欣ꎬ李书琴ꎬ刘㊀斌.基于多尺度残差神经网络的葡萄叶片病害识别[Ｊ].计算机工程ꎬ２０２１ꎬ４７(５):２８５￣２９１ꎬ３００. [１１]ＣＨＥＮＪꎬＷＡＮＧＷꎬＺＨＡＮＧＤꎬｅｔａｌ.Ａｔｔｅｎｔｉｏｎｅｍｂｅｄｄｅｄｌｉｇｈｔ￣ｗｅｉｇｈｔｎｅｔｗｏｒｋｆｏｒｍａｉｚｅｄｉｓｅａｓｅｒｅｃｏｇｎｉｔｉｏｎ[Ｊ].ＰｌａｎｔＰａｔｈｏｌｏｇｙꎬ２０２１ꎬ７０(３):６３０￣６４２.[１２]ＴＨＡＫＵＲＰＳꎬＳＨＥＯＲＥＹＴꎬＯＪＨＡＡ.ＶＧＧ￣ＩＣＮＮ:ａｌｉｇｈｔ￣ｗｅｉｇｈｔＣＮＮｍｏｄｅｌｆｏｒｃｒｏｐｄｉｓｅａｓｅｉｄｅｎｔｉｆｉｃａｔｉｏｎ[Ｊ].ＭｕｌｔｉｍｅｄｉａＴｏｏｌｓａｎｄＡｐｐｌｉｃａｔｉｏｎｓꎬ２０２３ꎬ８２:４９７￣５２０.[１３]孟㊀亮ꎬ郭小燕ꎬ杜佳举ꎬ等.一种轻量级ＣＮＮ农作物病害图像识别模型[Ｊ].江苏农业学报ꎬ２０２１ꎬ３７(５):１１４３￣１１５０. [１４]ＦＡＮＸꎬＬＵＯＰꎬＭＵＹꎬｅｔａｌ.Ｌｅａｆｉｍａｇｅｂａｓｅｄｐｌａｎｔｄｉｓｅａｓｅｉ￣ｄｅｎｔｉｆｉｃａｔｉｏｎｕｓｉｎｇｔｒａｎｓｆｅｒｌｅａｒｎｉｎｇａｎｄｆｅａｔｕｒｅｆｕｓｉｏｎ[Ｊ].Ｃｏｍ￣ｐｕｔｅｒｓａｎｄＥｌｅｃｔｒｏｎｉｃｓｉｎＡｇｒｉｃｕｌｔｕｒｅꎬ２０２２ꎬ１９６.ＤＯＩ:１０.１０１６/ｊ.ｃｏｍｐａｇ.２０２２.１０６８９２.[１５]ＡＨＭＥＤＫＲ.Ｓｍａｒｔｐｏｔｈｏｌｅｄｅｔｅｃｔｉｏｎｕｓｉｎｇｄｅｅｐｌｅａｒｎｉｎｇｂａｓｅｄｏｎｄｉｌａｔｅｄｃｏｎｖｏｌｕｔｉｏｎ[Ｊ].Ｓｅｎｓｏｒｓꎬ２０２１ꎬ２１(２４):８４０６. [１６]ＦＡＮＧＹꎬＬＩＹꎬＴＵＸꎬｅｔａｌ.Ｆａｃｅｃｏｍｐｌｅｔｉｏｎｗｉｔｈｈｙｂｒｉｄｄｉｌａｔｅｄｃｏｎｖｏｌｕｔｉｏｎ[Ｊ].ＳｉｇｎａｌＰｒｏｃｅｓｓｉｎｇ:ＩｍａｇｅＣｏｍｍｕｎｉｃａｔｉｏｎꎬ２０２０ꎬ８０:１１５６６４.[１７]ＰＡＯＬＥＴＴＩＭＥꎬＨＡＵＴＪＭꎬＦＥＲＮＡＮＤＥＺ￣ＢＥＬＴＲＡＮＲꎬｅｔａｌ.Ｃａｐｓｕｌｅｎｅｔｗｏｒｋｓｆｏｒｈｙｐｅｒｓｐｅｃｔｒａｌｉｍａｇｅｃｌａｓｓｉｆｉｃａｔｉｏｎ[Ｊ].ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＧｅｏｓｃｉｅｎｃｅ＆ＲｅｍｏｔｅＳｅｎｓｉｎｇꎬ２０１９ꎬ５７(４):２１４５￣２１６０.[１８]ＪＡＮＡＫＩＲＡＭＡＩＡＨＢꎬＫＡＬＹＡＮＩＧꎬＰＲＡＳＡＤＬꎬｅｔａｌ.Ｉｎｔｅｌｌｉ￣ｇｅｎｔｓｙｓｔｅｍｆｏｒｌｅａｆｄｉｓｅａｓｅｄｅｔｅｃｔｉｏｎｕｓｉｎｇｃａｐｓｕｌｅｎｅｔｗｏｒｋｓｆｏｒｈｏｒｔｉｃｕｌｔｕｒｅ[Ｊ].ＪｏｕｒｎａｌｏｆＩｎｔｅｌｌｉｇｅｎｔ＆ＦｕｚｚｙＳｙｓｔｅｍｓ:Ａｐｐｌｉｃａ￣ｔｉｏｎｓｉｎＥｎｇｉｎｅｅｒｉｎｇａｎｄＴｅｃｈｎｏｌｏｇｙꎬ２０２１ꎬ４１(６):６６９７￣６７１３. [１９]钱心筠ꎬ王友仁ꎬ赵亚磊.基于孪生网络的行星齿轮箱故障诊断方法[Ｊ].机械制造与自动化ꎬ２０２２ꎬ５１(４):１１６￣１１９. [２０]ＣＨＩＣＣＯＤ.Ｓｉａｍｅｓｅｎｅｕｒａｌｎｅｔｗｏｒｋｓ:ａｎｏｖｅｒｖｉｅｗ[Ｊ].ＡｒｔｉｆｉｃｉａｌＮｅｕｒａｌＮｅｔｗｏｒｋｓꎬ２０２０.ＤＯＩ:１０.１００７/９７８￣１￣０７１６￣０８２６￣５＿３. [２１]ＺＨＯＵＳꎬＺＨＯＵＹꎬＬＩＵＢ.ＵｓｉｎｇＳｉａｍｅｓｅｃａｐｓｕｌｅｎｅｔｗｏｒｋｓｆｏｒｒｅｍｏｔｅｓｅｎｓｉｎｇｓｃｅｎｅｃｌａｓｓｉｆｉｃａｔｉｏｎ[Ｊ].ＲｅｍｏｔｅＳｅｎｓｉｎｇＬｅｔｔｅｒｓꎬ２０２０ꎬ１１(８):７５７￣７６６.[２２]鲍文霞ꎬ黄雪峰ꎬ胡根生ꎬ等.基于改进卷积神经网络模型的玉米叶部病害识别[Ｊ].农业工程学报ꎬ２０２１ꎬ３７(６):１６０￣１６７.(责任编辑:王㊀妮)３３８１张善文等:基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法。
博士学位论文-中国科学技术大学
Studies on the dynamical properties of traffic processes on
complex networks
Author’s Name：
Xiang LBiblioteka ngspeciality：
Thermophysics Engineering
Supervisor： Prof.QingSong Wu, Assoc.Prof.MaoBin Hu
系统的研究了节点容量有限时，无标度网络、小世界网络和规则网络上出现 的交通迟滞现象。我们发现在使用局部路由策略时，小世界网络和规则网络中出
I
摘要
现迟滞现象的原因是由于节点处理能力有限，但在无标度网络中出现迟滞现象和 节点处理能力没有关系。在使用全局路由策略时，无标度网络、小世界网络和规 则网络上出现的迟滞现象和节点处理能力没有关系，即使节点处理能力无限大， 迟滞现象同样会出现。
The purpose of studying complex networks is to understand the function of the real network system. The characteristics of traffic dynamic processes taking place on networks is a focus of network study. In this paper, we have systematically investigated the dynamics of information traffic on complex networks, including routing strategies, hysteresis phenomena, traffic resource allocation and the optimization routing under the condition of limited resources. The main work for this paper is as follows.
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
September 2011, 18(Suppl. 1): 48–52/science/journal/10058885 The Journal of China Universities of Posts and TelecommunicationsDynamic routing network algorithm of low voltagebased on link qualityLI Jie ( ), ZHOU Ming, SONG Jun-deSchool of Computer Science, Beijing University of Posts and Telecommunications, Beijing 100876, ChinaAbstractPower distribution network is a dedicated transmission network for power. If you want to use it as communication network, the network traffic will become very complicated. Low voltage distribution network topology is very complex and quite different in different distribution network; and different load leave or input power line will cause a strong performance of channel variability. It is because of these features of low voltage distribution network that it difficult to networking and the network is easily destroyed. This paper presents a dynamic routing algorithm, based on link quality, to improve the reliability of low voltage distribution networks. The algorithm can eventually form an 8-layer structure of the tree topology, and can achieve network self-healing capabilities when the link is broken. The algorithm is suitable for a variety of network physical topology, and finally be able to form a tree structure, thus effectively improving the reliability and system flexibility of communication network. Through the simulation on NS2, we can see the feasibility and network self-healing of the algorithm.Keywords PLC network, self-healing, link quality1 IntroductionCommunication technology is becoming more and more widely in the field of distribution network. Through the PLC (power line communication) technology, we can run the distribution network for monitoring and associated control, such as: we can use it to control street lighting, automatic meter reading and remote monitoring, etc. This shows that power line network communication has very broad prospect [1–4]. The traditional methods of improving the reliability of PLC networks include increasing the transmission power, optimizing the coupling devices. Increasing the transmission power should not be used, mainly because of its development in the opposite direction with PLC. And we can optimize the coupling device to prevent the channel attenuation, but impedance characteristics of the low power voltage distribution networks are time-varying [5]. Through a large number of experiments in PLC, the dynamic routing algorithm to select the relay nodes methods, in order toReceived date: 25-07-2011Corresponding author: LI Jie, E-mail: jli@ DOI: 10.1016/S1005-8885(10)60211-7improve the reliability of power line communication, is feasible measures. Relay nodes and common nodes are identical, and the relay node is only added dynamic routing in software, which can adapt to the characteristics of changes in network [6–7]. In view of this, a dynamic group routing algorithm [8–10] to improve the communication reliability is proposed. Firstly, the node who wants to join the network broadcast a packet to get the link quality around it, and then select the best signal quality as its parent node in the network, notify the NC node of join information. At last, the NC node will assigned a node ID and network ID to the node who wants to join in the network.2 Protocol descriptionThe resulting network topology is tree structure, which includes the NC node and RS (remote station) [11–13].The network has a total of eight levels, in which 0 level has only one node (concentrator NC), the other levels can have 256 nodes and their child nodes, but the 7th level nodes can not have child nodes. NC node functions as a network concentrator, which is the root of the tree, since the launch ofSupplement 1 LI Jie, et al. / Dynamic routing network algorithm of low voltage based on link quality 49the NC node a logical subnet network ID will be assigned. The entire network topology is developed from the NC node. Except the NC node, the others are RS nodes.2.1 Address serviceEach node module has a MAC address provided by the manufacturer, which is a 6-byte code. However, you can not use this 6-byte code for data communications network. The MAC address is used when applying the node ID in the network. Therefore, each node in the network must have a network ID and node ID, which are obtained in the course of networking, and not be set by man [14–16] .1) Network ID service:Each logical subnet has a network ID, the network ID assigned to the NC node firstly. If a new node joins the logical subnet, the NC node will assigned the network ID to the corresponding child nodes. A new network can maintain the uniqueness of different logical subnet by choosing the unique network address.2) Node ID ServiceA node to node by node and node ID makes the link between the phase node by node ID to carry out normal communication. Node ID format is 4B of code, the first two bytes of the logical subnet of the node in which a layer, the last two bytes identify the node, the first few nodes in the layer. For example, if a node ID is 0X0102, it means that the nodes in the network's first layer of the second node.2.2 Joining of NodeThe process of the network topology formation is the process that all nodes join the network, then a tree network topology, the root of which is NC node, is formed. A network ID will be assigned to the NC node When it turns on, and the own node ID is set to 0X0000. If a node applies to join the logical subnet, the node will broadcast a link signal quality of the data packets SQP (Signal Quality Packet) to the nodes that can communicate directly, format of the packet is shown in Table 2. The node can get the signal quality between the nodes that can communicate directly. SQP packet contains a Node Filter parameter and a responsive type parameter. The new join node’s parent node can’t at the 7 level, because the top level of the network is 7.In the SQP packet, the value of Node Filter is set to 0x06FF and the response value is set to 1, which means the nodes in level 0 to 6 can respond to all nodes. The specific meaning of Node Filter parameter is shown in Table 1:Table 1 Channel quality tableValue ofNodeFilterValue of response type is 1 Value of response type is 10X0001Only the concentrator isallowed to answer the joinrequest.Only the concentrator isallowed to answer the joinrequest0X01FFOnly the concentrator and allnodes on level 1 are allowedto answer the join requestall nodes on level 1 areallowed to answer the joinrequest0X03FFOnly the concentrator and allnodes on level 1-3 areallowed to answer the joinrequestall nodes on level 3 areallowed to answer the joinrequest0X030AOnly the concentrator andthe nodes with the id 0x01 –0x0A on the levels 1-3 areallowed to answer the joinrequestthe nodes with the id 0x01–0x0A on the levels 3 areallowed to answer the joinrequest0X06FFOnly the concentrator and allnodes on level 1-6 areallowed to answer the joinrequestall nodes on level 6 areallowed to answer the joinrequestThe nodes who receive SQP data packets, through the detecting of node link quality between nodes, respectively, return link quality data packet to the nodes that want to join networks by RSQP. The format of RSQP is shown in Table 3. After the node obtain the link quality with its directly communicating nodes, select the best link quality nodes as its parent node, and establish the route to NC node, from which we can determine the parent nodes of new joining node, and sends a notification packet JNP1 to its parents. The format of JNP1 is shown in Table 4.When the parent nodes receive the join packet, it saves the MAC address of the new joining node. If the parent node is NC node, it will sent node id to the new joining node directly through SNP packet, adding the new ID to the tree network topology. Otherwise, the new joining node will send data packet (Get Node ID packet, GNP) to NC node in order to get the node ID, the format of GNP packet is shown in Table 5,which inform the NC node a new node want to join the network. The NC node determine the new joining node belongs to which level by judging the ID of sending node, and assigned a node ID to the new joining node. When all the relay nodes receive the GNP packet, they temporarily store the sending node ID, which is to establish the tree topology of relay nodes.After the NC node receives the request, it sends a node-assigned id packet (Assign Node ID packet, ANP) to50 The Journal of China Universities of Posts and Telecommunications 2011parent node, the format of ANP packet is shown in Table 6. Before the ANP packet arrive the node of sending GNP, all the relay nodes would establish a route to the new joining node, and add the new joining node ID as the temporarily-stored node’s child node, in order to establish the new corresponding tree topology. When the ANP packet arrive at the node of sending GNP packet (destination node) at last, the destination node will establish a route to the new joining node, then send the node ID to the child node through SNP packet. At this point, the new joining node has obtained its node ID. The format of SNP packet is shown in Table 7.Table 2 Format of SQPType Mac address of source node Data service type NodeFilter Response type Checksum 01⩒broadcast⩓Table 3 Format of RSQPType ⩒1⩓Mac address of source node Mac address of destination Value of link quality checksumTable 4 Format of JNP1Type⩒2⩓Mac address of source node Mac address of destination ChecksumTable 5 Format of GNPType⩒3⩓Source node ID Next hop ID Destination ID metricTable 6 Format of ANPType⩒4⩓Source of node id Next hop ID Destination ID Metric The new ID assigned to each levelTable 7 Format of SNPType ⩒5⩓Mac address of source node Mac address of destination Child node ID ChecksumThe new network tree topology is built through the above algorithm at last, and the new joining node obtains the relevant node ID.3 The function of self-healingBecause the channel of low voltage distribution network is time-varying, easily to break a link, an algorithm which has the function of self-healing is needed to reestablish the network topology. For example, when two nodes communicate with each other, after the sending node sends a data packet to the destination node, it will wait for a confirmation from the destination node. If the sending node receive confirmation, we are sure the link is not broken; if not receive the confirmation, then re-sent the data packet two times. However, if the sending node does not receive confirmation of the next hop node within three times, the link is considered to have been broken, and need to re-establish the route. If the broken link is shown in fig.8, the node 0X0201 will send a link break off packet (LBP) to NC node, the NC node and all the relay nodes will remove the unreachable nodes after they receive LBP packet. Then, node 0X0302 and node 0x0301 call the algorithm described in the previous section, and re-apply to join the network. The format of LBP packet is shown in Fig. 1.The other broken link situation is shown in Fig. 2. Node 0X0101 and node 0X0201 can’t communicate with each other, then node 0X0101 sends LBP packet to NC node. All the nodes have received LBP packet will remove the tree topology whose root node ID is 0X0101. Subsequently the tree topology whose root ID’s 0X0201 will re-join in the network and node 0X0201 will send a SQP packet to the up level. Because the node 0X0201 is on level 2, the Node Filter can be set to 0X01FF and response type is set to 0, which indicate only the first level can response the join request. The nodes who receive SQP packet will response RSQP packet. After the node 0X0201 receive RSQP packet, it will compare the value of link quality, and choose the node that is the source of best link as its parent node, establishing a route to NC node.The tree network topology whose root ID is 0x0201 does not need to re-apply the node ID from the NC node, and can use their original node ID, because the level where they be on have no other nodes join or leave, and if re-apply the node ID, the resulting node ID is also the same as their original ID. The node 0X0201 send a JNP2 (join notification packet) to NC node, and JNP2 packet contains the level of the sub-tree network topology and the nodes ID on this level. The format of JNP2 is shown in Fig. 11. After relay node and NC node receive the JNP2 packet, they will add the new joining node to their sub-tree network topology, and establish the route to new joining nodes, thus completing the self-healing network process.Supplement 1 LI Jie, et al. / Dynamic routing network algorithm of low voltage based on linkquality 51Fig. 1The situation of link broken 1Fig. 2 The situation of link broken 2Table 8 Format of LBPTypeSource node IDUnreachable node IDlifetime7Table 9 Format of JNP2TypeSource node IDNext hop IDDestination IDlevelNodes id of each level8NC node4 SimulationWe select the NS-2 simulation platform. The simulationobject, a physical structure, is shown in Fig. 3. Assuming that node 0 is the master plans, also called NC node, the nodes can directly communicate with it are ^`1,61,7,33,22; the nodes can directly communicate with node 61 are ^`10,20; the nodes can directly communicate with node 33 are ^`21,4,6,the nodes can directly communicate with node 7 are ^`33,21;the nodes can directly communicate with node 21 are ^`4,62;the nodes can directly communicate with node 22 are ^`32,57. The adjacent nodes can directly communicate with each other. We assume that the signal quality is inversely proportional to the distance between nodes. The simulation shows the resulting tree topology in Fig. 4.In the network topology, NC node 0 saved the entire network topology, and all the RS nodes saved the route to NC node. We can see each node is assigned the node ID and network ID from Fig. 5. Because all the nodes are in the samelogical network, they have the same network ID.Fig. 3 The physical network structureIt is assumed that the power line have more load, strong noise and channel attenuation during the evening peak, which result in number 10 node and number 20 node cannot communicate with each other, so number 20 node can’t communicate with NC node, and re-initiate the process in order to re-join the network. The number 20 node sends routing request packet by multicast, who receive the routing request packet will give the response. Suppose the relay node52 The Journal of China Universities of Posts and Telecommunications 201121 is damaged, leading the number 4 node can’t communicate with the NC node, then number 4 node re-multicast route request packet. Through the simulation in NS2 platform, the resulting network topology after self-healing is shown inFig.5.Fig. 4The resulting tree network topologyFig. 5 Tree topology after self-healing5 ConclusionsIn the limited communication capabilities circumstances of PLC, in order to improve the reliability and stability of PLC, the routing in the upper layer can improve the various properties. This paper describes the dynamic routing network algorithm based on link quality in detail. The simulation shows that the algorithm can eventually form an 8-layer structure of the tree topology, and can achieve network self-healing capabilities when the link is broken. The algorithm is suitable for a variety of network physical topology, and finally be able to form a tree structure, thus effectively improving the reliability and system flexibility of communication network.AcknowledgementsThis work was supported by the National Key project of Scientific and Technical Supporting Programs of China (2008BAH24B04, 2008BAH21B03, 2009BAH39B03); The National Natural Science Foundation of China (61072060); The Program for New Century Excellent Talents in University (NECET-08-0738); Engineering Research Center of Information Networks, Ministry of Education.References1.Xiaojun Tang. Development of ower line carrier communication during “tenone five”. Electric Power System, 2006, 27(10): 75-792.Wei Xu, Zhi Li, An adaptive routing algorithm for power linecommunication. Information and Electronic Engineering, 2010, (6): 738-741 3.Shengli Li, Shaohua Jiao, Lijun Qin. The research of power linecommunication. Electrical Measurement & Instrumentation, 2003, 5(11): 29-334.CavdarI HˊPerformance analysis of FSK power line communicationssystems over the time-varying channels: measurements and modeling. IEEE Transactions on Power Delivery, 2004, 19(1): 111-1175.Morosi S, Marabissi D, Enrico Del Re, et al. A rate adaptive bit-loadingalgorithm for in-building power-line communications based on DMT-modulated systems. IEEE Transactions on Power Delivery, 2006, 21(4): 1892-18976.Wenbing Lu, Xiufan Ma, Yingchun You. Design of digital carrier meterreading system and analysis of its communication performance. Electric Power Automation Equipment, 2003, 23(11): 23-267.Ricard Aquilue, Arc Deumal, Oan Llus Pijoan, Laura Corbeira. A LowComplexity Multicarrier Proposal for Medium Rate Demanding Automatic Meter Reading Systems. IEEE, 2007, 11-168.Sitong Wang, Ruiming Wang, Zhijie Sun. The technology of low voltagepower line carrier and its implementation in automated meter reading system. Electrical Measurement & Instrumentation, 2008(03)9.Weiqing Tao, Xiong chen. Realization of Low Voltage Carrier AutomaticMeter Reading System Based on Intelligent Relay Technology. Electrical Measurement & Instrumentation, 2009(05)10.Chaoqun Xu, Yuming Gao. Application of Low-voltage power linecommunication in a city street lamp long-distance intelligent monitoring system. Telecommunication Engineering, 2006(06)11.Xiaosheng Liu, Jiajin Qi, Qitao Song, etc. Method of Constructing PowerLine Communication Networks Over Low-voltage Distribution Networks Based on Ant Colony Optimization. Proceedings of the CSEE, 2008, 28(l): 71-7612.Tianbai Yu, Jiangxin Yang, Yuxi Zhao. A routing algorithm of edificecontrol system based on power line signal carrier. Power System Technology, 2006, 30(9): 88-9113.Zhang Subing, Liu Zemin. A QoS routing algorithm based on antalgorithm.25th, Annual IEEE Conference on Computer Network (LCN2000): TamPa, FL, USA 2000: 574-57814.Dorigo M, Maniezo Vittorio, Colomi Alberto. The Ant System:Optimization by A colony of cooperating agents. IEEE Transactions on Systems Man and Cybernetes-PartB, 1996, 26(l): l-1315.Dorigo M, Gambardella L M, Ant Colony system. A cooperative Leamintapproach to the traveling sales man problem. IEEE Transaetions on evolutionary Computation, 1997, 31(l): 53-6616.Maekawa T, Tada H, Wakamiya N. An Ant-based Routing ProtocolusingUnidirectional Links for Heterogeneous Mobile Ad-hoc Networks.Internation Conference on Wirelessand Mobile Communieations (ICWMC’06). IEEE Computer Soeiety Press (Roumania), 2006。