8Research on Clustering Routing Algorithms in Wireless Sensor Networks

合集下载

数据挖掘中聚类算法研究进展_周涛

一书中，即 “物以类聚，人以群分” ，聚类这个古老的问题，它伴随着人类社会的产生和发展而不断深化，人类要认识世界就必须区分不同的事物并认识事物间的相似性。数据挖掘的重要任务之一就是发现大型数据中的积聚现象，并加以定量化描述。聚类分
基金项目：国家自然科学基金（No.81160183）；宁夏自然科学基金（11105）；陕西省教育厅项目（No.2010JK466）；宁夏卫生厅重点科研项目（No.2011033）；宁夏高等学校科学研究重点项目（宁教高 [2011]263 号）；宁夏医科大学特殊人才项目（No.XT2011004）；宁夏医科大学青年基金项目（No.XQ2011011）。作者简介：周涛（1977—），男，回族，博士，副教授，硕士生导师，主要研究方向为医学图像处理、数据挖掘、软计算理论等；陆惠玲（1976—），女，讲师，主要研究方向为数据挖掘、医学图像处理。收稿日期： 2011-10-18 修回日期： 2011-12-21 DOI： 10.3778/j.issn.1002-8331.2012.12.021
[6] 提出 k- 模（k-modes）方法，它扩展了 k- 平均方法，用
则矩阵 μ = ( μij) 具有如下性质：
μij Î{0 1} 且 å μij = 1 ( j = 1 2 n)
i=1 c
设 ni 表示第 i 类中所包含的样本个数，则
ni = å μij (i = 1 2 c) 设 xi Î ÂN 表示第 i 类的中心，则 xi = μij x j å j=1 μij å j=1
[8] Application based upon Randomized Search）算法

工业机器人的路径规划算法考核试卷

15.以下哪种算法在路径规划中通常用于处理动态环境？（）
A. A*算法
B. D*算法
C. Floyd算法
D. Bresenham算法
16.在路径规划中，以下哪种数据结构用于存储已访问的节点？（）
A.开放列表
B.关闭列表
C.路径列表
D.邻接矩阵
17.关于工业机器人的路径规划，以下哪个描述是错误的？（）
B. Dijkstra算法
C. D*算法
D. RRT算法
13.关于路径规划中的碰撞检测，以下哪种方法计算量相对较小？（）
A.精确碰撞检测
B.粗略碰撞检测
C.迭代最近点法（ICP）
D.点到线段距离计算
14.以下哪项不是路径平滑处理的目的？（）
A.降低路径长度
B.减少运动时间
C.减少能量消耗
D.增加路径上的拐点
3.用于评估路径规划算法性能的指标通常包括路径长度、规划时间和______。
4.在路径规划中，为了减少计算量，常用的空间划分技术有四叉树和______。
5.机器人路径规划中的碰撞检测可以通过______和基于几何的碰撞检测两种方法实现。
6.路径平滑处理的目的是为了减少路径的拐点，提高路径的______和可执行性。
A. A*算法
B. Dijkstra算法
C. Floyd算法
D. Bresenham算法
2.下列哪种算法不属于路径规划中的启发式搜索算法？（）
A. A*算法
B. D*算法
C. IDA*算法
D. Breadth First Search算法
3.在A*算法中，H(n)代表什么？（）
A.从起始点到当前点的代价
4.在工业机器人路径规划中，如何处理动态障碍物？请提出一种算法或策略，并解释其工作原理和有效性。

聚类分析文献英文翻译

电气信息工程学院外文翻译英文名称：Data mining-clustering译文名称：数据挖掘—聚类分析专业：自动化姓名：****班级学号：****指导教师：******译文出处：Data mining：Ian H.Witten, EibeFrank 著二○一○年四月二十六日Clustering5.1 INTRODUCTIONClustering is similar to classification in that data are grouped. However, unlike classification, the groups are not predefined. Instead, the grouping is accomplished by finding similarities between data according to characteristics found in the actual data. The groups are called clusters. Some authors view clustering as a special type of classification. In this text, however, we follow a more conventional view in that the two are different. Many definitions for clusters have been proposed:●Set of like elements. Elements from different clusters are not alike.●The distance between points in a cluster is less than the distance betweena point in the cluster and any point outside it.A term similar to clustering is database segmentation, where like tuple (record) in a database are grouped together. This is done to partition or segment the database into components that then give the user a more general view of the data. In this case text, we do not differentiate between segmentation and clustering. A simple example of clustering is found in Example 5.1. This example illustrates the fact that that determining how to do the clustering is not straightforward.As illustrated in Figure 5.1, a given set of data may be clustered on different attributes. Here a group of homes in a geographic area is shown. The first floor type of clustering is based on the location of the home. Homes that are geographically close to each other are clustered together. In the second clustering, homes are grouped based on the size of the house.Clustering has been used in many application domains, including biology, medicine, anthropology, marketing, and economics. Clustering applications include plant and animal classification, disease classification, image processing, pattern recognition, and document retrieval. One of the first domains in which clustering was used was biological taxonomy. Recent uses include examining Web log data to detect usage patterns.When clustering is applied to a real-world database, many interesting problems occur:●Outlier handling is difficult. Here the elements do not naturally fallinto any cluster. They can be viewed as solitary clusters. However, if aclustering algorithm attempts to find larger clusters, these outliers will beforced to be placed in some cluster. This process may result in the creationof poor clusters by combining two existing clusters and leaving the outlier in its own cluster.● Dynamic data in the database implies that cluster membership may change over time.● Interpreting the semantic meaning of each cluster may be difficult. With classification, the labeling of the classes is known ahead of time. However, with clustering, this may not be the case. Thus, when the clustering process finishes creating a set of clusters, the exact meaning of each cluster may not be obvious. Here is where a domain expert is needed to assign a label or interpretation for each cluster.● There is no one correct answer to a clustering problem. In fact, many answers may be found. The exact number of clusters required is not easy to determine. Again, a domain expert may be required. For example, suppose we have a set of data about plants that have been collected during a field trip. Without any prior knowledge of plant classification, if we attempt to divide this set of data into similar groupings, it would not be clear how many groups should be created.● Another related issue is what data should be used of clustering. Unlike learning during a classification process, where there is some a priori knowledge concerning what the attributes of each classification should be, in clustering we have no supervised learning to aid the process. Indeed, clustering can be viewed as similar to unsupervised learning.We can then summarize some basic features of clustering (as opposed to classification):● The (best) number of clusters is not known.● There may not be any a priori knowledge concerning the clusters.● Cluster results are dynamic.The clustering problem is stated as shown in Definition 5.1. Here we assume that the number of clusters to be created is an input value, k. The actual content (and interpretation) of each cluster,j k ,1j k ≤≤, is determined as a result of the function definition. Without loss of generality, we will view that the result of solving a clustering problem is that a set of clusters is created: K={12,,...,k k k k }.D EFINITION 5.1.Given a database D ={12,,...,n t t t } of tuples and an integer value k , the clustering problem is to define a mapping f : {1,...,}D k → where each i t is assigned to one cluster j K ,1j k ≤≤. A cluster j K , contains precisely those tuples mapped to it; that is, j K ={|(),1,i i j t f t K i n =≤≤and i t D ∈}.A classification of the different types of clustering algorithms is shown in Figure 5.2. Clustering algorithms themselves may be viewed as hierarchical or partitional. With hierarchical clustering, a nested set of clusters is created. Each level in the hierarchy has a separate set of clusters. At the lowest level, each item is in its own unique cluster. At the highest level, all items belong to the same cluster. With hierarchical clustering, the desired number of clusters is not input. With partitional clustering, the algorithm creates only one set of clusters. These approaches use the desired number of clusters to drive how the final set is created. Traditional clustering algorithms tend to be targeted to small numeric database that fit into memory .There are, however, more recent clustering algorithms that look at categorical data and are targeted to larger, perhaps dynamic, databases. Algorithms targeted to larger databases may adapt to memory constraints by either sampling the database or using data structures, which can be compressed or pruned to fit into memory regardless of the size of the database. Clustering algorithms may also differ based on whether they produce overlapping or nonoverlapping clusters. Even though we consider only nonoverlapping clusters, it is possible to place an item in multiple clusters. In turn, nonoverlapping clusters can be viewed as extrinsic or intrinsic. Extrinsic techniques use labeling of the items to assist in the classification process. These algorithms are the traditional classification supervised learning algorithms in which a special input training set is used. Intrinsic algorithms do not use any a priori category labels, but depend only on the adjacency matrix containing the distance between objects. All algorithms we examine in this chapter fall into the intrinsic class.The types of clustering algorithms can be furthered classified based on the implementation technique used. Hierarchical algorithms can becategorized as agglomerative or divisive. ”Agglomerative ” implies that the clusters are created in a bottom-up fashion, while divisive algorithms work in a top-down fashion. Although both hierarchical and partitional algorithms could be described using the agglomerative vs. divisive label, it typically is more associated with hierarchical algorithms. Another descriptive tag indicates whether each individual element is handled one by one, serial (sometimes called incremental), or whether all items are examined together, simultaneous. If a specific tuple is viewed as having attribute values for all attributes in the schema, then clustering algorithms could differ as to how the attribute values are examined. As is usually done with decision tree classification techniques, some algorithms examine attribute values one at a time, monothetic. Polythetic algorithms consider all attribute values at one time. Finally, clustering algorithms can be labeled base on the mathematical formulation given to the algorithm: graph theoretic or matrix algebra. In this chapter we generally use the graph approach and describe the input to the clustering algorithm as an adjacency matrix labeled with distance measure.We discuss many clustering algorithms in the following sections. This is only a representative subset of the many algorithms that have been proposed in the literature. Before looking at these algorithms, we first examine possible similarity measures and examine the impact of outliers.5.2 SIMILARITY AND DISTANCE MEASURESThere are many desirable properties for the clusters created by a solution to a specific clustering problem. The most important one is that a tuple within one cluster is more like tuples within that cluster than it is similar to tuples outside it. As with classification, then, we assume the definition of a similarity measure, sim(,i l t t ), defined between any two tuples, ,i l t t D . This provides a more strict and alternative clustering definition, as found in Definition 5.2. Unless otherwise stated, we use the first definition rather than the second. Keep in mind that the similarity relationship stated within the second definition is a desirable, although not always obtainable, property.A distance measure, dis(,i j t t ), as opposed to similarity, is often used inclustering. The clustering problem then has the desirable property that given a cluster,j K ,,jl jm j t t K ∀∈ and ,(,)(,)i j jl jm jl i t K sim t t dis t t ∉≤.Some clustering algorithms look only at numeric data, usually assuming metric data points. Metric attributes satisfy the triangular inequality. The cluster can then be described by using several characteristic values. Given a cluster, m K of N points { 12,,...,m m mN t t t }, we make the following definitions [ZRL96]:Here the centroid is the “middle ” of the cluster; it need not be an actual point in the cluster. Some clustering algorithms alternatively assume that the cluster is represented by one centrally located object in the cluster called a medoid . The radius is the square root of the average mean squared distance from any point in the cluster to the centroid, and of points in the cluster. We use the notation m M to indicate the medoid for cluster m K .Many clustering algorithms require that the distance between clusters (rather than elements) be determined. This is not an easy task given that there are many interpretations for distance between clusters. Given clusters i K and j K , there are several standard alternatives to calculate the distance between clusters. A representative list is:● Single link : Smallest distance between an element in onecluster and an element in the other. We thus havedis(,i j K K )=min((,))il jm il i j dis t t t K K ∀∈∉and jm j i t K K ∀∈∉.● Complete link : Largest distance between an element in onecluster and an element in the other. We thus havedis(,i j K K )=max((,))il jm il i j dis t t t K K ∀∈∉and jm j i t K K ∀∈∉.● Average : Average distance between an element in onecluster and an element in the other. We thus havedis(,i j K K )=((,))il jm il i j mean dis t t t K K ∀∈∉and jm j i t K K ∀∈∉.● Centroid : If cluster have a representative centroid, then thecentroid distance is defined as the distance between the centroids.We thus have dis(,i j K K )=dis(,i j C C ), where i C is the centroidfor i K and similarly for j C .Medoid : Using a medoid to represent each cluster, thedistance between the clusters can be defined by the distancebetween the medoids: dis(,i j K K )=(,)i j dis M M5.3 OUTLIERSAs mentioned earlier, outliers are sample points with values much different from those of the remaining set of data. Outliers may represent errors in the data (perhaps a malfunctioning sensor recorded an incorrect data value) or could be correct data values that are simply much different from the remaining data. A person who is 2.5 meters tall is much taller than most people. In analyzing the height of individuals, this value probably would be viewed as an outlier.Some clustering techniques do not perform well with the presence of outliers. This problem is illustrated in Figure 5.3. Here if three clusters are found (solid line), the outlier will occur in a cluster by itself. However, if two clusters are found (dashed line), the two (obviously) different sets of data will be placed in one cluster because they are closer together than the outlier. This problem is complicated by the fact that many clustering algorithms actually have as input the number of desired clusters to be found.Clustering algorithms may actually find and remove outliers to ensure that they perform better. However, care must be taken in actually removing outliers. For example, suppose that the data mining problem is to predict flooding. Extremely high water level values occur very infrequently, and when compared with the normal water level values may seem to be outliers. However, removing these values may not allow the data mining algorithms to work effectively because there would be no data that showed that floods ever actually occurred.Outlier detection, or outlier mining, is the process of identifying outliers in a set of data. Clustering, or other data mining, algorithms may then choose to remove or treat these values differently. Some outlier detection techniques are based on statistical techniques. These usually assume that the set of data follows a known distribution and that outliers can be detected by well-known tests such as discordancy tests. However, thesetests are not very realistic for real-world data because real-world data values may not follow well-defined data distributions. Also, most of these tests assume single attribute value, and many attributes are involved in real-world datasets. Alternative detection techniques may be based on distance measures.聚类分析5.1简介聚类分析与分类数据分组类似。

5G网络高倒流问题优化研究

I G I T C W技术研究Technology Study44DIGITCW2023.10从当前的网络发展形势来看，5G/LTE （长期演进）双网共存并存仍然是通信的主流趋势。

从用户角度出发，由于5G 网络的带宽大、网速快，使用感知更好。

倘若网络优化不完善、参数设置不合理等因素，导致大量5G 终端用户仍然驻留在4G 网络上，造成4G 网络负荷高、5G 网络空闲的高倒流现象，无法将5G 网络资源有效转换成为收入。

结合实际案例分析，我们摸索出了一套通过配置无线侧网络参数来解决5G 网络高倒流问题的策略和方法，提升5G 用户驻留比，提高5G 网络投资效益。

1 5G网络高倒流原因分析5G 网络高倒流小区定义：在4G 网络和5G 网络共同覆盖的区域，当该区域内的5G 用户产生的4G 流量占比大于30%，即定义为高倒流小区。

由于目前5G 网络处于建设期，网络覆盖范围相对于4G 网络存在一定的差距，其中一部分由于参数设置不合理，导致5G 网络的业务会回落至4G 网络，造成5G 流量倒流，5G 基站和4G 基站的负荷不均衡[1]。

1.1 5G网络覆盖分析运营商在5G 网络工程建设当中，为充分利用现有4G 杆塔资源进行建设，同时为降低投资，4G 网络与5G 网络基本都是共址建设。

前期在4G 网络建设过程中，为保证网络覆盖能力，天线多是占用最优势的点位，而5G 网络建设时只能依据现有天面剩余资源进行建设。

因此5G 天线安装时往往无法使用最佳天面位置，5G 网络覆盖能力与4G 网络相比存在较大差距，造成5G 网络深度覆盖能力弱于4G 网络。

1.2 无线参数分析4G/5G 的无线参数对5G 小区高倒流也存在影响，4G/5G 的互操作参数的配置策略会收缩5G 网络真实的覆盖范围。

包括空闲态的5G 用户驻留策略（4G/5G 网络间的小区重选流程）和连接态的驻留策略（4G/5G网络间的异系统切换与覆盖重定向流程）[2]。

技能认证华为5G中级考试(习题卷2)

技能认证华为5G中级考试(习题卷2)说明：答案和解析在试卷最后第1部分：单项选择题，共50题，每题只有一个正确答案,多选或少选均不得分。

1.[单选题]终端初始的TA（time advance)调整量是通过以下哪条消息获取的?A)Random access responseB)SIBIC)RRC connection setupD)RRC Reconfiguration2.[单选题]SA组网架构下，5GRAN使用哪层完成Qos映射？A)SDAPB)PDCPC)NGAPD)RLC3.[单选题]S.A网络中,关于NR小区重选的准则描述错误的是哪项?A)触发最好小区重选时需要满足Rs<RnB)R_S=Qmeas,s+QhystC)R_n=Qmeas,n-QoffsetD)满足判决条件至少5s后才会触发重选4.[单选题]以下哪项不是对CPE做NR下行峰值调测时的建议操作?A)时隙配比设置为4:1B)调制阶数设置支持256QAMC)MIMO层数设置为4流D)把CPE终端放置在离AAU两米处5.[单选题]关于公共的PDCCH搜索空间,其最小的CCE格式是以下哪种?A)16CCEB)2CCEC)4CCED)8CCE6.[单选题]以下哪个信号可以反映NR网络上行的覆盖?A)PDCCH DMRSB)PBCH DMRSC)SRSD)csI-Rs7.[单选题]以下关于5QI与QFI关系的描述，错误的是哪一项?C)如果5QI作为QoS流的QFI，该QoS流必须承载Non-GBRd业务OD)QFI取值长度6bits,而5QI长度8bits8.[单选题]华为RAN3.1低频最大支持频段内多少载波聚合？A)4B)2C)3D)59.[单选题]5G 系统支持ExtendCP的子载波间隔是多少?A)15KHzB)30KHzC)60KHzD)24KHz10.[单选题]gNodeB根据UE上报的CQI，将其转换为几位长的MCS?A)2bitB)5bitC)4bitD)3bit11.[单选题]5G 小区中，UE的激活BWP带宽为100RB，则相对应关联的CSI-RS资源的最小RB个数是A)24B)100C)32D)6412.[单选题]NR小区载波聚合在哪一层判断数据是否需要分流?A)PDCPB)SDAPC)MACD)RLC13.[单选题]5GSA小区进行异系统移动性测量时,可下发最大多少个LTE测量频点?A)32B)8C)16D)6414.[单选题]一NR小区SSB波束采用默认模式，天线挂高35米，机械下倾角为3°，数字下倾配置为0°，则此小区主覆盖波瓣的下沿(近点)距离基站大约是多少米?A)330米B)1200米C)670米D)150米15.[单选题]5G RAN3. 1中，ChMeas. MCS. DuCel1主要是用来测量以下哪类指标？A)小区业务量16.[单选题]NSA组网中，要CPE1.0达到下行1000Mbps峰值，以下哪一项为小区下行速率的最低要求？A)900MbpsB)860MbpsC)800MbpsD)700Mbps17.[单选题]MIB消息中的哪个参数指示了CORESET0的时域符号数目？A)PDCCH- configSIB1高4位B)ssb-subcarrier offsetC)system frame numberD)PDCCH-configSIB1低4位18.[单选题]数传测试过程中发现商用终端一直保持RANK4但是MCS很低，可能是什么原因导致的？A)多径信道相关性太强B)RANK参数被固定C)空口质差D)MCS参数设置错误19.[单选题]以下关于切换失败惩罚的描述，错误的是哪一项?A)切换准备失败惩罚按原因分为资源类和非资源类B)资源类失败原因，通过设置惩罚定时器限制切换重试C)UE向目标小区切换失败超过2次，则不再向该目标小区发起切换D)非资源类失败原因，通过惩罚次数限制UE切换重试20.[单选题]在同频小区重选过程中，如果想实现终端从服务小区到某个特定邻区重选更容易，那么该如何修改参数?A)增加QoffsetB)增加QhystC)减小QoffsetD)诚小Qhyst21.[单选题]广播信息RMSI和随机接入响应消息RAR，默认采用那种MSC传输?A)MSC0B)MSC1C)MSC2D)MSC322.[单选题]5GSA组网中，以下哪种RRC状态转换流程是不支持的？A)RRC去激活到RRC连接B)RRC空闲到RRC去激活C)RRC去激活到RRC空闲D)RRC空闲到RRC连接23.[单选题]在5GC中，以下哪个模块用于用户的鉴权管理?A)ANFB)AUSFC)PCFD)SMFB)波束恢复C)RRC连接重建D)上行数据到达25.[单选题]超远邻区识别的自优化功能在优化邻区属性时，不会参考以下哪一项信息?A)掉线率B)切换次数C)邻区间距离D)切换成功率26.[单选题]以下哪个参数不用于PRACH功率控制?A)preamblereceivedt argetpowerB)preambletransmaxC)PRACH-CONFI guration indexD)SSB Tx Power27.[单选题]N.R小区峰值测试中,UE支持256QAM时MCS最高能达到多少阶?A)26阶B)27阶C)29阶D)28阶28.[单选题]如果下行带宽内PRB个数为100，RBG的大小是多少？A)2B)16C)4D)829.[单选题]以下关于5GPRACH相关的描述,正确是哪一项?A)PRACH中GT保护时间的长度与PUSCHSCS有关B)PUSCHSCS不会影响小区半径的计算C)PRACH中CP的时间长度与PRACHSCS有关D)PRACHSCS和PUSCHSCS要一致30.[单选题]5G RAN3.1小区上/下行平均吞吐率话统指标关于哪层进行统计?A)RLCB)SDAPC)MACD)PDCP31.[单选题]SA小区的系统内同频切换不存在以下哪个环节?A)切换环节B)判决环节C)触发环节D)测量环节32.[单选题]5G空口灌包时测出来的速率是哪个协议层的速率?A)PDCP33.[单选题]NR小区的其他下行信道都是基于以下哪个功率值作为基准来提供功率的?A)CSI-RS功率B)MaxTranmitPowerC)PDCCH功率D)PBCH功率34.[单选题]PDSCH DMRS和TRS之间使用哪种类型的QCL?A)TypeBB)TypeAC)TypeCD)TypeD35.[单选题]在NR组网下，为了用户能获得接近上行最高速率，其MCS值最低要求应该是多少?A)25B)16C)20D)3236.[单选题]3GPP标准定义的5G传播模型中，对杆站和宏站天线的典型高度定义分别是多少？A)20米，40米B)10米，40米C)15米，25米D)10米，25米37.[单选题]华为RAN3.132TRX的AAU最大支持多少层PDCCH?A)16层B)8层C)4层D)2层38.[单选题]华为RAN3.1NR小区使用哪一项作为PDU资源变更失败的打点？A)PDU SESSION RESOURCE SETUPFATLURE的原因值B)PDU SESSION RESOURCE SETUPREJECTC)PDU SESSION RESOURCE SETUPFAILURED)PDU SESSION RESOURCE SETUPRESPONSE的原因值39.[单选题]N.SA场景当终端检测到NR侧上行RLC达到最大重传时会触发什么流程?A)主动上报SCGFailureB)主动发起随机接入C)主动发起重建D)主动发起上行重同步40.[单选题]在5G 到4G 的重选过程中，UE通过哪条消息获取4G 频率的重选优先级?A)SIB6B)SIB4C)SIB7D)SIB5B)主动发起随机接入C)主动发起重建D)主动发起上行重同步42.[单选题]关于NSA组网的gNodeB添加流程,以下哪个指标只能在gNodeB侧统计?A)随机接入成功次数B)gNodeB添加成功次数C)gNodeB添加尝试次数D)gNodeB添加拒绝次数43.[单选题]以下哪一个测量对象可以反映特定两小区的切换性能，统计系统内特定两两小区之间的切换指标?A)NRCELLtoECELLB)NRDUCELLC)NRDUCELLTRPD)NRCellRelation44.[单选题]在NSA组网中,如果eNdoB侧的配置的gNOdeBID长度和gNode侧配置的不一致,会导致以下哪个问题A)eNodeB不发起eNode添加请求B)eNodeB不回复添加请求C)eNodeB拒绝添加请求D)eNdoeB不下发NR的测量配置45.[单选题]64T64RAA支持的NR广播波束的水平3dB波宽，最大可以支持多少？A)65°B)90C)45°D)11046.[单选题]NR2.6GHzSCS=30KHz小区和LTE-TDD2.6GHz共同组网场景，当LTE小区采用DSUDD，SSP7，NR小区采用8：2配比，SS54时，还需要设多少偏置才能保证帧结构对齐？A)4msB)3msC)1msD)2ms47.[单选题]对于NSA组网gNodeB必须广播的系统消息是用一项A)MIBB)SIB1C)SIB5D)SIB248.[单选题]用NR覆盖高层楼宇时，NR广播波束场景化建议配置成以下哪项?A)SCENARIO_1B)SCENARIO_0C)SCENARIO_6D)SCENARIO_1349.[单选题]SIB2消息的调度信息，是在哪一条消息中获取？D)SIB450.[单选题]以下哪---种SSC Mode 不提供IP连续性，适用于网页浏览等无连续性需求应用?A)SSC2B)SSC3C)SSC4D)SSC1第2部分：多项选择题，共34题，每题至少两个正确答案,多选或少选均不得分。

基于随机几何与排队论的无线网络性能研究

基于随机几何与排队论的无线网络性能研究基于随机几何与排队论的无线网络性能研究随着无线网络的飞速发展，人们对网络性能的要求也越来越高。

为了提高网络的性能，研究人员们开始利用随机几何和排队论等理论进行无线网络性能研究，这些研究对于无线网络的优化与改进具有重要意义。

随机几何是研究空间中随机分布的点与物体之间关系的数学学科，对于无线网络而言，可以利用随机几何来建模网络中用户的空间分布。

无线网络中的用户可以看作是在空间中的随机分布的点，而基站则可以看作是放置在空间上的固定点。

研究人员通过利用随机几何理论，可以对无线网络中用户之间的距离、用户到基站的距离等进行建模，从而对网络性能进行分析与优化。

排队论是研究顾客到达和服务之间的关系的数学学科，对于无线网络而言，可以利用排队论来研究网络中用户的排队现象。

在无线网络中，用户需要不断地向基站发送数据请求，而基站需要对这些请求进行处理。

如果基站处理速度跟不上用户的数据请求速度，就会出现排队现象，用户需要等待其他用户的数据请求处理完毕才能得到回应。

利用排队论，研究人员可以对无线网络中用户的排队时间、排队长度等进行建模，从而对网络性能进行分析与优化。

通过结合随机几何和排队论等理论，可以对无线网络的性能进行研究和优化。

具体研究内容可围绕以下几个方面展开：首先，可以基于随机几何理论进行网络拓扑建模。

通过建立网络中用户与基站之间的空间距离模型，可以分析和优化用户之间的信号干扰，提高网络的容量和覆盖范围。

其次，可以基于排队论进行网络流量建模。

通过分析用户的数据请求和基站的处理速度，可以建立网络中的排队模型，从而分析和优化网络的延迟和吞吐量，提高网络的性能。

此外，还可以利用随机几何和排队论等理论进行无线资源的优化分配。

通过研究用户在网络中的分布情况和数据请求情况，可以合理地分配无线资源，提高网络的利用率和性能。

最后，可以通过仿真实验验证理论模型的正确性和有效性。

利用计算机仿真技术，可以模拟真实的无线网络环境，并进行各种性能测试和优化实验，以验证随机几何和排队论等理论的可行性和实用性。

cluster_louvain加权算法原理

cluster_louvain加权算法原理
cluster_louvain加权算法是一种用于社区发现的算法，它基于
贪心策略，旨在最大化整个网络的模块度。

其原理如下：
1. 初始化：将每个节点作为一个单独的社区。

2. 计算每对节点之间的边的权重：将节点对之间的边的权重设置为节点间的连接权重之和。

3. 对于每个节点，计算它与所有其他社区的连边权重之和。

这个过程可以用于寻找与指定节点相邻的其他社区。

4. 对于每个节点，将它移动到与其相邻的社区中，如果这个移动能够使模块度增加。

5. 重复步骤3和4，直到不能再进行移动为止。

在每次迭代中，都会重新计算每对节点之间的边的权重，以反映节点移动后的变化。

6. 最后，将所有节点按照他们所在的社区进行聚类，输出最终的社区划分结果。

cluster_louvain加权算法通过不断迭代节点的移动，将相邻节
点划分到同一个社区，并尽量减少社区之间的边的权重，从而提高整个网络的模块度。

模块度是一个衡量社区结构优劣的指标，用于评估社区内部的紧密性和社区之间的松散程度。

因此，
cluster_louvain加权算法可以用于发现网络中的社区结构，并提供一种有效的社区划分方法。

2024年电信5G基站建设理论考试题库(附答案)

2024年电信5G基站建设理论考试题库（附答案）一、单选题1.在赛事保障值守过程中，出现网络突发故障，需要启用红黄蓝应急预案进行应急保障，确保快速处理和恢复。

红黄蓝应急预案的应急逻辑顺序为（)A、网络安全->用户感知->网络性能B、网络性能->用户感知->网络安全C、用户感知->网络安全->网络性能D、用户感知->网络性能->网络安全参考答案：D2.2.1G规划，通过制定三步走共享实施方案，降配置，省TCO不包含哪项工作?A、低业务小区并网B、低业务小区关小区C、低业务小区拆小区D、高业务小区覆盖增强参考答案：D3.Type2-PDCCHmonsearchspaceset是用于（）。

A、A)OthersysteminformationB、B)PagingC、C)RARD、D)RMSI参考答案：B4.SRIOV与OVS谁的转发性能高A、OVSB、SRIOVC、一样D、分场景，不一定参考答案：B5.用NR覆盖高层楼宇时，NR广播波束场景化建议配置成以下哪项？A、SCENARTO_1B、SCENARIO_0C、SCENARIO_13D、SCENARIO_6参考答案：C6.NR的频域资源分配使用哪种方式？A、仅在低层配置(非RRC)B、使用k0、k1和k2参数以实现分配灵活性C、使用SLIV控制符号级别的分配D、使用与LTE非常相似的RIV或bitmap分配参考答案：D7.SDN控制器可以使用下列哪种协议来发现SDN交换机之间的链路？A、HTTPB、BGPC、OSPFD、LLDP参考答案：D8.NR协议规定,采用Min-slot调度时，支持符号长度不包括哪种A、2B、4C、7D、9参考答案：D9.5G控制信道采用预定义的权值会生成以下那种波束？A、动态波束B、静态波束C、半静态波束D、宽波束参考答案：B10.TS38.211ONNR是下面哪个协议（）A、PhysicalchannelsandmodulationB、NRandNG-RANOverallDescriptionC、RadioResourceControl(RRC)ProtocolD、BaseStation(BS)radiotransmissionandreception参考答案：A11.在NFV架构中，哪个组件完成网络服务（NS）的生命周期管理？A、NFV-OB、VNF-MC、VIMD、PIM参考答案：A12.5G需要满足1000倍的传输容量,则需要在多个维度进行提升，不包括下面哪个（）A、更高的频谱效率B、更多的站点C、更多的频谱资源D、更低的传输时延参考答案：D13.GW-C和GW-U之间采用Sx接口，采用下列哪种协议A、GTP-CB、HTTPC、DiameterD、PFCP参考答案：D14.NR的频域资源分配使用哪种方式？A、仅在低层配置(非RRC)B、使用k0、k1和k2参数以实现分配灵活性C、使用SLIV控制符号级别的分配D、使用与LTE非常相似的RIV或bitmap分配参考答案：D15.下列哪个开源项目旨在将电信中心机房改造为下一代数据中心？A、OPNFVB、ONFC、CORDD、OpenDaylight参考答案：C16.NR中LongTruncated/LongBSR的MACCE包含几个bit（）A、4B、8C、2D、6参考答案：B17.对于SCS120kHz，一个子帧内包含几个SlotA、1B、2C、4D、8参考答案：D18.SA组网中，UE做小区搜索的第一步是以下哪项？A、获取小区其他信息B、获取小区信号质量C、帧同步，获取PCI组编号D、半帧同步，获取PCI组内ID参考答案：D19.SA组网时，5G终端接入时需要选择融合网关，融合网关在DNS域名的'app-protocol'name添加什么后缀？A、+nc-nrB、+nr-ncC、+nr-nrD、+nc-nc参考答案：A20.NSAOption3x组网时，语音业务适合承载以下哪个承载上A、MCGBearB、SCGBearC、MCGSplitBearD、SCGSplitBear参考答案：A21.5G需要满足1000倍的传输容量,则需要在多个维度进行提升，不包括下面哪个（）A、更高的频谱效率B、更多的站点C、更多的频谱资源D、更低的传输时延参考答案：D22.以SCS30KHz，子帧配比7:3为例，1s内调度次数多少次，其中下行多少次。

Spectral Relaxation for K-means Clustering

Anei.dFaosr
matrix, a given
paanrdtitAioi nis
min-b(y1-)s,i,thi.ee.a,stshoeciaitthedcsluusmte-or fc-soqnutaariness
ss( ) = Xk Xsi ka(si) ? mik2; mi = Xsi a(si)=si;
Chris Ding & Horst Simon
NERSC Division Lawrence Berkeley National Lab.
UCfBcherqkdeinlegy,,hdBseirmkoelnegy@, ClbAl.g9o4v720
Ming Gu
Dept. of Mathematics UC Berkeley, Berkeley, CA 95472
Xk si
i=1
Aie si
2 = Xk sikmik2;
i=1
a weighted sum of the squared Euclidean norms of the mean vector of each clusters.
Remark. If we consider the elements of the Gram matrix AT A as measuring similarity between data vectors, then we have shown that Euclidean distance leads to Euclidean inner-product similarity. This inner-product can be replaced by a general Mercer kernel as is done in 9, 3].

磁共振图像非均匀场校正方法研究

ｃｕｎ，ｅａｎｄ
ｓｕ晌ｃｅ
ｆｉｔｔｉｎｇｂａＳｅｄ
ｏｎ
ｌｅａｓｔ－ｓｑｕａｒｅｓｍｅｔｈｏｄｔｏｉｎｔｅｒａｃｔｉｏｎａ１１ｄ
ｎ０
ｒｅｄｕｃｅｔｈｅｉｍａｇｅｅｎｔｒｏｐｙ．Ｂｅｓｉｄｅｓ，
ｔｈｅｍｅｔｌｌｏｄｒｅｑｕｉｒｅｓ
ｎｏｕｓｅｒ
ＩｌＩ
ｈａｓｂｅｅｎｕｓｅｄｔｏｃｏ盯ｅｃｔｔｈｅｓｉｍｕｌａｔｅｄｐｒｉ。ｒｌ（１１。ｗｌｅｄｇｅ，ａｎｄ
ｉｎｔｅｎｓｉｔｙｉ１１ｈｏｍｏｇｅｎｅ时ｉｎ
ＭＲｉｍａｇｅ，ｔｈｅｉｍａｇｅｑｕａｌ畸ｉｓ
ｏｂＶｉｏｕｒｓｌｙ．ＡｎｏｔｈｅｒｍｅｔｈｏｄｂａＳｅｄ
ｏｎ
ｅｎｔｒｏｐｙｍｉｎｉｍｉｚａｌｉｏｎｉｓａｌｓｏｐｒｏｐｏｓｅｄ，ｕｓｉｎｇｐａｎｉｃｌｅ
ｓｗ锄ｏｐｔｉｍｉｚａｔｉｏｎ，
ｔｈｅｉｍａｇｅｂｙｉｎｃｏ叩ｏｒａｔｅｄｔ：ｈｅｂｉａｓｍｏｄｅｌ，ａｒｅａｃｏｍｒｏｌｓｕｒｆａｃｅｆｉｔｔｉｎｇｂａＳｅｄｏｆ
ｏｎｉｎｆｏｍｌａｔｉｏｎ，ｃｕｒｖｅａＩｌｄ
ｌｅａＳｔ－ｓｑｕａｒｅｓｍｅｔｈｏｄ．Ｔｈｅｍｅｍｏｄｉｓｂｒａｉｎ
ｓｕｉ诎ｌｅ
ｆｏｒｃｏｒｒｅｃｔｉｏｎｉｍｐｒｏＶｅｄ
明，两种算法都可以取得很好的校正效果。
基于模糊均值聚类算法改进了基于灰度信息的模糊Ｃ均值（ＦＣＭ）算法，将偏移场模型、代表图像空间信息的邻域控制信息和最小二乘曲面拟合方法有机结合，能同时实现图像的校正和聚类，适用于灰度不均匀脑部磁共振图像的校正。实验结果表明校正后的图像质量得到较大提高。基于熵最小化的自动校正算法是一种新颖的自动校正算法，该算法以图像熵最小为目标，利用拟合曲面、粒子群算法实现对ＭＲ图像的亮度校正。该方法是一种全自动的无需先验知识的校正方法，在对ＢｒａｉｎＷｒｅｂ数据库仿真Ｍ对和真实ＭＲＩ校正中，取得了令人满意的校正结果，达到了实际应用的要求。本文提出的两种算法各有优点，基于模糊均值聚类的校正算法调整参数少，计算速度快，适用于组织较为简单的ＭＲ图像的校正；基于熵最小化的自动校

复杂网络中的社区发现与分析

复杂网络中的社区发现与分析人们在日常生活中经常会听到“社区”这个词，指的是一群共同具有某些特性、彼此有相互交往并且相对孤立于其他群体的人或组织。

而在复杂网络中，社区也有着类似的定义：指的是网络中由一些紧密相连的节点组成的一个子图，与其他子图相对孤立。

社区也被称为群组（clique）、簇（cluster）等。

在现实中，社区的发现对很多领域都有着重要的应用价值。

例如，在生命科学中可以通过社区发现来解析蛋白质复合物、基因调控网络等；在社会网络分析中，可以通过社区发现来分析朋友圈、领导小组、商业竞争等。

因此，如何有效地发现复杂网络中的社区，一直是研究的热点和难点。

社区发现的方法目前，社区发现的方法主要有以下几类：1. 基于聚类的方法基于聚类的方法是将网络中的节点划分到不同的簇中。

其中，最简单的方法是K-means，它将节点按照相似性分到不同的组中。

这种方法的优点是速度快，适用于规模较小的网络。

但是，缺点也很明显，随着网络规模增大，聚类结果会受到噪声的干扰，导致分类不准确。

2. 基于谱聚类的方法基于谱聚类的方法将节点之间的相似性表示为矩阵，并使用谱分解来找到最优的社区划分，它不仅适用于规模较小的网络，而且对噪声有很好的抗干扰能力。

但是，它也有缺点，例如当网络具有较多的噪声时会使得谱聚类的效果变差。

3. 基于模块度优化的方法基于模块度优化的方法是划分社区的一种常用方法，其基本思路是通过最大化社区内部的联系和最小化社区与社区之间的联系，来得到最优的社区划分。

其中，例如Newman的模块度最大化法、GN算法等，都是基于模块度优化的方法。

这种方法的优势在于时间效率高，但是对于社区分布不均匀或社区重叠等情况，会降低其准确性。

4. 基于深度学习的方法近年来，深度学习在社区发现中的应用越来越广泛。

基于深度学习的方法通过训练神经网络，来预测节点所属的社区。

例如CN-Ke-GAE、SDNE等方法，都是基于深度学习的方法。

相对于其他方法，它在对规模较大、社区之间重叠等问题有着更好的应对能力。

INTERNATIONAL JOURNAL OF WIRELESS AND MOBILE COMPUTING (IJWMC) 1 A Biologically Inspired Qo

A Biologically Inspired QoS Routing Algorithm forMobile Ad Hoc NetworksZhenyu Liu,Marta Z.Kwiatkowska,and Costas ConstantinouAbstract—This paper presents an Emergent Ad hoc Routing Algorithm with QoS provision(EARA-QoS).This ad hoc QoS routing algorithm is based on a swarm intelligence inspired routing infrastructure.In this algorithm,the principle of swarm intelligence is used to evolutionally maintain routing information. The biological concept of stigmergy is applied to reduce the amount of control trafﬁc.This algorithm adopts the cross-layer optimisation concept to use parameters from different layers to determine routing.A lightweight QoS scheme is proposed to provide service-classiﬁed trafﬁc control based on the data packet characteristics.The simulation results show that this novel routing algorithm performs well in a variety of network conditions.Index Terms—MANET,routing,QoS,swarm intelligence.I.I NTRODUCTIONM OBILE ad hoc networks(MANETs)are wireless mo-bile networks formed munication in such a decentralised network typically involves temporary multi-hop relays,with the nodes using each other as the relay routers without anyﬁxed infrastructure.This kind of network is veryﬂexible and suitable for applications such as temporary information sharing in conferences,military actions and disaster rescues.However,multi-hop routing,random movement of mobile nodes and other features unique to MANETs lead to enormous overheads for route discovery and maintenance.Furthermore, compared with the traditional networks,MANETs suffer from the resource constraints in energy,computational capacities and bandwidth.To address the routing challenge in MANETs,many ap-proaches have been proposed in the literature.Based on the routing mechanism for the traditional networks,the proactive approaches attempt to maintain routing information for each node in the network at all times[1]–[3],whereas the reactive approaches onlyﬁnd new routes when required[4]–[6].Other approaches make use of geographical location information for routing[7],[8].Those previous works only provide a basic “best effort”routing functionality that is sufﬁcient for con-ventional applications such asﬁle transfer or email download. To support real-time applications such as V oIP and video stream in MANETs,which have a higher requirement for delay,jitter and packet losses,provision of Quality-of-Service (QoS)is necessary in addition to basic routing functionality. Z.Liu and M.Z.Kwiatkowska is with School of Computer Science,The University of Birmingham,Birmingham,England B152TT.C.Constantinou is with the Department of Electronic Electrical and Computer Engineering,The University of Birmingham,Birmingham,England B152TT.Given the nature of MANETs,it is difﬁcult to support real-time applications with appropriate QoS.In some cases it may be even impossible to guarantee strict QoS requirements.But at the same time,QoS is of great importance in MANETs since it can improve performance and allow critical information to ﬂow even under difﬁcult conditions.At present,the most fundamental challenges of QoS support in MANETs concern how to obtain the available bandwidth and maintain accurate values of link state information during the dynamic evolution of such a network[9].Based on common techniques for QoS provision in the Internet,some researchers proposed the integration of QoS provision into the routing protocols[10],[11].However,since these works implicitly assumed the same link concept as the one in wired networks,they still do not fully address the QoS problem for MANETs.In this paper,we propose a new version of the self-organised Emergent Ad hoc Routing Algorithm with QoS provisioning(EARA-QoS).This QoS routing algorithm uses information from not only the network layer but also the MAC layer to compute routes and selects different paths to a destination depending on the packet characteristics.The underlying routing infrastructure,EARA originally proposed in[12],is a probabilistic multi-path algorithm inspired by the foraging behaviour of biological ants.The biological concept of stigmergy in an ant colony is used for the interaction of local nodes to reduce the amount of control trafﬁc.Local wireless medium information from the MAC layer is used as the artiﬁcial pheromone(a chemical used in ant communications) to reinforce optimal/sub-optimal paths without the knowledge of the global topology.One of the optimisations of EARA-QoS over EARA is the use of metrics from different layers to make routing decisions. This algorithm design concept is termed as the cross-layer design approach.Research[13]has shown the importance of cross-layer optimisations in MANETs,as the optimisation at a particular single layer might produce non-intuitive side-effects that will degrade the overall system performance.Moreover, the multiple-criteria routing decisions allow for the better usage of network characteristics in selecting best routes among multiple available routes to avoid forwarding additional data trafﬁc through the congested areas,since the wireless medium over those hotspots is already very busy.The parameters for measuring wireless medium around a node depend largely on the MAC layer.In this paper,we focus on the IEEE802.11 DCF mode[14],since it is the most widely used in both cellular wireless networks and in MANETs.This cross-layer technique of using MAC layer information can be appliedeasily to other MAC protocols.In addition to the basic routing functionality,EARA-QoS supports an integrated lightweight QoS provision scheme.In this scheme,trafﬁc ﬂows are classiﬁed into different service classes.The classiﬁcation is based on their relative delay bounds.Therefore,the delay sensitive trafﬁc is given a higher priority than other insensitive trafﬁc ﬂows.The core technique of the QoS provision scheme is a token bucket queuing scheme,which is used to provide the high priority to the real-time trafﬁc,and also to protect the lower-priority trafﬁc from star-vation.Experimental results from simulation of mobile ad hoc networks show that this QoS routing algorithm performs well over a variety of environmental conditions,such as network size,nodal mobility and trafﬁc loads.II.B ACKGROUNDIn this section,we give a brief introduction to background knowledge on ant colony heuristics,and the QoS provision techniques in MANETs.A.Foraging Strategies in AntsOne famous example of biological swarm social behaviour is the ant colony foraging [15](see Figure 1).Many ant species have a trail-laying,trail-following behaviour when foraging:individual ants deposit a chemical substance called pheromone as they move from a food source to their nest,and foragers follow such pheromone trails.Subsequently,more ants are attracted by these pheromone trails and in turn reinforce them even more.As a result of this auto-catalytic effect,the optimal solution emerges rapidly.In this food searching process a phenomenon called stigmergy plays a key role in developing and manipulating local information.It describes the indirect communication of individuals through modifying theenvironment.Fig.1.All Ants Attempt to Take the Shortest PathFrom the self-organisation theory point of view,the be-haviour of the social ant can be modelled based on four elements:positive feedback,negative feedback,randomness and multiple interactions [16].This model of social ants using self-organisation theories provides powerful tools to transfer knowledge about the social insects to the design of intelligent decentralised problem-solving systems.B.Quality-of-Service in MANETsQuality-of-Service (QoS)provision techniques are used to provide some guarantee on network performance,such as average delay,jitter,etc.In wired networks,QoS provision can generally be achieved with the over-provisioning of re-sources and with network trafﬁc engineering [17].With the over-provisioning approach,resources are upgraded (e.g.ﬁbre optic data link,advanced routers and network cards)to make networks more resistant to resource demanding applications.The advantage of this approach is that it is easy to be implemented.The main disadvantage of this approach is that all the applications still have the same priority,and the network may become unpredictable during times of bursting and peak trafﬁc.In contrast,the idea of the trafﬁc engineering approach is to classify applications into service classes and handle each class with a different priority.This approach overcomes the defect of the former since everyone is following a certain rule within the network.The trafﬁc engineering approach has two complemen-tary means to achieve QoS provisioning,Integrated Services (IntServ)and Differentiated Services (DiffServ).IntServ [18]provides guaranteed bandwidth for ﬂows,while DiffServ [19]provides hard guarantees for service classes.Both of the approaches rely on the possibility to make bandwidth reservations.The former was used in ATM (Asynchronous Transfer Mode)[20]and is today the method of achieving QoS in RSVP-IntServ [21].On the other hand,in the DiffServ approach,no reservation is done within the network.Instead,QoS is achieved by mechanisms such as Admission Control ,Policy Manager ,Trafﬁc Classes and Queuing Schedulers .These mechanisms are used to mark a packet to receive a particular forwarding or dropping treatment at each node.Based on QoS provision techniques in wired networks,many QoS approaches are proposed to provide QoS services for MANETs.Flexible QoS Model for MANETs (FQMM)[22],is the ﬁrst QoS approach for MANETs,which combines knowledge on IntServ/DiffServ in wired networks with con-sideration of MANETs.As an essential component to achieve the QoS provisioning,QoS routing algorithms tightly integrate QoS provisioning into routing protocols.The QoS version of AODV (QoS-AODV)[23],the Core-Extraction Distributed Ad Hoc Routing (CEDAR)protocol [10],the Multimedia Support for Mobile Wireless Networks (MMWN)protocol [11],and the ticket-based protocols [24]are examples of QoS routing algorithms proposed for MANETs.On the other hand,QoS signaling techniques are inde-pendent of the underlying routing protocols.The In-band Signalling for QoS in Ad-Hoc Mobile Networks (INSIGNIA)algorithm [25]is the typical signaling protocol designed exclusively for MANETS.The idea of CEDAR,MMWN,and ticket-based protocols is to disseminate link-state information across the network in order to enable other nodes to ﬁnd routes that meet certain QoS criteria,like the minimum bandwidth.On the other hand,INSIGNIA piggybacks resource reservations onto data packets,which can be modiﬁed by intermediate nodes to inform the communication endpoint nodes in case of lack ofresources.All those approaches are based on the idea that the wireless links between mobile nodes have certain QoS related properties,in particular a known amount of available bandwidth,and that nodes are able to give guarantees for trafﬁc traversing these links.III.C RITIQUE OF E XISTING Q O S A PPROACHES INMANET SNowadays,most of the QoS provisioning techniques are derived from the QoS approaches of the wired networks. However,QoS support approaches proposed in wired networks are based on the assumption that the link characteristics such as bandwidth,delay,loss rate and error rate must be available and manageable.However,given the challenges of MANETs, e.g.dynamic topology and time-varying link capacity,this assumption does not apply any longer.Thus,applying the concepts of wired trafﬁc engineering QoS approaches directly to MANETs is extremely difﬁcult.Generally,the situation in MANETs is completely different from those in wired networks.In wireless networks,the available bandwidth undergoes fast time-scale variations due to channel fading and errors from physical obstacles.These effects are not present in wired networks.In MANETs,the wireless channel is a shared-access medium,and the available bandwidth even varies with the number of hosts contending for the channel.Below we analyse why the IntServ/DiffServ models are not appropriate for MANETs respectively. IntServ based approaches are not applicable for MANETs mainly due to two factors,huge resource consumption and computation power limitation.Firstly,to support IntServ,a huge amount of link state information has to be built and main-tained for each mobile node.The amount of state information increases proportionally with the number ofﬂows,which is also a problem with the current IntServ QoS scheme.Secondly, current wireless networks employ two major MAC techniques, the single-channel approach and the multiple channel ap-proach.With single-channel approach(e.g.IEEE802.11[14]), all nodes share the same channel and therefore potentially interfere with each other.With a multiple-channel approach (e.g.Bluetooth[26]or CDMA[27]),nodes can communicate on several channels simultaneously.Both of the two MAC techniques have a similar bandwidth reservation mechanism. This common mechanism requires a transmission schedule to deﬁne time slots,in which nodes take their turns periodically. For each slot,its duration and a set of possible simultaneous transmissions must be deﬁned.However,in wireless networks, the problem ofﬁnding an optimal schedule is proved to be NP-complete[28],which is a fundamental limitation of QoS provisioning in wireless networks.On the other hand,the DiffServ approach is a lightweight QoS model for interior routers since individual stateﬂows are aggregated into sets of service classes whose packets are treated differently at the routing nodes.This makes routing a lot easier in the network.Thus this approach could be a potential solution for MANETs.Even though it is not practical to provide a hard separation of different service classes in MANETs,relative prioritisation is possible in such a way that trafﬁc of a certain class is given a higher or lower priority than trafﬁc of other service classes.One solution would be to divide the trafﬁc into a predeﬁned set of service classes that are deﬁned by their relative delay bounds,such as delay sensitive(realtime)and insensitive(bulk)trafﬁc.Realtime trafﬁc should be given higher priority than bulk trafﬁc.No absolute bandwidth guarantees are provided.Some work based on service differentiation rather than resource reservations in MANETs already exists[29].IV.D ESCRIPTION OF EARA-Q O SEARA-QoS is an on-demand multipath routing algorithm for MANETs,inspired by the ant foraging intelligence.This algorithm incorporates positive feedback,negative feedback and randomness into the routing computation.Positive feed-back originates from destination nodes to reinforce the existing pheromone on good paths.Ant-like packets,analogous to the ant foragers,are used to locallyﬁnd new paths.Artiﬁcial pheromone is laid on the communication links between nodes and data packets are biased towards strong pheromone,but the next hop is chosen probabilistically.To prevent old routing solutions from remaining in the current network status,expo-nential pheromone decay is adopted as the negative feedback. Each node using this algorithm maintains a probabilistic routing table.In this routing table,each route entry for the destination is associated with a list of neighbour nodes.A probability value in the list expresses the goodness of node as the next hop to the destination.For each neighbour, the shortest hop distance to the destination and the largest sequence number seen so far are also recorded.In addition to the routing table,each node also possesses a pheromone table.This table tracks the amount of pheromone on each neighbour link.The table may be viewed as a ma-trix with rows corresponding to neighbourhood and columns to destinations.There are three threshold values controlling the bounds on pheromone in the table.They are the upper pheromone that prevents extreme differences in pheromone, the lower pheromone,below which data trafﬁc cannot be forwarded,and the initial pheromone that is assigned when a new route is found.In addition to the routing data structures present above,the following control packets are used in EARA-QoS to perform routing computation:Route Request Packet(RQ)containing destination ad-dress,source address and broadcast ID.Route Reply Packet(RP)containing source address,des-tination address,sequence number,hop account and life-time.Reinforcement Signal(RS)containing destination ad-dress,pheromone value and sequence number.Local Foraging Ant(LFA)containing source address (the node that sent LFA),the least hop distance from the source to the destination,stack of intermediate node address and hop count.Hello Packet(HELLO)containing source(the node that sent Hello)address and hop count(set to0).A.Parameters of Lower Layers1)The Average MAC Layer Utilisation:Theﬁrst metric is the average MAC layer utilisation for a node.This metric measures the usage of the wireless medium around that node. As the instantaneous MAC layer utilisation at a node is either (busy)or(idle),we average this value over a period of time window as follows:(1) where is the time when the medium is busy in the window.This average MAC utilisation indicates the degree to which the wireless medium around that node is busy or idle.We consider the instantaneous MAC layer utilisation level at a node to be1when the wireless medium around that node either detects physical carrier to be present or is deferring due to virtual carrier sensing,inter-frame spacing,or backoff.In addition,we also consider the medium is busy at any time when the node has at least one packet in the transmission queue.2)The Transmission Queue Heuristic:The second metric isa heuristic value that is calculated with the network interface transmission queue length in the current node.Apart from the media status,the transmission queue length is also a key factor that can affect the packet latency or packet drop due to the size limit on the queue length.We deﬁne the heuristic value with the following rules.If the outgoing network interface employs a single queue scheme,the heuristic value is deﬁned as:(2) where is the length(in bytes waiting to be sent)of the interface queue in node,and is the maximum packet bytes allowed in the queue.If the network interface employs the multiple virtual queue scheme for each outgoing link,the heuristic value is deﬁned as:(3)where is the length(in bytes waiting to be sent)of the virtual queue of the link in node and denotes the neighbourhood of node as a next-hop to some destination.3)The Average MAC Layer Delay:The last metric is the MAC layer delay for the link.The MAC layer delay is deﬁned as the interval from when the RTS frame is sent at node to when the data frame is received successfully at node.The average MAC delay is obtained by averaging these values over a time window as follows:(4)where is the time interval in the window,and is a coefﬁcient.This average MAC delay indicates the degree of interference.In regions where there is a lot of interference from other nodes,MAC delay is high due to the contentionof the channel.B.Data PropagationWhen multiple virtual queue scheme is employed,the rout-ing probability value is computed by the composition ofthe pheromone values,the local heuristic values and the linkdelays as follows:(5) where,and()are tunable parametersthat control the relative weight of pheromone trail,MAC delay and heuristic value,and is the neighbourhood as a next-hop to some destination.Incorporating the heuristic value and link delay in the rout-ing computation makes this algorithm possess the congestionawareness property.Based on the probabilistic routing table, data trafﬁc will be distributed according to the probabilitiesfor each neighbour in the routing table.The routing algorithmexhibits load balancing behaviour.Nodes with a large number of packets in the buffer are avoided.The EARA-QoS algorithm consists of several components.They are the route discovery procedure,the positive and neg-ative reinforcement,and the local connectivity management.C.Route DiscoveryWe use a similar route discovery procedure as describedin[12].On initialisation,a neighbourhood for each node is built using the single-hop HELLO messages.Whenever atrafﬁc source needs a route to a destination,it broadcastsroute request packets(RQ)across the network.Rather than simplyﬂooding the RQ packets,we adopt the probabilisticbroadcast scheme explored in[30]combined with the MAClayer utilisation.When a nodeﬁrst receives a packet,with probability it broadcasts the packet to its neighbours,andwith probability it discards the packet.The probabilityvalue is calculated as(6) where()is the coefﬁcient.This broadcast scheme helps to discover new routes avoiding congestion areas,but atthe cost of missing potential routes to the destination. During the course ofﬂooding RQ packets to the destination ,the intermediate node receiving a RQ packetﬁrst sets up reverse paths to the source by recording the source addressand the previous hop node in the message cache.If a validroute to the destination is available,that is,there is at least one link associated with the pheromone trail greater than the lower pheromone bound,the intermediate node generates a route reply(RP).The RP is routed back to the source via the reverse paths.Otherwise,the RQ is rebroadcast.Other than just establishing a single forward path,whenthe destination node receives RQs it will send a RP to allthe neighbours from which it sees a RQ.In order to maintain multiple loop-free paths at each intermediate node,node(b) Path Reinforcement(c) Local Repair(a) Initial Pheromone Setup Fig.2.Illustrating Working Mechanism of EARA-QoSmust record all new forward paths that possess the latest sequence number but hold a lower hop-count in its routing table,and also send a RP to all the neighbours from which it saw a RQ.During the course of the RP tracking back to the source,an initial pheromone value is assigned to the corresponding neighbour node,which indicates a valid route to the destination.This process is illustrated in Figure2(a).D.Route ReinforcementAfter the destination node receives the data trafﬁc sent by the source node,it begins to reinforce some good neighbour(s)in order to“pull”more data trafﬁc through the good path(s)by sending reinforcement signal packets(RS) whenever it detects new good paths.When node receives a RS,it knows it has an outgoing link toward the destination ,which is currently deemed a good path.Subsequently, node updates the corresponding pheromone table entry with the value and forwards a RS packet to(at least one) selected neighbour locally based on its message cache,e.g.the neighbour(s)that saw the least hops of the incoming packets. The amount of the pheromone used to positively rein-force the previous hop neighbour is computed as follows.If the RS packet is sent by the destination to node,then is calculated using the upper bound pheromone value ,(7) If the RS packet is sent by an intermediate node towards node,the is calculated using the current largest pheromone value max()in node with the next hop to the destination in the pheromone table,max(8) where,and are parameters that control the relative weight of the relative source hop distance,the rel-ative packet number and the local queue heuristic. Incorporating the congestion-measuring metric into the reinforcement can lead data trafﬁc to avoid the congestion areas.The relative source hop distance is calculated as follows:(9) where is the shortest hop distance from the source to the current node through node,and is the shortest hop distance from to.This parameter is used to ensure that paths with shorter hop distance from the source node to the current node are reinforced with more pheromone.The relative packet number is calculated as follows:(10) where is the number of incoming packets from neighbour to the destination,and is the total number of incomingpackets towards the destination.This parameter is used to indicate that the data forwarding capacity of a link also affects the reinforcement.The more data arrives,the stronger reinforcement is generated for the corresponding link.On receiving the RS from a neighbour,node needs to positively increase the pheromone of the link towards node.If the sequence number in the RS is greater than the one recorded in the pheromone table,node updates its corresponding pheromone with the value of carried on the RS:(11) If the sequence number is equal to the current one,then:ifotherwise(12)If the sequence number in RS is less than the current one in the pheromone table,then this RS is just discarded.Node also has to decide to reinforce(at least)one of its neighbours by sending the RS message based on its own message cache.This process will continue until reaching the source node.As a result of this reinforcement,good quality routes emerge,which is illustrated in Figure2(b).The same procedure can apply to any intermediate node to perform local link error repair as long as it has pheromone value that is greater than the lower bound.For instance,if an intermediate node detects a link failure from one of its upstream links, it can apply the reinforcement rules to discover an alternative path as shown in Figure2(c).There is also an implicit negative reinforcement for the pheromone values.Within every time interval,if there is no data towards a neighbour node,its corresponding pheromone value decays by a factor as follows:(13)E.Local Foraging AntsIn a dynamic network like MANET,the changes of the net-work topology create chances for new good paths to emerge.In order to make use of this phenomenon,this algorithm launcheslocal foraging ants(LFA)with a time interval to locallysearch for new routes whenever all the pheromone trails of a node towards some destination drop below the threshold.The LFA will take a random walk from its original node. During the course of its walk,if the LFA detects congestionaround a node(the average channel utilisation is greaterthan a predeﬁned threshold value),then the LFA dies to avoid increasingly use the wireless medium.Otherwise,theLFA pushes the address of the nodes that it has travelledinto its memory stack.To avoid forming of loops,LFA will not choose to travel to the node that is already in.Before reaching the maximum hop,if LFA canﬁnd a node with pheromone trails greater than and the hop distanceto destination not greater than the one from its original nest,itreturns to its’nest’following its memory stack and updates the corresponding paths with.Otherwise,it simply dies.F.Local Connectivity ManagementNodes maintain their local connectivity in two ways.When-ever a node receives a packet from a neighbour,it updates its local connectivity information to ensure that it includes thisneighbour.In the event that a node has not sent any packets toits neighbours within a time interval,it has to broadcast a HELLO packet to its neighbours.Failure to receive packetsfrom the neighbourhood in indicates changes in the local connectivity.If HELLO packets are not received from the nexthop along an active path,the node that uses that next hop issent notiﬁcation of link failure.In case of a route failure occurring at node,cannot for-ward a data packet to the next hop for the intended destination .Node sends a RS message that sets ROUTE RERR tag to inform upstream nodes of the link failure.This RS signalassigns to the corresponding links the lower bound.Here, RS plays the role of an explicit negative feedback signal to negatively reinforce the upstream nodes along the failure path. This negative feedback avoids causing buffer overﬂow due to caching on-ﬂight packets from upstream nodes. Moreover,the use of HELLO packets can also help to ensure that only nodes with bidirectional connectivity are deemed as neighbours.For this purpose,the HELLO packet sent by a node has an option to list the nodes from which it has heard HELLO packets,and nodes that receive the HELLO check to ensure that it uses only routes to neighbours that have sent HELLO packets.G.The QoS Provision SchemeThis section describes a lightweight approach to DiffServ. The basic idea is to classifyﬂows into a predeﬁned set of service classes by their relative delay bounds.Admission control only works at the source node.There is no session orﬂow state information maintained at intermediate nodes. Once a realtime session is admitted,its packets are marked as RT(realtime service)and otherwise they are considered as best-effort bulk packets.As depicted in Figure3,each of these trafﬁc classes is buffered in a logically separate queue.A simple novel queuing strategy,based on the token bucket scheme,provides high priority to realtime trafﬁc,and also protects the lower-priority trafﬁc from starvation.No absolute bandwidth guarantees are provided in this scheme.We explain this queuing strategy and its novelty below.The queues are scheduled according to a token bucket scheme.In this scheme,prioritisation is achieved with token balancing.Each trafﬁc class has a balance of tokens,and the class with higher balance has a higher priority when dequeuing the next packet for transmission.For each transmission of a packet of class,an amount of tokens is subtracted from the class’token balance and an equal fraction thereof is added to every other class’balance such that the sum of all tokens is always the same.The weight value reﬂects the delay sensitivity assigned to the different classes.A higher weight value corresponds to a lower delay sensitivity.The size of the token balance together with the value determines the maximal length of a burst of trafﬁc from one class.In this scheme,as long as the amount of delay-sensitive trafﬁc does not grow too large,it is forwarded as quickly as possible,and if it does grow too large,starvation of other trafﬁc classes is prevented.Setting the upper bound of a class’token balance depending on its delay-sensitivity enables further tuning of the describedmethod.Fig.3.Overview of Service Differentiation SchemeIn this packet scheduling scheme,routing protocol pack-ets are given unconditional priority before other packets. Moreover,realtime applications normally have stringent delay bounds for their trafﬁc.This means that packets arriving too late are useless.From the application’s point of view,there is no difference between late and lost packets.This implies that it is actually useless to forward realtime packets that stay in a router for more than a threshold amount of time,because they will be discarded at the destination anyway.Dropping those packets instead has the advantage of reducing the load in the network.To our knowledge,this service classiﬁcation based queuing scheme is the simplest implemented QoS provisioning technique designed exclusively for MANETs so far.V.C HARACTERISTICS OF THE A LGORITHMThis proposed protocol,implementing the cross-layer design concept,exhibits some properties that show itsﬁtness as a solution for mobile ad hoc networks:Loop-freeness:during the route discovery phase,the nodes record the unique sequence number of RP packets.。

基于邻居信息聚合的子图同构匹配算法

2021⁃01⁃10计算机应用,Journal of Computer Applications 2021,41(1):43-47ISSN 1001⁃9081CODEN JYIIDU http ：//基于邻居信息聚合的子图同构匹配算法徐周波，李珍，刘华东*，李萍（广西可信软件重点实验室（桂林电子科技大学），广西桂林541004）（∗通信作者电子邮箱ldd@ ）摘要：图匹配在现实中被广泛运用，而子图同构匹配是其中的研究热点，具有重要的科学意义与实践价值。

现有子图同构匹配算法大多基于邻居关系来构建约束条件，而忽略了节点的局部邻域信息。

对此，提出了一种基于邻居信息聚合的子图同构匹配算法。

首先，将图的属性和结构导入到改进的图卷积神经网络中进行特征向量的表示学习，从而得到聚合后的节点局部邻域信息；然后，根据图的标签、度等特征对匹配顺序进行优化，以提高算法的效率；最后，将得到的特征向量和优化的匹配顺序与搜索算法相结合，建立子图同构的约束满足问题（CSP ）模型，并结合CSP 回溯算法对模型进行求解。

实验结果表明，与经典的树搜索算法和约束求解算法相比，该算法可以有效地提高子图同构的求解效率。

关键词：子图同构；约束满足问题；图卷积神经网络；信息聚合；图匹配中图分类号：TP391文献标志码：ASubgraph isomorphism matching algorithm based on neighbor informationaggregationXU Zhoubo ，LI Zhen ，LIU Huadong *，LI Ping（Guangxi Key Laboratory of Trusted Software （Guilin University of Electronic Technology ），Guilin Guangxi 541004，China ）Abstract:Graph matching is widely used in reality ，of which subgraph isomorphic matching is a research hotspot and has important scientific significance and practical value.Most existing subgraph isomorphism algorithms build constraints based on neighbor relationships ，ignoring the local neighborhood information of nodes.In order to solve the problem ，a subgraph isomorphism matching algorithm based on neighbor information aggregation was proposed.Firstly ，the aggregated local neighborhood information of the nodes was obtained by importing the graph attributes and structure into the improved graph convolutional neural network to perform the representation learning of feature vector.Then ，the efficiency of the algorithm was improved by optimizing the matching order according to the characteristics such as the label and degree of the graph.Finally ，the Constraint Satisfaction Problem （CSP ）model of subgraph isomorphism was established by combining the obtained feature vector and the optimized matching order with the search algorithm ，and the model was solved by using the CSP backtracking algorithm.Experimental results show that the proposed algorithm significantly improves the solving efficiency of subgraph isomorphism compared with the traditional tree search algorithm and constraint solving algorithm.Key words:subgraph isomorphism;Constraint Satisfaction Problem (CSP);graph convolutional neural network;information aggregation;graph matching0引言图匹配技术被广泛地应用于社交网络、网络安全、计算生物学和化学等领域［1］中。

用于启用非去交错信道估计的方法及设备[发明专利]

专利名称：用于启用非去交错信道估计的方法及设备
专利类型：发明专利
发明人：丽贝卡·文玲·袁,拉古·N·沙拉,于远宁,迈克尔·L·麦克劳德
申请号：CN201380011066.4
申请日：20130308
公开号：CN104160671A
公开日：
20141119
专利内容由知识产权出版社提供
摘要：本发明的方面通常涉及无线通信，且更具体地说涉及执行具有修改以改进系统性能的信道估计。

方面通常包含在用户设备UE处接收当前子帧中的来自基站的参考信号，及执行信道估计，其中所述信道估计至少部分基于所述当前子帧中接收的所述参考信号、所述UE的移动性特性及在所述当前子帧之前的子帧的配置。

申请人：高通股份有限公司
地址：美国加利福尼亚州
国籍：US
代理机构：北京律盟知识产权代理有限责任公司
代理人：宋献涛
更多信息请下载全文后查看。

agglomerativeclustering原理

agglomerativeclustering原理（原创版）目录1.概述2.agglomerative clustering 的定义3.agglomerative clustering 的基本原理4.agglomerative clustering 的步骤5.agglomerative clustering 的优点与缺点6.应用案例正文1.概述聚类分析是一种无监督学习方法，其主要目的是将相似的数据点划分到同一类别中。

在众多聚类算法中，agglomerative clustering（聚合聚类）是一种较为常见的方法。

本文将详细介绍 agglomerative clustering 的原理及其应用。

2.agglomerative clustering 的定义agglomerative clustering，又称为聚合聚类，是一种基于距离的聚类方法。

该方法通过计算数据点之间的距离，逐步将距离较小的数据点合并成簇，从而实现聚类的目的。

3.agglomerative clustering 的基本原理agglomerative clustering 的基本原理是：从所有数据点开始，将距离最近的两个数据点合并成一个簇，然后计算新簇与剩余数据点的距离，重复上述过程，直至所有数据点都被合并成一个簇。

4.agglomerative clustering 的步骤agglomerative clustering 主要分为两个步骤：（1）初始化：将每个数据点视为一个独立的簇，即所有数据点都属于簇 0。

（2）合并：根据数据点之间的距离，依次将距离最近的两个簇合并，形成一个新的簇。

重复这一过程，直至所有数据点都属于同一个簇。

5.agglomerative clustering 的优点与缺点优点：（1）能够较好地处理不同形状的聚类结构；（2）对于不同密度的聚类结构具有一定的适应性；（3）算法简单，易于实现。

缺点：（1）需要预先设定一个停止条件，如距离阈值或簇数量，否则算法可能无法结束；（2）对于大规模数据集，计算复杂度较高。

On-demand power management for ad hoc networks

On-demand power management for ad hoc networks qRong Zheng*,Robin KravetsDepartment of Compute Science,University of Illinois at Urbana-Champaign,Urbana,IL61801,USAReceived17August2003;accepted6September2003Available online26November2003AbstractBattery power is an important resource in ad hoc networks.It has been observed that in ad hoc networks,energy consumption does not reﬂect the communication activities in the network.Many existing energy conservation protocols based on electing a routing backbone for global connectivity are oblivious to traﬃc characteristics.In this paper,we propose an extensible on-demand power management framework for ad hoc networks that adapts to traﬃc load.Nodes maintain soft-state timers that determine power management transitions.By monitoring routing control messages and data transmission,these timers are set and refreshed on-demand.Nodes that are not involved in data delivery may go to sleep as supported by the MAC protocol.This soft state is aggregated across multipleﬂows and its maintenance re-quires no additional out-of-band messages.We implement a prototype of our framework in the ns-2simulator that uses the IEEE802.11MAC protocol.Simulation studies using our scheme with the Dynamic Source Routing protocol show a reduction in energy consumption near50%when compared to a network without power management under both long-lived CBR traﬃc and on–oﬀtraﬃc loads,with comparable throughput and latency.Preliminary results also show that it outperforms existing routing backbone election approaches.Ó2003Elsevier B.V.All rights reserved.Keywords:Ad hoc networks;Energy conservation;Power management;Soft state;On demand1.IntroductionWith the proliferation of portable computing platforms and small wireless devices,ad hoc wireless networks have received more and more attention as a means for providing data commu-nications among devices regardless of their physi-cal locations.Wireless communication has the advantage of allowing untethered communication, which implies reliance on portable power sources such as batteries.However,due to the slow advancement in battery technology,battery power continues to be a constrained resource.It has been observed that in ad hoc networks, energy consumption does not always reﬂect active communication in the network[1].Experimental results reveal that the energy consumption of wireless devices in an idle state is only slightly smaller than that in a transmitting or receiving state.Therefore,it is in general desirable to turn the radio oﬀwhen it is not in use,termed as powerq The work is sponsored in part by MURI/AFOSR under contract number F49620-00-1-0330and National Science Foundation grant number ANI-0081308.*Corresponding author.E-mail addresses:zheng4@(R.Zheng),rhk@(R.Kravets).1570-8705/$-see front matterÓ2003Elsevier B.V.All rights reserved.doi:10.1016/j.adhoc.2003.09.008Ad Hoc Networks3(2005)51–68/locate/adhocmanagement.Motivated by these observations, several energy conservation protocols[2,3]have been proposed to take advantage of the route redundancy in dense ad hoc networks by turning oﬀdevices that are not required for global network connectivity.However,in these protocols,the decision about which set of nodes to leave on is only based on geographical/topological informa-tion,thus is oblivious to the actual traﬃc load in the network.Since many applications of ad hoc networks are data-centric,maintenance of global connectivity is costly and unnecessary when no traﬃc or only localized traﬃc is present in the network.Various techniques,both in hardware and software,have been proposed to reduce energy consumption for mobile computing devices in wireless LANs[4,5].In contrast,power manage-ment in ad hoc networks is a more diﬃcult prob-lem for two reasons.First,in ad hoc networks,a node can be both a data source/sink and a router that forwards data for other nodes and partici-pates in high-level routing and control protocols. Additionally,the roles of a particular node may change over time.Second,there is no centralized entity such as an access point to control and maintain the power management mode of each node in the network,buﬀer data and wake up sleeping nodes.Therefore,power management in ad hoc networks must be done in a distributed and cooperative fashion.A major challenge to the de-sign of a power management framework for ad hoc networks is that energy conservation usually comes at the cost of degraded performance such as lower throughput or longer delay.A naive solution that only considers power savings at individual nodes may turn out to be detrimental to the operation of the whole network.In this paper,we propose an on-demand power management framework targeting generic ad hoc networks.To achieve reduced energy consumption while maintaining eﬀective communication,our framework integrates routing information from on-demand ad hoc routing protocols and power management capabilities from the MAC layer. Energy conservation is achieved by judiciously turning on and oﬀthe radios of speciﬁc nodes in the network.The novelty of our framework is that such power management decisions are driven by active communications in the network.For the purpose of energy conservation,connectivity is only main-tained between pairs of senders and receivers and along the route of data communication.Transitions between power management modes for each node are associated with a soft-state timer that is established and refreshed by data and con-trol messages in the network.Once the soft state is established,subsequent data delivery can be expedited without incurring additional delays from waking up sleeping nodes along the route.The length of the soft-state timer reﬂects the adap-tiveness of the power management framework to variations in traﬃc load.Since the operations of transmitting to a sleeping node and an active node are diﬀerent,we present mechanisms to discover a neighborÕs power management mode.In this con-text,neighbor discovery is challenging because a node in power-save mode cannot monitor the channel consistently.Therefore,any neighbor information may be ambiguous.This situation is even worse if nodes are mobile.Our framework is not limited to any speciﬁc routing or MAC protocols.This extensibility is a key beneﬁt of our design since it enables the use of our framework in various scenarios and allows the integration of new protocols as they become available.To verify our framework,we present a prototype using the IEEE802.11MAC protocol and evaluate it using Dynamic Source Routing (DSR)[6]and greedy geographical forward-ing protocol in the ns-2[7]simulator.Under a wide range of traﬃc patterns and load,our pro-totype achieves40–60%savings in power con-sumption as compared to a network without power management.In addition,our proto-type minimally increases latency during the initial setup stage,but achieves an average latency com-parable to a network without power manage-ment.The rest of the paper is organized as follows. Weﬁrst layout the design space for power man-agement protocols in ad hoc wireless networks and give a brief overview of existing approaches in Section2.Then we discuss how each approachﬁts into the design space.In Section3,we present the building blocks and technical details of our on-52R.Zheng,R.Kravets/Ad Hoc Networks3(2005)51–68demand power management framework.Section4 describes a prototype based on IEEE802.11 MAC.Extensive simulation results are presented in Section5.Finally,we conclude the paper and discuss future extensions in Section6.2.Design spacePower management in ad hoc networks spans all layers of the communication protocol stack. Each layer has access to diﬀerent types of infor-mation about the communication in the network, and thus uses diﬀerent mechanisms for power management.The MAC layer does power man-agement using local information while the network layer can take a more global approach based on topology or traﬃc characteristics.In this paper,we consider power management approaches that save energy by turning oﬀthe radios of nodes in the network.Other energy conservation mechanisms such as topology control and power controlled MAC protocols[8–10]are considered orthogonal and the beneﬁts can be combined.Similar to ad hoc routing protocols,power management schemes range from proactive to reactive.The extreme of proactive can be deﬁned as always-on(i.e.all nodes are active all the time) and the extreme of reactive can be deﬁned as al-ways-oﬀ(i.e.all nodes are in power saving mode by default)(see Fig.1).Given the dynamic nature of ad hoc networks,there needs to be a balance be-tween proactiveness,which generally provides more eﬃcient communication,and reactiveness, which generally provides better power saving.In this section,we outline the design space of power management in ad hoc networks and de-scribe where existing approachesﬁt into this de-sign space based on their adaptability to network traﬃc.2.1.MAC layer approachesAt the MAC layer,power management deci-sions are made based on local information.The time scale for power management can be per-packet or a short time interval.Such approaches are limited by the lack of access to information about the topology and traﬃc in the network.The PAMAS power-saving medium access protocol[11]turns oﬀa nodeÕs radio when it overhears a packet not addressed to it.The eﬀec-tiveness of PAMAS is limited to reducing the power consumption of processing unnecessary packets.Note that PAMAS alone can be consid-ered a proactive approach to power management, however it may be combined with most high level power management schemes that aim to reduce idle time energy consumption.The IEEE802.11MAC provides low-level support for power management such as buﬀering data for sleeping nodes and synchronizing nodes to wake up for data delivery.The network inter-face hasﬁve physical states:transmitting,receiv-ing,idle,sleeping and completely power-oﬀ. Energy consumption in the sleeping state is sig-niﬁcantly less than in the transmitting/receiving/ idle state.In the IEEE802.11speciﬁcation,a node can be in one of two power management modes, active mode(AM)or power-save mode(PS).In active mode,a node is awake and may receive data at any time.In power-save mode,a node wakes up periodically to check for incoming traﬃc.The transition between power management modes is left to higher-level power management protocols and is unspeciﬁed in the documentation.STEM[12]proposes a similar approach to the IEEE802.11power management,but uses an independent control channel to avoid the clock synchronization needed by IEEE802.11.STEM uses asynchronous beacon packets in a second control channel to wakeup intended receivers. After transmissions have ended(e.g.after a time-out,etc.),the node turns its radio oﬀin the data channel.Similar to IEEE802.11,sleeping nodes with traﬃc destined for them are woken upon R.Zheng,R.Kravets/Ad Hoc Networks3(2005)51–6853demand,but decisions about when a node should go back to sleep are based on local informa-tion.STEM does not provide mechanisms for indicating the power management state of a node. Instead,the power management state is only maintained on a per-link basis between nodes with active communication.Therefore,it is possible that an initiator node will experience the delay from waking up a receiver node,even if the re-ceiver is already awake due to recent communi-cation with a third node.In S-MAC[13],the authors propose a mecha-nism called message passing that modiﬁes a net-work allocation vector(NAV)for virtual channel reservation in IEEE802.11MAC type of proto-cols.The length of NAV is determined by the duration of a burst of messages.The virtual res-ervation serves two purposes:(1)it mandates the receiver to remain on throughout the transmission of the burst,and(2)it prevents other nodes from transmitting during this interval.Though message passing may be desirable for certain types of applications for sensor networks,it can be ineﬃ-cient for more generic ad hoc networks.Addi-tionally,the reservations may cause some nodes to be starved.A pure MAC layer approach as speciﬁed by the IEEE802.11MAC(i.e.nodes are always in power-save mode)can be considered as the most reactive approach to power management in our design space.In Section5,we demonstrate that a network that relies solely on the IEEE802.11MAC for power management can be highly ineﬃcient even though some communication is still possible.As future research,we will investigate the interaction between intelligent MAC layer approaches,such as STEM and S-MAC,with our on-demand frame-work.2.2.Connected dominating set approachesAt the network layer,power management schemes can take advantage of topological infor-mation.The connected dominating set approaches use neighborhood or global information to decide the set of nodes that form a connected dominating set(CDS)for the network,where all nodes are either a member of the CDS or a direct neighbor of one of the members.Nodes in the CDS serve as the ‘‘routing backbone’’and remain on all the time to maintain global connectivity.All other nodes can choose to sleep if necessary.CDS approaches such as GAF[3]and SPAN[2] conserve energy by reducing routing redundancy in dense networks.Selection and maintenance of the CDS requires local broadcast messages that may consume a signiﬁcant amount of energy[1].In addition,regardless of whether or not traﬃc is present in the network,all backbone nodes must be on all the time.Therefore,CDS approaches can be categorized as proactive.Based on these observations about MAC layer and CDS approaches,we propose an on-demand power management framework to explore the de-sign space between proactive and reactive by adapting to the traﬃc characteristics inside the network.3.On-demand power managementThe goal of on-demand power management is to base power management decisions on traﬃc patterns in the network.By reacting to changes in these patterns,nodes that do not carry any traﬃc can be dispensed from consuming a signiﬁcant amount of energy.Varying the adaptiveness to network load in our protocol balances the trade-oﬀbetween latency,throughput and energy con-sumption.The key idea of our on-demand power management framework is that transitions from power-save mode to active mode are triggered by communication events such as routing control messages or data packets and transitions from active mode to power save mode are determined by a soft-state timer.The soft-state timer is re-freshed by the same communication events that trigger a transition to active mode.A node keeps track of its neighborsÕpower management mode either by HELLO messages or by snooping transmissions over the air.For direct unicast messages,if the next hop is in active mode,the message is delivered immediately as allowed by the queuing discipline.54R.Zheng,R.Kravets/Ad Hoc Networks3(2005)51–683.1.A cross-layer design for power managementPower management in ad hoc networks can beneﬁt from a cross-layer design that leverage both network layer and MAC layer information. Knowledge about route setup and packet for-warding can provide hints about when power management should be performed.Since the route discovery phase of on-demand routing protocols determines the path subsequent packets will fol-low,nodes along this route should be as responsive as possible.On the other hand,any eﬀective power management protocol requires a mechanism to awaken a sleeping receiver when packet delivery is imminent.This is usually handled by low-level mechanisms at the MAC or physical layers. Higher-level power management techniques can beneﬁt from information about and access to the mechanisms used to provide such services.Our power management framework leverages the capability of modern MAC protocols,such as the IEEE802.11MAC,to switch power manage-ment states of nodes and buﬀer data if necessary for sleeping nodes.It also uses routing information to decide when to turn nodes on and oﬀ,which ties energy consumption with active communication in the network.The interaction of the power man-agement module with other communication layers is illustrated in Fig.2.3.2.Power management mode and state transitionIn our framework,a node can be in one of two power management modes:active mode(AM)and power-save mode(PS).In active mode,a node is awake and may receive data at any time.In power-save mode,a node is sleeping most of the time and wakes up periodically to check for pending mes-sages.Packets destined to a node in power-save mode will experience delay on the order of the length of the sleeping cycle.Transitions from power-save mode to active mode are triggered by communication events in the network.Transitions from active mode to power-save mode are determined by a soft-state keep-alive timer.Initially,all nodes are in power-save mode.Upon reception of packets,a node starts the keep-alive timer and switches to active mode.Timer values depend on the type of packet received.Upon expiration of the keep-alive timer, a node switches from active mode to power-save mode.If all packets trigger a node to stay awake with a keep-alive timer on the order of the network lifetime,our scheme degenerates to an always-on network without power management.On the other hand,if the keep-alive timer is always set to zero, our framework degenerates to the most reactive MAC layer approach discussed in Section 2. Therefore,the choice of diﬀerent keep-alive timer values varies the reactiveness of the protocol and strikes diﬀerent trade-oﬀs between energy con-sumption and data delivery eﬃciency.In an ad hoc network,if a path is going to be used,the nodes along that path should be awake as to not cause unnecessary delay for data trans-mission.If a path is not going to be used,the nodes should be allowed to sleep.During the lifetime of the network,diﬀerent messages will indicate diﬀerent levels of‘‘commitment’’to using a path.Knowledge of the semantics of such mes-sages can help make better power management decisions,which is a missing piece in most MAC layer power management approaches.On one end,most control messages(e.g.link state in table-driven ad hoc routing protocols, location updates in geographical routing,route request messages in on-demand routing protocols etc.)areﬂooded throughout the network and provide poor hints for the routing of data trans-missions.Such control messages should not trigger a node to stay in active mode.On the other end, data transmissions are usually bound to a path on relatively large time scales.Therefore,data transmissions are a good hint for guidingpower R.Zheng,R.Kravets/Ad Hoc Networks3(2005)51–6855management decisions.For data packets,the keep-alive timer should be set on the order of the packet inter-arrival time to ensure that nodes along the path do not go to sleep during active communica-tion.There are also some control messages,such as route reply messages in on-demand routing pro-tocols and query messages in sensor networks,that provide a strong indication that subsequent pack-ets will follow this route.Therefore,such messages should trigger a node to switch to active mode.The time scale of the keep-alive timer for such a tran-sition should be on the scale of the end-to-end de-lay from source to destination so the node does not transition back to power save mode before theﬁrst data packet arrives.One important feature of the keep-alive timer is that it is refreshed on demand.Whenever a node receives a routing message or a data packet,it sets the timer with the maximum of what is left for the current keep-alive timer and the value associated with the received message.Therefore,only per-node,instead of per-ﬂow,information is needed for power management.Ideally,the keep-alive values for data packets should be larger than the inter-arrival time of data packets.In reality,since a node can be both data source/sink and for-warding router simultaneously,its keep-alive timer is an aggregation of various timer values,i.e.,old timer will be extended when new communication events arrive.Therefore,the performance is quite insensitive to the choice of these timer values. 3.3.Obtaining neighbors’power management modeSince communication with a neighbor is only possible if the neighbor is in active mode,it is necessary for nodes to track power management modes of neighbors.In our framework,each node maintains a neighbor list that caches a neighborÕs power management mode and a time-stamp of the most recent update from this neighbor.A neighborÕs power management mode can be discovered in two ways.Theﬁrst way is through explicit local HELLO message exchanges with piggybacked information about the power man-agement mode of a node.HELLO messages should be transmitted atﬁxed intervals regardless of the power management mode of a node.Link failure is assumed if no HELLO messages have been received during successive intervals,since the loss of only one HELLO message may have been caused by a broadcast collision.The second way to discover a neighborÕs power management mode is via passive inference. Depending on the capability of the hardware and the MAC protocol,a node may be able to operate in promiscuous mode and passively snoop mes-sages in the air.With MAC layer support,a nodeÕs power management mode can be piggybacked in the control header of MAC layer data units.There are two challenges to using passive inference.First, nodes in power-save mode cannot hear messages from their neighbors and so do not have a good basis for determining the power management mode of their neighbors.Second,nodes in power-save mode may not be transmitting and so their neighbors will have diﬃculty diﬀerentiating nodes that are in power-save mode from nodes that are away or dead.Therefore,special care must be taken to distinguish between nodes that move away from ones that are in power-save mode.Since the use of HELLO messages is expensive, our framework uses two types of indicators for such passive inference.Theﬁrst indicator is a lack of communication during a time interval.When no communications have been observed from a node that was in active mode,the neighbor is assumed to be in power-save mode.The value of the this interval should be based on the keep-alive timer since the length of keep-alive timer indicates the maximum amount of time a node commits to be in active mode when no messages are received.If a node does not hear from its neighbor during the keep-alive period,it is very likely that either that neighbor moved away or it has switched to power-save mode.The other indicator is packet delivery failure to the neighbor(e.g.indicated by an RTS retry time out in IEEE802.11).Based on the observed power management mode of the neighbor,a packet delivery failure is treated in two stages.First,if the neighbor was originally in active mode,it is con-sidered to have switched to power-save mode. Second,if the neighbor was originally in power-save mode,it is now considered unreachable.Data for this node is discarded at the MAC layer.The56R.Zheng,R.Kravets/Ad Hoc Networks3(2005)51–68rationale for this two-stage process is that transi-tion to an intermediate stage provides a second chance to salvage data for a neighbor that has switched to power-save mode since the last update.Compared to using HELLO messages,passive inference does not rely on additional control messages,which is more desirable from an energy conservation perspective.However,the ambiguity of link failure and the power management mode of a neighbor can result in delayed data transmission, but as will be shown in Section5,this approach in general works well.4.A prototype based on the IEEE802.11MACIn this section,we present a prototype of our framework based on the IEEE802.11MAC.First, we give a brief overview of IEEE802.11power management functions and then we discuss the implementation details of our prototype.Note that our on-demand framework can be easily implemented over other MAC protocols including those using asynchronous wakeup mechanisms [14].4.1.Overview of IEEE802.11power management in ad hoc networksIn the IEEE802.11speciﬁcation,all nodes in the network are synchronized to wake up period-ically.Broadcast/multicast messages or unicast messages to a power-saving node areﬁrst an-nounced during the period when all nodes are awake.The announcement is done via an ad hoc traﬃc indication message(ATIM)inside a small interval at the beginning of the beacon interval called the ATIM window.During the ATIM window,nodes that have buﬀered data for sleeping nodes transmit an ATIM management frame that contains the iden-tity of the intended receivers.If a node receives a directed ATIM frame during the ATIM window (i.e.it is the designated receiver),it sends an acknowledgment and stays awake for the entire beacon interval waiting for the data to be trans-mitted.Broadcast/multicast messages announced in the ATIM need not be acknowledged.Imme-diately after the ATIM window,nodes can trans-mit buﬀered broadcast/multicast frames,data packets and management frames addressed to nodes that have acknowledged a previously transmitted ATIM frame.Following the trans-mission of all buﬀered data,nodes transmit data destined to nodes that are known to be in the ac-tive state for the current beacon interval.In IEEE 802.11,a nodeÕs power management status is indicated in the frame controlﬁeld of the MAC header for each packet.4.2.Our prototypeWe experiment with on-demand routing pro-tocols as DSR[6]or AODV[15]as well as stateless routing protocols such as greedy geographical routing protocols.Unless otherwise speciﬁed,the results are reported with using DSR protocol.The complete state transition diagram is shown in Fig. 3.Transitions between power-save and active mode are triggered by packet arrivals and expiration of the keep-alive timer.Sub-state tran-sitions inside power-save or active mode indicate the physical state of the node and are controlled by the IEEE802.11MAC power management func-tions.To maintain the neighbor list,the prototype uses passive inference to update neighborsÕpower management modes and link states.Nodes snoop transmissions in the air when they are awake and update their neighbor lists based on the control ﬁeld of the MAC header of packets.Entries for unreachable neighbors are purged periodically. Due to the use of beacon messages,changes in link availability can be detected proactively as follows.Let the beacon interval be I and let the degree of a node be bounded by d.The interval N(mea-sured in units of beacon intervals)of two succes-sive beacon messages sent by node n follows the geometric distribution N$ð1=dÞð1À1=dÞNÀ1 with mean d.Therefore,if a node has not heard any beacon messages from a particular neighbor for more than cÁd beacon intervals,where c is a protocol-speciﬁc constant,the node is likely away or‘‘dead’’.The degree of a node is obtained from the neighbor list and a node can use its own degree to approximate its neighborÕs degree.This methodR.Zheng,R.Kravets/Ad Hoc Networks3(2005)51–6857can be combined with events of packet delivery failure to better infer the availability of a link be-tween neighbors.The major beneﬁt of inferring a neighbor Õs state by snooping is that it does not incur out-of-band control messages and therefore scales to large networks.Table 1lists various timers and values for dif-ferent messages.To determine the values of vari-ous keep-alive timers,consider a k -hop route from source n 0to sink n k .Suppose the one-way delay from node n 0to node n i on this route is d i .The inter-arrival time of data packets at the source is 1=k .Let the beacon interval be I .Therefore,the time it takes for the route request message to reach node n i from n 0is i ÃI þd i on average.The time it takes node n i to receive the route reply message is ð2k þ1Ài ÞÃI þd k þ1þd i under the assumption of symmetric routes.Assuming the data source will immediately transmit the data upon reception of the route reply message,the time between the reception of the route reply message and the reception of the ﬁrst data packet at node n i isi ÃI þ2Ãd i .Finally,assuming no additional queuing at intermediate nodes,the packet inter-arrival time at node i is 1=k .Therefore,on the pessimistic side,the length of the keep-alive timers for diﬀerent message types should be chosen to be larger than these estimates to ensure low latency for data delivery.If available,information about the network dimensions and traﬃc patterns can be used to select these values based the above discussion.In our performance evaluation in Section 5,we do not assume availability of such knowledge.In our implementation,we set RTRQ_KEEPALIVE to 0,RTRL_KEEPALIVE to 5s,DATA_KEEPALIVE,SRC_KEEPALIVE and DST_KEEPALIVE to 2s.The REFRESH_INTERVAL is set to 5s.Since keep-alive timer values can be aggregated,the performance is quite insensitive to the choice of these values.As part of our future work,we will investigate techniques to adapt keep-alive timer values based on measure-ments in the network.5.Performance evaluationWe implemented our prototype in the ns-2[7]network simulator using the CMU wireless extension [16].To evaluate the eﬀectiveness of our proposed scheme,we conducted several simula-tions using diﬀerent traﬃc models in both static and mobilenetworks.Table 1Timers and messages Message typeValueRoute request RTRQ_KEEPALIVE Route replyRTRL_KEEPALIVE Data at intermediate node DATA_KEEPALIVE Data at source SRC_KEEPALIVE Data at sinkDST_KEEPALIVE58R.Zheng,R.Kravets /Ad Hoc Networks 3(2005)51–68。

cluster list 原理 -回复

cluster list 原理-回复
什么是Cluster List，其原理是什么？
Cluster List 是Redis 数据库中一种数据结构，它能够存储多个字符串类型的元素，这些元素被称为节点，而这些节点又被称为集群。

Cluster List 的实现原理是将多个节点连接在一起，形成一个链表。

每个节点都具有一个指向前一个节点和后一个节点的指针，这样整个链表就能够连接起来形成一个完整的集群。

Cluster List 的实现中，每个节点的数据都包含三部分：元素值、前驱指针、后继指针。

其中元素值是存储在节点中的实际数据，而前驱指针和后继指针则用来指向前一个节点和后一个节点，以实现链表的功能。

Cluster List 的主要特点是能够高效地插入和删除节点，其时间复杂度为O(1)。

这可以通过将新节点插入到已有节点的前面或后面来实现。

同时，Cluster List 还支持节点的随机访问，这是由于每个节点都有一个索引值，可以根据索引值来定位节点的位置，其时间复杂度也为O(1)。

Cluster List 在Redis 中的应用非常广泛，主要用于存储链表结构数据的场景，如任务队列、消息队列、日志文件等。

另外，Cluster List 还经常与Redis 的发布订阅功能一起使用，这样就可以实现集群通信和任务调
度等功能。

总之，Cluster List 是Redis 中一种高效的数据结构，它的实现原理非常简单，同时具有插入、删除和查找等强大的功能，因此在许多应用场景下表现出色，为Redis 数据库的高效处理提供了很好的支持。

判定非对称选择网活性及活性单调性的一个算法

判定非对称选择网活性及活性单调性的一个算法
宋文;陆维明
【期刊名称】《计算机科学》
【年(卷),期】2005(032)009
【摘要】活性是Petri网的重要行为特征之一.为了得到判定AC网活性有效的算法,本文利用分治的思想,在定义极小死锁的前、后归约子网的基础上,将较大问题分而治之,把未知问题转化为已知的FC网上的问题,从而得到了判定AC网活性及活性单调性的多项式时间的算法.
【总页数】4页(P18-20,57)
【作者】宋文;陆维明
【作者单位】中国科学院数学与系统科学研究院,北京,100080;西华大学计算机与数理学院,成都,610039;中国科学院数学与系统科学研究院,北京,100080
【正文语种】中文
【中图分类】TP3
【相关文献】
1.可分解非对称选择网的活性和家态 [J], 林贵献;陆维明
2.可分解非对称选择网的活性和有界性 [J], 徐静;陆维明
3.非对称选择网活性的一个多项式时间判定 [J], 焦莉;陆维明
4.扩展强化非对称选择网的活性和有界性 [J], 焦莉;陆维明
5.新扩展强化非对称选择网的活性 [J], 宋文;伊良忠;严兵
因版权原因，仅展示原文概要，查看原文内容请购买。

OLSR协议基本原理

8
12 OLSR 协议内
容
2.1 OLSR协议核心——MPR机制：节点选择部分邻节点作
为它的中继节点，只有被选择的中继节点转发节点的控制消中息继。节
N
P
N
点 P
M B
Q
M
B
Q
LF
C
GL
F
C
G
A
A
K
E
D
J I
纯扩散机
HK
E
D
H
J I
MPR机9制
12 OLSR 协议内
容
多跳范围内的消息转发仍然遵循MPR机制
T_dest T_last
Байду номын сангаас
C
\
BC
AB
TC更新后
这里令B发送TC 分组更新
12 OLSR 协议内
容 2.4.4 路由计算
OLSR路由协议采用Dijkstra最短路径选路算法进行选路。 22
2 O L S RO协pti议mized Link State Routing Protocol
OLSR协议优点和局限
17
12 OLSR 协议内
容 2.3.3 拓扑表
网络中的每一个节点维护一张拓扑表，记录从TC分组中得到的拓扑信息，并由此信息计算路由。节点将网络中其他节点的多点中继信息作为拓扑条目记录在拓扑表中。
拓扑条目格式
此条目说明了T_dest已经选择T_last作为MPR，而且 T_last已经发布了序列号为T_seq的MPR Selector 集信息。T_time作为保持时间，过期就删除该条目。
OLSR协议
2024年10月 15日 1
CONTENTS

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)。

Research on Clustering Routing Algorithms in Wireless Sensor NetworksQing Bian, Yan Zhang, Yanjuan ZhaoDepartment of Media Technology and CommunicationNortheast Dianli UniversityJiLin, Chinapfazy@, zhaoyanjuan1984@Abstract—compared with the traditional wireless networks, wireless sensor networks have energy constraints, low-data-rate of high redundant and data flow of high-to-one, and so on. Energy effectiveness is the key performance indicators of wireless sensor networks. Based on the analysis of energy management strategy in the wireless sensor networks,the main factors affecting energy consumption are:perceptual data, data processing and radio communications,the radio communication is the main part of energy consumption. In the wireless sensor networks, the realization of energy-efficiency could be improved while in the different layers of communication protocol stack. However, as the basis of the limitations of the physical layer, the improvement is focus on design and implementation of network-layer protocol.The researchers agreed that the clustering of nodes in wireless sensor networks is an effective program of energy conservation.This article dedicated to research the clustering routing algorithm in wireless sensor networks.Keywords-Wireless Sensor Networks；clustering routing algorithm；LEACH routing algorithmI.I NTRODUCTIONThe concept of clusters in wireless sensor networks was first proposed in the packet radio network, which was mainly used for hierarchical routing. With the continuous deepening of the study, to date, for constructing and maintaining the sub-cluster network structure, we had proposed a large number of clustering algorithms. Clustering-based network can reduce the cost of routing algorithm and the flooding broadcast. So we can easy to manage mobile nodes and control the channel access, and can improve the efficiency of network resources.(1) The effects of clustering structure in wireless sensor networksSpecifically, clustering in wireless sensor networks has the following functions:·Complete the routing function. Network organizational structure which is generated by clustering algorithm is the basis of the routing algorithm. First clusters are acted as a router, which maintenance and distribute of the routing information. Ordinary node will pass the needing groups to the first cluster node. In the sub-cluster structure it can also use the hierarchical routing algorithm, which priori routing algorithm is be used in the cluster. In the hierarchical routing algorithm node maintains the completing information within the other nodes.And in this algorithm it used the reactive inter-cluster routing protocol to reduce the communication and cost of the routing.·Node management such as the reusing of resources, bandwidth, QoS provisionAnother challenge of wireless sensor networks is how to allocate resources efficiently and thus can order the bandwidth by the way of quantitative or statistical.In cellular networks the resources allocation is relatively easy to be achieved,directly or through other communication nodes in the base station, each mobile node can obtain the bandwidth requirements, that is, it can carry out resource allocation through the base station.However, if the network is divided into the hierarchical structure which is based on the clusters, then it will extend the method of the cellular networks to the wireless sensor network. Within each cluster,the first cluster node can control business requirements and rational allocation of the bandwidth.·The role of data fusionThe first cluster can act as a data processing center; ordinary node will send the monitoring information to the first cluster node. The first cluster node will compress and fuse redundant information,so as to effectively reduce the network load and extend the network lifetime.(2) The performance index of clustering algorithmsAccording to the system requirements of the network clustering algorithm divided the network into several clusters. The basic performance indicators are as follows:·Algorithm should be fully distributed.Each node onlydepends on local information to make decisions independently.·Algorithm should have the low time complexity andmessage complexity.·After the algorithm runs,nodes in the network shouldbe uniquely identified,or cluster head, or ordinary node.·Algorithm should have good stability.In the sub-clusternetwork structure,dynamic movement nodes can frequently join or leave a cluster, because the battery power runs out, some nodes can lead to changes in network topology and thus affect the stability of the system.More serious, intense mobility of the cluster nodecan sometimes result in the update of first cluster nodeand reconfiguration of the network, frequent update of thefirst cluster node will introduce a larger cost of the computation and communication. This can seriously2010 International Conference on Intelligent Computation Technology and Automationaffect the performance of other network protocols, such as packet scheduling, routing and resource management. ·Clustering algorithm should be simple and efficient. The goal of the clustering algorithm is to construct a cluster set which can cover the whole user nodes, and can better support for resource management and interconnected of the routing protocols. In order to reduce communication and computational overhead this is caused by clustering algorithm, when have only a few nodes slower movement and topology changes, clustering mechanism should try to maintain the original structure. Thereby it can reduce the cost of the introducing of the re-generation clusters to introduce and improve the overall effectiveness of the network.II.S UB -C LUSTER R OUTING A LGORITHMIn the existing clustering algorithm of the wireless sensor network, LEACH algorithm is representative.1) LEACH algorithm processLEACH is low-power, adaptive and clustering routing algorithm, which is made for WSN by Chandrakasan in MIT. And compared to the general flat multi-hop routing protocols and static clustering algorithm, LEACH can extend the network life-cycle of 15%, mainly through the randomly selected sub-cluster leader and shared equally between the relay communications services to achieve the network life-cycle. LEACH defines the concept of "round". And one round consists of two phases which is the initialization phase and stability phase. In order to avoid the extra processing cost, the stabilization phase lasting a relatively long time.In the initialization phase, cluster head election process is as follows: Sensor node generates a random number range 0-1. If this random number is less than the threshold T (n), then it releases the information that he is the cluster head node to the nodes within the cluster. In each round of circulation, if the node has been elected as cluster head, then put the T (n) to 0, so that the node will not be elected as cluster head in the next round of re-elected. For the nodes, which had not elected as cluster head node, it will be elect as the cluster head node by the probability of T (n). As the number of elected cluster head nodes increased, the threshold T (n) is even greater for the remaining node which had not be elected as the cluster head node, the probability of generating the random number which is less than T (n) is greater, therefore, the probability of one node to be cluster head node is greater. When only a node is not elected, T(n) = 1, so this node will be elected. T (n) formula can be expressed as:Where P is the percentage of cluster head node in all nodes, r is an election rounds, r mod (1 / P) is on behalf of the number of node which was elected as cluster head nodes in the cycle, G is the node set which is not elected as a cluster head node in this cycle.After node is selected as cluster head, it will broadcast the information that he is the cluster head to the rest of the nodes in the same cluster. The remaining nodes decide to join the cluster according to the size of the received signal, and return the join signal. When the cluster head receives all join messages, it will allocate TDMA time slot information to all the nodes in the same cluster, notice nodes within the same cluster to send a TDNA message to the cluster head in its own time slot. In order to avoid signal interference near the cluster, cluster head can determine the CDMA codes which all nodes used. The CDMA codes which is used in the current phase and TDMA timing information will be sent together. When nodes within the cluster receive the message, they will send data to the cluster head in their own time slot. Algorithm will enter a stable phase.Work in a stable phase, member nodes continuous collected monitoring data, and send data to the cluster head node in their own time slots. While the other time, it can turn off the radio module, into hibernation, and it is one of the main ways to save energy for LEACH. After the cluster head node received the data which is from its member, it will do the necessary processing of data fusion. Then the information is sent to the sink node. After a period of time of data transmission, nodes enter to a new work round, to re-select a new cluster head, and constantly circulating.2) Advantages and disadvantages of the LEACH algorithmIt can be seen that LEACH algorithm as a typical sub-cluster routing protocol has the following advantages: · The hierarchy, path selection and routing information is relatively simple, and the sensor nodes do not need to store large amounts of routing information, and do not need complex functions.·The cluster head node is randomly selected, the opportunity of each node is equal, and the load of whole network is balance.LEACH algorithm randomly selected cluster-heads; it can evenly distribute the high energy consumption to all nodes on the network by the using of the rotation of the election. So that nodes, which have no energy, can be randomly distributed. And the LEACH algorithm uses hierarchical structure; Cluster-heads reduce the energy consumption of data transmission through the data fusion mechanism, and therefore compared to the general multi-hop routing protocols and static clustering algorithm, LEACH can extend the network life-cycle of 15%. However, there are a number of deficiencies in LEACH algorithm, such as:Because the cluster head in LEACH protocol are randomly generated, energy consumption can be evenly⎟⎠⎞⎜⎝⎛×−P r P P1mod 1 if n ∈G0 otherwisedistributed in the network; however, it ignores residual energy of nodes, geographic location and other information in the election of cluster head node. So it can easily lead to exhaust the energy quickly in cluster head nodes.LEACH assumes that all the nodes can be directly communicate with the cluster head node and the base station node, the actual network of base stations are usually far away from the sensing area,this would make the cluster head which is far from the base station is easier to fail. Therefore, expansion of the network is not strong, and is not suitable for large networks.Because the distribution of cluster head is totally dependent on the random number, so the number of the cluster-heads can be big at a regional, and the number of the cluster-heads will be little at other regional. In the cluster-heads centralized regional, the number of the general node is very little, and this can have lost the meaning of hierarchical routing; in cluster-heads sparse region, cluster head node is responsible for too much data, and the distance from the cluster head is far, transmitted signal energy consumption is too large.As shown in Figure 1 is the two kinds of scenarios of the cluster head nodes in sensor networks. In the first round of the bad case scenario, all the cluster head nodes are distributed in the right circles, while in the second round, the entire cluster head node are also distributed in the left border. So when a node within cluster transmits messages to the cluster head node, the energy consumption is high. According to data from thesis, so when transmitting messages to all nodes, the energy consumed by the sum of 3215nJ/bit. In the good case scenario, although each one of the cluster head node is not the best distribution, but relatively uniform. The close distance between cluster head node and node within cluster ensures the average energy consumption of this scenario is smaller than the bad scenario, to 905nJ/bit.The balance mechanism of energy consumption in LEACH protocol requires that the initial energy of all nodes isthe same.Figure 1. A bad-case-scenario and a good-base-scenario example in LEACH3) The improvement of the LEACH algorithmBased on the shortage of above algorithm in LEACH, we can improve the LEACH algorithm that is mainly from three aspects of the LEACH algorithm such as the cluster shape, clustered approach and the choice of the cluster head node.In the LEACH algorithm, it will produce the cost in the process of the clusters establishment. If you can reduce the cost of this part, then it can make more energy to be consumed on the data transmission. The improved algorithm uses a fixed sub-cluster approach, that is, in the initial stage, after the clusters are divided, the node within the cluster will no longer be changed; meanwhile, in order to make energy consumption evenly distributed across all nodes, it need to rotate cluster head within each cluster. Cluster head node has an important position in wireless sensor networks; the energy consumption of cluster head node is much higher than normal sensor nodes. The energy consumption of cluster head node, including nodes within the cluster head and cluster communication, cluster head and cluster node, the energy consumed by communications The energy consumption of cluster head node, including the energy consumption when cluster head node communicates to nodes within cluster, and when cluster head node communicates to aggregation nodes; for the cluster node which is near to the aggregation nodes, because of the large distance, the cluster node which is far from the aggregation nodes can consume higher energy. The improved algorithm uses non-uniform sub-cluster method (as shown in Figure 3-3), that is, the convergence radius, which is near to the aggregation node, is larger; and away from the aggregation node, the convergence radius is smaller.The cluster head node, which is near to the aggregation node, consumes greater energy within the cluster than the cluster head node which is far away from the aggregation node. That can make all the energy consumed by cluster head nodes are close to the same. Thus it can balance the energyconsumption of the cluster head node.Figure 2. Non-uniform clusteringThe selection of the cluster head node needs to make minimize the energy consumption. As well as the dynamic selection cluster head node in order to avoid the energy consumption of a single cluster head node.From these two points, which are the minimized energy consumption within the cluster and the largest energy consumption in the cluster head node, we select cluster head node according the residual energy of each node.III.C ONCLUDERouting algorithm in wireless sensor networks is a very hot research topic, because it has great research significance in saving energy and prolonging network life-cycle. This paper first described the role of sub-clusters structure in wireless sensor networks and sub-cluster algorithm performance indicators; second, described the core ideas and analysis model of LEACH algorithm. The focus of this article is to study the advantages and disadvantages of the LEACH algorithm, and improve the algorithm for those disadvantages.R EFERENCES[1]J. ELSon and D. Estrin, Time Synchronization for Wireless SensorNetworks, Parallel and Distributed Processing Symposium, Proceedings 15th International, 2004.4.[2]Y.Yao and J.Gehrke, The cougar approach to in-network queryprocessing in sensor networks in SIGMOD Record, 2005.9.[3] C.Y Chong, S.Kumar, “Sensor networks evolution, Opportunities, andChallenge, ” Proceeding of the IEEE, 2003.[4]Nordman M.M, Lehtonen M, A wireless sensor concept for managingelectrical distribution.[5]Pottie G J, Wireless sensor networks. Information Theory Workshop.2008.[6]Deborah Estrin,John Heidemann,and Wei Ye, “An. Energy-EfficientMAC Protocol for Wireless Sensor Networks,” IEEE INFOCOM 2003. [7]Kottapalli V A, Kiremidjian A S, Lynch J P et al, Two-tiered wirelesssensor network architecture for structural health monitoring. SPIE’s 10th Annual International Symposium on Smart Structures and Materials. San Diego, CA, USA. March 2005.7.[8]Ramanathan R, Rosales-Hain R, Topology control of multihop wirelessnetworks using transmit power adjustment. Proc 9th Joint Conf on IEEE Computer and Communications Societies(INFOCOM)．Tel-Aviv, Israel. 2006.3.[9] B.Karp and H.T.Kung, GPRS: Greedy Perimeter statelessRouting for Wireless Networks, 2006.[10]Nieulesu D,Nath B, Ad Hod positioning System(APS) using AOA.In:Proc 22 Annual Joint Conference IEEE Computer and Communication.IEEE，Vol.3，2005.[11]S.O Doumit, D.P Agrwal, Bio-inspired mobility in environment awarewireless sensor networks[C], Pevrasive Computing and Communications, 2003,l:514-517.[12]21 Ideas for the 21st Century [J].Business Week,2008,30(08):78-167.[13] D Koller, 10 Emerging Technologies that Will Change the World[J].Technol, 2007, 106(01):33-49.[14]Mark Weiswer, Tine Computer for the Twenty-First Century [J].Scientific Americna, 2006, 20(09):94-10.[15]G Schuster, Standford Mathematical Feophysics Summer SchoolLectures:Basics of Exploration Seismology and Tomography./Stanford/ch.html. January 2008.。