Dynamic Load Balancing Based on Restricted Multicast Tree in Homogeneous Multiprocessor Systems

To decrease the cost of exchanging load information among processors,a dynamic load-balancing(DLB)algorithm which adopts multicast tree technology is proposed.The multicast tree construction rules are also proposed to avoid wrongly transferred or redundant DLB messages due to the overlapping of multicast trees.The proposed DLB algorithm is distributed controlled,sender initiated and can help heavily loaded processors with complete distribution of redundant loads with minimum number of executions.Experiments were executed to compare the effects of the proposed DLB algorithm and other three ones,the results prove the effectivity and practicability of the proposed algorithm in dealing with great scale compute-intensive tasks.     

An optimization algorithm for locomotive secondary spring load adjustment based on artificial immune

In order to control the locomotive wheel （axle） load distribution, a shimming process to adjust the locomotive secondary spring loads was heretofore developed. An immune dominance clonal selection multi-objective algorithm based on the artificial immune system was presented to further improve the performance of the optimization algorithm for locomotive secondary spring load adjustment, especially to solve the lack of control on the output shim quantity. The algorithm was designed into a two-level optimization structure according to the preferences of the problem, and the priori knowledge of the problem was used as the immune dominance. Experiments on various types of locomotives show that owing to the novel algorithm, the shim quantity is cut down by 30%-60% and the calculation time is about 90% less while the secondary spring load distribution is controlled on the same level as before. The application of this optimization algorithm can significantly improve the availability and efficiency of the secondary spring adjustment process.     

Improved single-neuron PID controller for DC motor applied in robot system

An improved single-neuron proportional integral derivative (PID) controller and a new method to build the DC motor system were presented in the article. In the simulation, the robot arm is considered as an external load to DC motor. Both the motor module and the load module are created in Simulink to achieve simulation results closer to real robot system. In this way, it can well verify the performance of the improved single-neuron PID controller, which is a combined controller of normal PID controller and single-neuron PID controller. Besides, an intelligent switcher can help to realize the function of choosing a better control algorithm according to motor’s velocity output. Simulated results confirm the rapid and stable response of the improved PID controller. Moreover, the improved single-neuron PID controller has an excellent ability to overcome the load impact and supress the jamming signals. At last, a GUI interface platform is built to make the controller easier to be applied in other robot systems.     

Numerical simulation of dense particle-gas two-phase flow using the minimal potential energy principle

A simulation method of dense particle-gas two-phase flow has been developed. The binding force is introduced to present the impact of particle clustering and its expression is deduced according to the principle of minimal potential energy. The cluster collision, break-up and coalescence models are proposed based on the assumption that the particle cluster are treated as one discrete phase. These models are used to numerically study the two-phase flow field in a circulating fluidized bed （CFB）. Detailed results of the cluster structure, cluster size, particle volume fraction, gas velocity, and particle velocity are obtained. The correlation between the simulation results and experimental data justifies that these models and algorithm are reasonable, and can be used to efficiently study the dense particle-gas two-phase flow.     

Algebraic algorithm for reliability index weighted capacity of communication network

Communication network has communication capacity and connection reliability of the links. They can be independently defined and can be used separately, and when the reliability of a communication network is analyzed from a macroscopical angle of view, it is more objective to express the performance index of a communication network as a whole. The reliability index weighted capacity is just obtained by integrating these two parameters. It is necessary to further study the algorithm to calculate the reliability index of the communication network with a complicated topologic structure and a whole algebraic algorithm is therefore proposed for calculation of the reliability index weighted capacity of a communication network with a topologic structure. The whole computational procedure of the algorithm is illustrated with a typical example.     

Graph Partitioning Method to Determine Servers Placement in CDN

To determine CDN cache servers’ placement reasonably, an idea that using graph partitioning to solve the problem was put forward through theoretical analysis and the specific algorithm of partitioning was researched. The concept of graph partitioning for CDN was defined. The conditions of graph partitioning for CDN were demonstrated: the sum of the weights of the nodes in each subarea is as close as possible; edge cut between the subareas is as large as possible; internal nodes in each subarea are connected as far as possible. By reference to light vertex matching algorithm of graph partitioning for network simulation, a multilevel k-way algorithm of graph partitioning for CDN was proposed. The maximized edge cut k-way KL refinement algorithm was discussed. Graph partitioning is a feasible way to solve the problem of CDN servers’ placement. Multilevel k-way algorithm is a feasible algorithm for CDN graph partitioning.     

Retraction control of motorized seat belt system with linear state observer

A design and verification of linear state observers which estimate state information such as angular velocity and load torque for retraction control of the motorized seat belt (MSB) system were described. The motorized seat belt system provides functions to protect passengers and improve passenger’s convenience. Each MSB function has its own required belt tension which is determined by the function’s purpose. To realize the MSB functions, state information, such as seat belt winding velocity and seat belt tension are required. Using a linear state observer, the state information for MSB operations can be estimated without sensors. To design the linear state observer, the motorized seat belt system is analyzed and represented as a state space model which contains load torque as an augmented state. Based on the state space model, a linear state observer was designed and verified by experiments. Also, the retraction control of the MSB algorithm using linear state observer was designed and verified on the test bench. With the designed retraction control algorithm using the linear state observer, it is possible to realize various types of MSB functions.     

Ground vibration analysis due to a high-speed moving load in a tunnel

The surface response of an infinite viscous-elastic half-space due to a moving load in the tunnel is analyzed. The tunnel is modeled as an inforite long Euler-Bernoulli beam without thickness and the concept of the equivalent stiffness is introduced to simulate the half-space. The inverse Fourier transformation and the relative coordinate transform are utilized to transfer a double infinite integral to a double definite integral, which improves the operational efficiency. Then, the analytic solution of the surface response of a half-space due to a moving load in the tunnel is obtained. Finally, the laws of ground vibration responses induced by moving loads in the tunnel are analyzed, considering different tunnel embedded depths and different moving speeds. Results show that the displacement distortion can be obtained by at some special velocities. A theoretical explaination of this phenomenon is provided as well.     

A genetic algorithm for the pareto optimal solution set of multi-objective shortest path problem

Unlike the shortest path problem that has only one optimal solution and can be solved in polynomial time, the muhi-objective shortest path problem （ MSPP ） has a set of pareto optimal solutions and cannot be solved in polynomial time. The present algorithms focused mainly on how to obtain a precisely pareto optimal solution for MSPP resulting in a long time to obtain multiple pareto optimal solutions with them. In order to obtain a set of satisfied solutions for MSPP in reasonable time to meet the demand of a decision maker, a genetic algo- rithm MSPP-GA is presented to solve the MSPP with typically competing objectives, cost and time, in this pa- per. The encoding of the solution and the operators such as crossover, mutation and selection are developed. The algorithm introduced pareto domination tournament and sharing based selection operator, which can not only directly search the pareto optimal frontier but also maintain the diversity of populations in the process of evolutionary computation. Experimental results show that MSPP-GA can obtain most efficient solutions distributed all along the pareto frontier in less time than an exact algorithm. The algorithm proposed in this paper provides a new and effective method of how to obtain the set of pareto optimal solutions for other multiple objective optimization problems in a short time.     

Two-way Markov random walk transductive learning algorithm简

Researchers face many class prediction challenges stemming from a small size of training data vis-a-vis a large number of unlabeled samples to be predicted. Transductive learning is proposed to utilize information about unlabeled data to estimate labels of the unlabeled data for this condition. This work presents a new transductive learning method called two-way Markov random walk （TMRW） algorithm. The algorithm uses information about labeled and unlabeled data to predict the labels of the unlabeled data by taking random walks between the labeled and unlabeled data where data points are viewed as nodes of a graph. The labeled points correlate to unlabeled points and vice versa according to a transition probability matrix. We can get the predicted labels of unlabeled samples by combining the results of the two-way walks. Finally, ensemble learning is combined with transductive learning, and Adboost.MH is taken as the study framework to improve the performance of TMRW, which is the basic learner. Experiments show that this algorithm can predict labels of unlabeled data well.     

基于混合策略的集群负载均衡算法研究

传统的负载均衡策略仅对某些类型的站点有效,而不能满足各种站点类型的要求。在分析已有方法的基础上,针对不同类型的用户请求特点,提出了一种基于混合策略的负载均衡算法。该算法通过对用户请求的分类调度来获得较高的cache命中率,同时还引入了会话保持技术和反馈环节来解决会话失效和负载不均衡的问题。实验结果表明,这种该算法能有效提高集群系统的整体性能。     

基于服务质量感知的异构无线网络负载均衡算法

为了实现合理的网络资源调度,提高异构无线网络资源利用率,提出了一种基于服务质量(Quality of service,QoS)感知的网间负载均衡算法。算法基于无线业务特征,定义了适用于异构网络的一般化终端收益函数和网络效用函数,分别用来表征终端的QoS体验和网络的负荷情况。算法迭代地把负荷最重的网络中QoS收益和资源利用效率较低的终端调度到能够提高该终端QoS收益的负荷最轻的网络中,直至达到网络负载平衡,实现网络资源的均衡利用,并有利于保证终端的QoS。仿真结果表明:算法能够有效地提高网络的吞吐量、降低业务的时延和丢包率。     

面向复杂数据应用的可扩展负载平衡中间件研究

现有的负载平衡中间件大都采用单一的基于对象的负载平衡粒度,更多地关注于同一对象组的不同副本之间的平衡,不能满足雷达仿真所涉及的复杂数据应用的要求.基于通用对象请求代理体系结构(CORBA)技术,采用基于移动对象适配器(POA)的负载平衡粒度,提出了一种高可扩展的负载平衡中间件模型,阐述了该模型相关的若干关键技术、解决方法及其实现,包括负载策略的扩展、负载度量的扩展等.实验表明:该模型能够有效降低平均响应时间,并增加总的吞吐量.     

基于模糊聚类的云计算集群资源调度算法

针对现有的云计算集群资源调度算法具有的负载不均衡和在线动态适应能力不强的缺点,提出了一种基于模糊聚类的云计算动态集群资源调度算法。首先,构建了云计算环境下的资源调度模型。然后采用模糊聚类对云计算集群资源进行聚类,根据节点与所有聚类中心的距离判断是否需要增减聚类数量。当新任务到来时,自动计算其到各个聚类中心的距离,将具有最小聚类距离的聚类中心分配给该任务。在Cloudsim环境下进行仿真试验,结果表明该方法能有效地实现云计算集群资源的动态调度,且较其它方法相比,具有反应实时和负载均衡的优点,是一种适合云计算环境的可行任务调度方法。     

一种新的无线Ad hoc网络群头生成算法

从网络安全的角度出发提出了一种新的群头选择算法,并结合相应的负载平衡措施改善该算法的性能。该算法以图论为理论背景,使用Kruskal算法求出无线Ad hoc网络拓扑结构的最小生成树,在最小生成树上生成群,确保群内结点间通讯的代价保持在一个较低的水平。该算法采取的负载平衡措施最大限度地延长了群头的生命周期,并可在新老群头交替时保持整个网络的稳定性。     

Fuzzy clustering based load balancing algorithm with feature weighted

Focusing on the data fusion problem of various loads, a fuzzy clustering based load balancing algorithm with feature weighted is proposed. First of all, various system resources are considered as dimensions for load metrics, and features for different dimensions are weighted so as to quantify comprehensive loads; then, this algorithm introduces fuzzy clustering, optimizes weight constraints, and adds penalty terms. Hence, the most suitable objective node cluster for load transferring is resolved through fuzzy clustering. Experimental results show that this algorithm can effectively fuse multidimensional load data and reduce standard deviation for node loads within the cluster by 21％ compared with existing algorithms.     

基于OPNET的集群负载均衡仿真

为了分析比较集群负载均衡中随机调度算法、轮转算法、加权轮转算法等3种基本算法,介绍了OPNET的建模方法,给出了基于OPNET进行网络仿真的建模层次和步骤,并运用OPNET建立集群系统模型,对其进行仿真测试．仿真结果表明：3种算法均能实现负载的平衡分配,在同构集群中,轮转算法有较好的负载均衡效果;但在异构的集群系统中,加权轮转算法具有最好负载均衡效果．     

一种负载均衡的无线传感器网络自适应分簇算法

在分簇算法中,有效的簇首选取策略可以提高网络负载均衡和簇首均匀分布程度.针对簇首选取问题,文章提出了一种负载均衡的无线传感器网络自适应分簇算法,该算法使用簇半径、节点剩余能量和簇首间距作为参数选取簇首,网络中簇内成员到簇首的通信以及簇首之间的通信都基于自由空间模型的低能量衰减,簇首与Sink节点采用多跳的方式进行通信.仿真结果表明,与LEACH算法比较,该算法有效地实现了网络负载均衡和簇首均匀分布,延长了网络生存时间.     

多重请求调度的分布式Web缓存集群设计

设计了一种采用多重用户请求调度机制的、基于内容交换的WebCache集群.集群前端使用基于第4层的负载平衡策略将用户请求分发给后端;后端采用第7层交换技术,按内容转发请求到最佳匹配后端,并直接应答用户.此系统综合了当前第4层和第7层调度技术的优点,避免了前端瓶颈问题,提高了整个集群的转发效率和缓存的命中率.