Resource allocation for multi-class services in multipath networks

In multipath networks, multiple paths are available for each pair of source and destination and can be used to carry data packets parallelly. It has been recognized that using multipath could promote the transmission reliability and fault tolerance, and improve the performance of increasingly bandwidth-hungry multi-media applications. In this paper we propose the resource allocation model for multi-class services in multipath networks with the objective of utility maximization, which is an intrinsically difficult problem of nonconvex optimization. We firstly analyze the model for only elastic services and obtain the optimal rate allocation for them. Then we also discuss the model for inelastic services with nonconcave (sigmoidal or discontinuous) utilities which share common links with elastic ones, and obtain some sufficient conditions under which the global optimum for both elastic and inelastic services can be obtained. For the nonconvex optimization problem, we present a heuristic algorithm using Particle Swarm Optimization (PSO), which can lead to improved solutions over existing approaches. Finally, some numerical examples are given to verify the results obtained.     

基于条件信息卷积生成对抗网络的图像识别

传统的图像识别方法需要大量有标签样本进行训练，且模型训练难以达到稳定。针对这些问题，结合条件生成网络和信息最大化生成网络的结构优势建立了条件信息卷积生成网络（C-Info-DCGAN）。模型增加图像的类别信息和潜在信息作为输入数据，然后利用Q网络去更好地发挥类别信息和潜在信息对训练的引导作用，并且利用深度卷积网络来加强对图像特征的提取能力。实验结果表明，该方法能够加快模型训练收敛速度，并有效提高图像识别的准确率。     

Automated coronal hole segmentation from Solar EUV Images using the watershed transform

Region of interest segmentation in solar images is the subject of frequent research in solar physics. This study outlines watershed by immersion segmentation to identify coronal hole areas in solar images acquired using the Extreme UV Imaging Telescope (EIT). Solutions presented here produce highly accurate segmentation results of coronal holes of irregular shape, and what is more, they do so for images representing varied solar activity, recorded in different years and months. In addition, the solutions presented here make all the methods used operate very quickly. These methods include: the preprocessing step before the watershed segmentation, the watershed segmentation itself, and also the postprocessing of solar images after the watershed segmentation. The mean duration of the entire segmentation process of solar images amounts to 342 ms for a single coronal hole, without the parallel implementation of the methods used. The experiments were carried out on a computer with an Intel Core i7 CPU @ 2 GHz and 4 GB RAM. After the seed point is identified inside the coronal hole, the segmentation runs automatically.     

珠江新城规划和发展回顾

广州市珠江新城是广州未来最重要的城市中心区,其发展源于广州"东进"与"国际化"的战略,从"城市中心区"上升为"21世纪CBD",现在又再次面临亚运所带来的发展机遇。在这个过程中,广州市先后编制了控制性详细规划、城市设计、规划检讨等多种城市规划方案指引珠江新城建设。如今的珠江新城已经逐渐发展成为商务发展氛围成熟国际化、公共服务设施多样高质量和市政基础配套高效多功能的现代化CBD。珠江新城的规划体现了一种动态完善的思路,其未来的发展仍然主要取决于广州的整体国际化程度,但珠江新城必然将对此有重要的贡献。     

Coverage‐aware connectivity‐constrained unattended sensor deployment in underwater acoustic sensor networks

Careful deployment of nodes in underwater acoustic sensor networks in a distributed manner with the goal of maximized coverage and guaranteed connectivity is a challenging problem because it is very difficult and costly to access the 3D underwater environment. This paper presents a novel algorithm for self‐deployment of nodes in underwater acoustic sensor networks assuming that the nodes are randomly dropped to the water surface and form a densely populated connected network at the water surface. The idea of the algorithm is based on calculating an optimized depth for each node in the network in such a way that the possible sensing coverage overlaps are minimized and the connectivity of final topology is guaranteed. The algorithm has three main phases. In the first phase, nodes are organized in a tree structure that is rooted at the surface station. In the second phase, the depths for all nodes are computed iteratively at surface station. In the final phase, the calculated depths are distributed to nodes so that the nodes start sinking. The performance of the proposed approach is validated through simulation. We observed that the proposed approach performs at least 10% better in terms of network coverage than contemporary schemes in the literature. Copyright © 2016 John Wiley & Sons, Ltd.     

基于隐马尔科夫模型的CPM信号盲Turbo均衡算法

针对高阶连续相位调制信号(CPM)盲均衡中存在的均衡性能较差以及不容易收敛等问题,从隐马尔科夫模型(HMM)的角度出发,建立一种新的EM-SOVA信道盲均衡方法,并结合Turbo均衡思想,提出了一种适用于高阶CPM信号的盲Turbo均衡算法。该算法通过将SOVA内嵌到EM算法的迭代过程中,有效改善了信道均衡效果,同时使用Turbo均衡的软信息迭代处理来进一步提高低信噪比条件下系统性能。理论分析和仿真结果表明,所提算法具有良好的盲均衡性能以及收敛性。     

基于γ图像重建放射源三维几何形状的模拟研究

准确评估待检测对象内放射性物质的几何形状,在核设施退役和核安全等方面都具有重要的意义。利用MCNP模拟屏蔽体内不同几何形状的体源,针孔γ相机围绕被探测区域获取投影数据。根据针孔成像的机理,建立系数矩阵,并通过最大似然期望最大(MLEM)算法重建体源的三维位置和形状分布。模拟数据结果表明:该方法可准确地重建出放射性体源的三维位置和形状分布,体源在三维方向上形状的尺寸偏差不大于10%。     

SHAPE OPTIMIZATION FOR FATIGUE PROBLEMS IN MECHANICAL ENGINEERING

High Cycle Fatigue (HCF) is a very important topic in mechanical engineering problems. Almost any machine part contains some changes in cross section or boreholes connected with stress concentration. It is the task of shape optimization to improve the strength of machine parts by changing the design at free boundaries. In the past, some work was done with the so-called classical stress concentration factors just to minimize the mentioned stress concentration, also for fatigue. This paper represents a more sophisticated view by including modern Continuum Damage Mechanics (CDM) to model material behavior in HCF. The central question is to design machinery parts so that maximum lifetime is obtained. The whole context for this is developed in this paper, including non-linear finite element algorithms. The paper concludes with numerical and experimental tests for the shape optimization of a notched tension bar in HCF.     

A GENERAL APPROACH TO THE STUDY OF SPATIAL SYSTEMS II. Two Examples of Application

The emphasis in this work is on the practical application of ideas set forth in Part I. Two studies involving spatial distribution are discussed: the distribution of human population in the Indianapolis, Indiana Region, and the world distribution of mammals. In the first, populations over the period 1890–1980 of Ihe 393 townships making up the study area are used to produce classifications involving various numbers of classes of township types. A test of the theoretical structure posed in Part I is then detailed. On the theory that the underlying relations involved cannot sustain a real world structure involving more than five functional classes, it is hypothesized that classifications involving more than five classes will be associated with township distribution patterns that conflict with Central Place Theory. This is shown to be clearly so. In the second study, the derivation of a five region partitioning of world mammalian distribution patterns is described. Discussion includes the introduction of a wholly new rationale for taxonomic revision based on biogeographic criteria (and ultimately the principles described in Part I). The study closes with a short argument regarding why the model described in Part I should not be labelled as either social physics or anti-humanistic.