Time filter

Source Type

Luo X.,Xian Jiaotong University | Luo X.,Shaanxi University of Science and Technology | Xin G.,AVIC Computing Technique Research Institute | Wang Y.,Nanjing Southeast University | And 2 more authors.
Cluster Computing | Year: 2015

Load balance and power proportionality are both important aspects in constructing high-performance and cost-effective distributed storage systems. However, traditional replica placement strategies which provide load balance usually produce scattered replica layouts, thus disabling power proportionality, while recent strategies that enable power proportionality are typically based on uniform replication, thus compromising the ability of load balance. In this article, we introduce Superset (an organized non-uniform replica placement strategy) which takes both load balance and power proportionality into consideration. The main idea is to partition the whole system into several uniform replication based subsystems with the accommodated file subsets satisfying the ‘superset’ condition. We proved Superset to be theoretically effective, and conducted a series of simulations with real-world distributions of data popularity. Our results show that, compared to state of the art solutions, Superset consumes less energy to fulfill the same performance requirement while offers higher performance subject to the same energy consumption constraint. © 2015 Springer Science+Business Media New York Source

Zhang J.-J.,Xidian University | Fang M.,Xidian University | Wang H.,Xidian University | Wang H.,AVIC Computing Technique Research Institute | Li X.,Xidian University
Engineering Applications of Artificial Intelligence | Year: 2015

High dimensionality of label space poses crucial challenge to efficient multi-label classification. Therefore, it is needed to reduce the dimensionality of label space. In this paper, we propose a new algorithm, called dependence maximization based label space reduction (DMLR), which maximizes the dependence between feature vectors and code vectors via Hilbert-Schmidt independence criterion while minimizing the encoding loss of labels. Two different kinds of instance kernel are discussed. The global kernel for DMLRG and the local kernel for DMLRL take global information and locality information into consideration respectively. Experimental results over six categorization problems validate the superiority of the proposed algorithm to state-of-art label space dimension reduction methods in improving performance at the cost of a very short time. © 2015 Elsevier Ltd. All rights reserved. Source

Zhao J.,AVIC Computing Technique Research Institute | An J.,Xian University of Science and Technology
International Journal of Pattern Recognition and Artificial Intelligence | Year: 2012

The contours and segments of objects in digital images have many important applications. Contour extractions of gray images can be converted into contour extractions of binary images. This paper presents a novel contour-extraction algorithm for binary images and provides a deduction theory for this algorithm. First, we discuss the method used to construct convex hulls of regions of objects. The contour of an object evolves from a convex polygon until the exact boundary is obtained. Second, the projection methods from lines to objects are studied, in which, a polygon iteration method is presented using linear projection. The result of the iteration is the contour of the object region. Lastly, addressing the problem that direct projections probably cannot find correct projection points, an effective discrete ray-projection method is presented. Comparisons with other contour deformation algorithms show that the algorithm in the present paper is very robust with respect to the shapes of the object regions. Numerical tests show that time consumption is primarily concentrated on convex hull computation, and the implementation efficiency of the program can satisfy the requirement of interactive operations. © World Scientific Publishing Company. Source

Wang H.,Xidian University | Wang X.,AVIC Computing Technique Research Institute | Liu S.F.,Xidian University | Li P.,Xidian University | Shi X.W.,Xidian University
Microwave and Optical Technology Letters | Year: 2015

A broadband circularly polarized (CP) antenna with simple structure and wide angular coverage is proposed. The interlaced coplanar dipole is excited by a coaxial cable to achieve the CP wave. To obtain a high gain and low backward radiation, a rectangular cavity-backed reflector is used. An initial design at around 6.0 GHz is presented to show the performance of this configuration. An impedance bandwidth and axial ratio bandwidth of about 35.6 and 33.5% can be achieved at this frequency. Also, a stable gain of about 10 ± 0.5 dBi and a wide axial ratio beamwidth of about 200° over the frequency of interest are obtained. Then, a parametric analysis is conducted to estimate the significance of the design parameters that affect the antenna's performance. Finally, a prototype at 11.0 GHz is fabricated to validate the design. The measured results are in good agreement with the simulated ones, which shows that the design can be easily applied to other frequency and that the design may be a good candidate for the CP antenna family. Source

Zhou M.,Huazhong University of Science and Technology | Li R.,Huazhong University of Science and Technology | Shang L.,Beihang University | Zhang L.,AVIC Computing Technique Research Institute
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013

This paper proposes the design of a fault-tolerant controller area network (CAN). Several candidate redundancy strategies are introduced to improve the reliability. To choose a suitable one wisely, it is meaningful to make comparison on the reliability between the candidates. However, as there are redundant channels, the reliability of the network depends on the transmission requirements, which is difficult to evaluate. In this paper, a general approach based on minimal path set is proposed to model the reliability of complex fault-tolerant systems. With the proposed approach, the transmission-aware reliability of the redundant CAN network is evaluated. Comparisons on the reliability are carried out between the candidate strategies. In addition, comparisons on the cost, complexity and latency are also carried out. The results show the channel redundancy is superior to other strategies due to its high reliability, low latency and acceptable cost. © Springer-Verlag 2013. Source

Discover hidden collaborations