首页> 外文期刊>Applied Mathematical Modelling >An Algorithm for Matrix Recovery of High-loss-rate Network Traffic Data
【24h】

An Algorithm for Matrix Recovery of High-loss-rate Network Traffic Data

机译:一种矩阵恢复高损耗率网络流量数据的算法

获取原文
获取原文并翻译 | 示例

摘要

In network engineering, it is significant to recover the complete traffic flows from extremely limited sampled data. To carry out this task, we propose a linear-constrained nuclear norm minimization model, which can exploit the underlying properties of the traffic matrix, namely the spatio-temporal low-rankness, the time-continuity and the nonnegativ-ity. To efficiently solve the proposed model, we subsequently design a Schur complement based semi-proximal alternating direction method of multipliers algorithm. The proposed algorithm is proved to be globally convergent. Moreover, at each iteration step, the optimums of all the block variables have closed forms. This advantage makes the iterations fast and accurate. To weaken the parameter sensitivity, we also adopt cross validation, an effective technique in machine learning, to find suitable parameters in advance. By absorbing these techniques, our method substantially improves the recovery accuracy in comparison with others in the vast majority of data loss scenarios. Besides, the required computational time of our algorithm is also far less than the actual time duration.
机译:在网络工程中,从极其有限的采样数据恢复完整的流量很大。为了执行这项任务,我们提出了一种线性约束的核规范最小化模型,它可以利用交通矩阵的潜在特性,即时空低秩,时间连续性和非空白问题。为了有效地解决所提出的模型,我们随后设计了一种基于SCUR补充的乘法算法的半近端交替方向方法。已证明该算法被证明是全局收敛。此外,在每个迭代步骤中,所有块变量的最佳度都具有封闭形式。这种优势使迭代快速准确。为了削弱参数灵敏度,我们还采用交叉验证,机器学习中的有效技术,提前找到合适的参数。通过吸收这些技术,我们的方法与绝大多数数据丢失方案中的其他方法相比,我们的方法大大提高了恢复精度。此外,我们算法的所需计算时间也远远低于实际持续时间。

著录项

  • 来源
    《Applied Mathematical Modelling》 |2021年第8期|645-656|共12页
  • 作者单位

    Department of Mathematical Sciences Tsinghua University Beijing 100084 China Theory Lab Central Research Institute 2012 Labs Huawei Technologies Co. Ltd Hong Kong;

    Department of Mathematical Sciences Tsinghua University Beijing 100084 China;

    Theory Lab Central Research Institute 2012 Labs Huawei Technologies Co. Ltd Hong Kong;

    Theory Lab Central Research Institute 2012 Labs Huawei Technologies Co. Ltd Hong Kong;

  • 收录信息 美国《科学引文索引》(SCI);美国《工程索引》(EI);
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

    Network traffic; Nuclear norm; Alternating direction method of multipliers; Cross validation;

    机译:网络流量;核标准;交替方向乘法器方法;交叉验证;

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号