ID 原文 译文
20695 然后在获得初始支撑集的前提下，先循环删除支撑集内的元素，再从支撑集补集中依据与当前残差内积值最大来匹配添加新元素，直至残差内积基本不变； Then, underthe premise of having obtained initial support set, the elements in the support set are removed cyclically. Inaddition, according to the maximum value of the current residual within the product, the remaining elementsare used to match and add the new elements until the residual product is the same.
20696 最后利用时延值与稀疏支撑集的关系得到了时延的估计值。 Finally, the estimate of theTOA is obtained using the relationship between the time delay value and the sparse support set.
20697 仿真结果表明，所提算法相比于传统稀疏重构时延估计算法具有更高的估计精度。 The simulation results show that the proposed algorithm has higher estimation accuracy than the traditional sparse reconstruction time delay estimation algorithm.
20698 同时基于USRP平台，利用实际信号对所提算法进行了有效性验证。 At the same time, based on the USRP platform, theeffectiveness of the proposed algorithm is verified by the actual signal.
20699 服务功能链的服务性能取决于功能的部署位置和数据传输路径的选择。 The efficiency of Service Function Chain (SFC) depends closely on where functions are deployed and how to select paths for data transmission.
20700 针对资源有限的网络中的服务功能链部署问题，该文设计了一种基于最长有效功能序列(LEFS)的服务功能链部署算法， For the problem of SFC deployment in a resource-constrainednetwork, this paper proposes an optimization algorithm for SFC deployment based on the Longest EffectiveFunction Sequence (LEFS).
20701 以功能复用和带宽需求联合优化为目标，在控制路径长度的同时根据LEFS逐步搜索中继节点，直至满足服务请求。 To optimize function deployment and bandwidth requirement jointly, the upperbound of path length is set and relay nodes are searched incrementally on the basis of LEFS until the service request is satisfied.
20702 仿真结果表明，该算法能够均衡网络资源，同时优化网络的功能部署率和带宽利用率， Simulation results show that, the proposed algorithm can balance network resource andoptimize the function deploymen rate and bandwidth utilization.
20703 与其他算法相比，网络资源利用率降低了10%，可以支持更多的服务请求，且算法计算复杂度低，可以实现对服务请求的快速响应。 Compared with other algorithms, the utilization of network resource decreases 10%, so that more service requests can be supported. What is more,the algorithm has a lower computation complexity and can response to service requests quickly.
20704 针对动作特征在卷积神经网络模型传输时的损失问题以及网络模型过拟合的问题，该文提出一种跨层融合模型和多个模型投票的动作识别方法。 To solve the problem of the loss in the motion features during the transmission of deep convolution neural networks and the overfitting of the network model, a cross layer fusion model and a multi-model voting action recognition method are proposed.