Relay in full-duplex(FD) mode can achieve higher spectrum efficiency than that in half-duplex mode,while it is crucial to suppress relay self-interference to ensure transmission quality which requires instantaneous ch...Relay in full-duplex(FD) mode can achieve higher spectrum efficiency than that in half-duplex mode,while it is crucial to suppress relay self-interference to ensure transmission quality which requires instantaneous channel state information(CSI). In this paper,the channel estimation issue in FD amplify-andforward relay networks is considered,where the training-based estimation technique is adopted. Firstly,the least square(LS) estimation is implemented to obtain composite channel coefficients of source-relay-destination(SRD) channel and relay loop-interference(LI) channel in order to assist destination in performing data detection. Secondly,both LS and maximum likelihood estimation methods are utilized to perform individual channel estimation aiming at supporting successive interference cancelation at destination. Finally,simulation results demonstrate the effectiveness of both composite and individual channel estimation,and the presented ML method can achieve lower MSEs than LS solution.展开更多
真实场景点云不仅具有点云的空间几何信息,还具有三维物体的颜色信息,现有的网络无法有效利用真实场景的局部特征以及空间几何特征信息,因此提出了一种双通道特征融合的真实场景点云语义分割方法DCFNet(dual-channel feature fusion of ...真实场景点云不仅具有点云的空间几何信息,还具有三维物体的颜色信息,现有的网络无法有效利用真实场景的局部特征以及空间几何特征信息,因此提出了一种双通道特征融合的真实场景点云语义分割方法DCFNet(dual-channel feature fusion of real scene for point cloud semantic segmentation)可用于不同场景下的室内外场景语义分割。更具体地说,为了解决不能充分提取真实场景点云颜色信息的问题,该方法采用上下两个输入通道,通道均采用相同的特征提取网络结构,其中上通道的输入是完整RGB颜色和点云坐标信息,该通道主要关注于复杂物体对象场景特征,下通道仅输入点云坐标信息,该通道主要关注于点云的空间几何特征;在每个通道中为了更好地提取局部与全局信息,改善网络性能,引入了层间融合模块和Transformer通道特征扩充模块;同时,针对现有的三维点云语义分割方法缺乏关注局部特征与全局特征的联系,导致对复杂场景的分割效果不佳的问题,对上下两个通道所提取的特征通过DCFFS(dual-channel feature fusion segmentation)模块进行融合,并对真实场景进行语义分割。对室内复杂场景和大规模室内外场景点云分割基准进行了实验,实验结果表明,提出的DCFNet分割方法在S3DIS Area5室内场景数据集以及STPLS3D室外场景数据集上,平均交并比(MIOU)分别达到71.18%和48.87%,平均准确率(MACC)和整体准确率(OACC)分别达到77.01%与86.91%,实现了真实场景的高精度点云语义分割。展开更多
为实时传输海量船舶图像,为有效监控船舶实时航行状况提供保障,研究基于云计算的船舶图像网络数据低时延传输方法。结合云计算技术,构建船舶图像网络数据传输平台,通过平台数据接收单元将大量船舶网络数据存入云端服务器内;数据处理单...为实时传输海量船舶图像,为有效监控船舶实时航行状况提供保障,研究基于云计算的船舶图像网络数据低时延传输方法。结合云计算技术,构建船舶图像网络数据传输平台,通过平台数据接收单元将大量船舶网络数据存入云端服务器内;数据处理单元基于云计算的分布式处理特点,结合并行运行框架Map Reduce和并行化K-means聚类算法,并行挖掘云端服务器内的船舶图像网络数据;信道均衡单元结合所构建的船舶通信网络节点能量消耗模型,均衡船舶通信网络信道,运用均衡信道实现所挖掘船舶图像网络数据的传输。结果显示,该方法的数据挖掘速率高,数据传输过程中各信道能量开销均衡,且可低至3.5 k J,传输船舶图像网络数据的时延可低至10.23 ms,实现了数据的低时延传输。展开更多
基金supported in part by the National High Technology Research and Development Program of China(Grant No.2014AA01A707)the Beijing Natural Science Foundation(Grant No.4131003)+1 种基金the Specialized Research Fund for the Doctoral Program of Higher Education (SRFDP)(Grant No.20120005140002)the Key Program of Science and Technology Development Project of Beijing Municipal Education Commission of China (KZ201511232036)
文摘Relay in full-duplex(FD) mode can achieve higher spectrum efficiency than that in half-duplex mode,while it is crucial to suppress relay self-interference to ensure transmission quality which requires instantaneous channel state information(CSI). In this paper,the channel estimation issue in FD amplify-andforward relay networks is considered,where the training-based estimation technique is adopted. Firstly,the least square(LS) estimation is implemented to obtain composite channel coefficients of source-relay-destination(SRD) channel and relay loop-interference(LI) channel in order to assist destination in performing data detection. Secondly,both LS and maximum likelihood estimation methods are utilized to perform individual channel estimation aiming at supporting successive interference cancelation at destination. Finally,simulation results demonstrate the effectiveness of both composite and individual channel estimation,and the presented ML method can achieve lower MSEs than LS solution.
文摘真实场景点云不仅具有点云的空间几何信息,还具有三维物体的颜色信息,现有的网络无法有效利用真实场景的局部特征以及空间几何特征信息,因此提出了一种双通道特征融合的真实场景点云语义分割方法DCFNet(dual-channel feature fusion of real scene for point cloud semantic segmentation)可用于不同场景下的室内外场景语义分割。更具体地说,为了解决不能充分提取真实场景点云颜色信息的问题,该方法采用上下两个输入通道,通道均采用相同的特征提取网络结构,其中上通道的输入是完整RGB颜色和点云坐标信息,该通道主要关注于复杂物体对象场景特征,下通道仅输入点云坐标信息,该通道主要关注于点云的空间几何特征;在每个通道中为了更好地提取局部与全局信息,改善网络性能,引入了层间融合模块和Transformer通道特征扩充模块;同时,针对现有的三维点云语义分割方法缺乏关注局部特征与全局特征的联系,导致对复杂场景的分割效果不佳的问题,对上下两个通道所提取的特征通过DCFFS(dual-channel feature fusion segmentation)模块进行融合,并对真实场景进行语义分割。对室内复杂场景和大规模室内外场景点云分割基准进行了实验,实验结果表明,提出的DCFNet分割方法在S3DIS Area5室内场景数据集以及STPLS3D室外场景数据集上,平均交并比(MIOU)分别达到71.18%和48.87%,平均准确率(MACC)和整体准确率(OACC)分别达到77.01%与86.91%,实现了真实场景的高精度点云语义分割。
文摘为实时传输海量船舶图像,为有效监控船舶实时航行状况提供保障,研究基于云计算的船舶图像网络数据低时延传输方法。结合云计算技术,构建船舶图像网络数据传输平台,通过平台数据接收单元将大量船舶网络数据存入云端服务器内;数据处理单元基于云计算的分布式处理特点,结合并行运行框架Map Reduce和并行化K-means聚类算法,并行挖掘云端服务器内的船舶图像网络数据;信道均衡单元结合所构建的船舶通信网络节点能量消耗模型,均衡船舶通信网络信道,运用均衡信道实现所挖掘船舶图像网络数据的传输。结果显示,该方法的数据挖掘速率高,数据传输过程中各信道能量开销均衡,且可低至3.5 k J,传输船舶图像网络数据的时延可低至10.23 ms,实现了数据的低时延传输。