The traditional Convolutional Neural Network (CNN) cannot directly process point cloud data, and the point cloud data must be converted into a multi-view or voxelized grid, which leads to a complicated process and low point cloud recognition accuracy. Aiming at the problem, a new point cloud classification and segmentation network called Linked-Spider CNN was proposed. Firstly, the deep features of point cloud were extracted by adding more Spider convolution layers based on Spider CNN. Secondly, by introducing the idea of residual network, short links were added to every Spider convolution layer to form residual blocks. Thirdly, the output features of each layer of residual blocks were spliced and fused to form the point cloud features. Finally, the point cloud features were classified by three-layer fully connected layers or segmented by multiple convolution layers. The proposed network was compared with other networks such as PointNet, PointNet++ and Spider CNN on ModelNet40 and ShapeNet Parts datasets. The experimental results show that the proposed network can improve the classification accuracy and segmentation effect of point clouds, and it has faster convergence speed and stronger robustness.

Since relation and its fragments in the distributed database have multiple copies and multi-site storage characteristics, which increases the time and space complexity of query and results in lower search efficiency of Query Execution Plan (QEP), a Parallel Genetic Algorithm and Max-Min Ant System (PGA-MMAS) based on design principles of Fragments Site Selector (FSS) was proposed. Firstly, based on the design requirement of the distributed information management system for actual business, FSS was designed, which selected the best one from many copies of relationship heuristically to decrease query join cost and search space of PGA-MMAS. Secondly, Genetic Algorithm (GA) encoded the final join relations and conducted parallel genetic operation to get a set of relative optimal QEPs by taking advantage of quick convergence of GA. Then, QEPs were transformed into the initial pheromone distribution of Max-Min Ant System (MMAS) to obtain the optimal QEP quickly and efficiently. Finally, simulation experiments were conducted under different number of relation conditions, and the results show that PGA-MMAS based on FSS searches optimal QEP more efficiently than original GA, Fragments Site Selector-Genetic Algorithm (FSS-GA), Fragments Site Selector-Max-Min Ant System (FSS-MMAS) and Fragments Site Selector-Genetic Algorithm-Max-Min Ant System (FSS-GA-MMAS). And in the actual engineering application, the proposed algorithm can search high-quality QEP to improve the efficiency of multi-join query in distributed database.

In cellular networks, users requesting the same data from base station can be grouped into a cluster, and Device-to-Device (D2D) links can be utilized to improve data dissemination efficiency. However, the difference of D2D links may be the bottleneck that results in conservative resource utilization. To solve this problem, a priority-based D2D relay scheme was proposed. The proposed scheme first set an optimal threshold by traversing the D2D links quality matrix, and then preferentially selected the D2D links with high achievable data rates to forward the data. In simulation experiment, compard with the traditional scheme without priority, the proposed scheme consumed less time and frequency resources with a higher efficiency especially when the ratio of ACK users and NACK users decreasing; meanwhile, the size of the cluster also influenced the improvement of resource efficiency. The results show that the proposed priority-based scheme improves resource efficiency when there is a small number of users, especially the users of NACK are more than that of ACK.