Terahertz wave has attracted significant attention in recent years, and terahertz devices have been applied in various fields. However, the complicated and time-consuming spectrum prediction and structure design issues have hindered the widespread application of terahertz science. In this work, we propose a new method to use neural networks to predict the reflection spectrum in the terahertz band, and more importantly, design a micro-nano structure with an on-demand optical response. To verify the effectiveness, we select a terahertz metasurface as an example for discussion. After the neural networks are trained, the spectrum prediction can achieve high precision, and the neural network also has encouraging performance when solving the design problem of micro-nano structure. Furthermore, we conclude that we can choose structure design neural networks with different complexity to satisfy different demands, and can optimize the networks to improve accuracy. Our work demonstrates that such a data-driven neural network can be applied to study the prediction and design problem of metasurface in the terahertz band and provide more opportunities for the terahertz devices in the future.

In recent years, 3D vision has gained increasing prominence in practical applications such as autonomous driving and robotics. However, the scarcity of large labeled point cloud datasets continues to be a bottleneck for deep networks. Self-supervised representation learning (SRL) has emerged as an effective approach to alleviate this issue by pre-training general feature encoders without requiring human annotations. Existing contrastive SRL methods for 3D point clouds have predominantly concentrated on object representations from a global or point perspective. They overlook essential local geometry information, thereby constraining the generalizability of pre-trained models. To address these challenges, we propose a local contrast module as an intermediate level between the scene and point levels. It is then integrated with a global contrast module to form a dual contrast method known as DCPoint. The local contrast module operates on point-wise representations of objects and designs contrastive pairs based on the spatial information of point clouds. It effectively addresses the challenges posed by the sparsity and irregularity of point clouds and imperfect partition issues. The point-wise local contrast module aims to enhance the internal connections between the components within the point cloud, while the global contrast module introduces semantic information about individual instances. Experimental results demonstrate the effectiveness of DCPoint across various downstream tasks on synthetic and real-world datasets. It consistently outperforms previously reported SRL methods and the randomly initialized counterparts. Additionally, the proposed local contrast module can enhance the performances of other SRL methods. Our source codes of this research are available at https://github.com/UnderTheMangoTree/DCPoint.git.

The fog radio access network (F-RAN) equipped with enhanced remote radio heads (eRRHs), which can pre-store some requested files in the edge cache and support mobile edge computing (MEC). To guarantee the quality-of-service (QoS) and energy efficiency of F-RAN, a proper content caching strategy is necessary to avoid coarse content storing locally in the cache or frequent fetching from a centralized baseband signal processing unit (BBU) pool via backhauls. In this paper we investigate the relationships among eRRH/terminal activities and content requesting in F-RANs, and propose an edge content caching strategy for eRRHs by mining out mobile network behavior information. Especially, to attain the inference for appropriate content caching, we establish a pre-mapping containing content preference information and geographical influence by an efficient non-uniformed accelerated matrix completion algorithm. The energy consumption analysis is given in order to discuss the energy saving properties of the proposed edge content caching strategy. Simulation results demonstrate our theoretical analysis on the inference validity of the pre-mapping construction method in static and dynamic cases, and show the energy efficiency achieved by the proposed edge content pre-caching strategy.

Mobile online social network (mOSN) is a burgeoning research area. However, most existing works referring to mOSNs deal with static network structures and simply encode whether relationships among entities exist or not. In contrast, relationships in signed mOSNs can be positive or negative and may be changed with time and locations. Applying certain global characteristics of social balance, in this paper, we aim to infer the unknown relationships in dynamic signed mOSNs and formulate this sign inference problem as a low-rank matrix estimation problem. Specifically, motivated by the Singular Value Thresholding (SVT) algorithm, a compact dictionary is selected from the observed dataset. Based on this compact dictionary, the relationships in the dynamic signed mOSNs are estimated via solving the formulated problem. Furthermore, the estimation accuracy is improved by employing a dictionary self-updating mechanism.

With centralized baseband unit (BBU) pool, Cloud-RAN (C-RAN) has powerful computation capability to support jointly processing signals for several remote radio heads (RRHs) simultaneously. It raises a challenge to fully utilize the computation resources and meet the hard deadline constraint posed by multi-media traffics. To overcome this challenge, in this work a real-time partitioned scheduling algorithm called Wi-EDF has been proposed. In order to improve the efficiency of Wi-EDF, a cross layer scheduling framework is firstly presented. Based on that, Wi-EDF is carrying out in a two level scheduling manner. In the first level, a beamforming based scheduling problem is formulated, which aims to minimize the system power consumption with the constraint of fronthual capacity. The problem can be iterative solved by the equivalent second order cone program (SOCP) problems. In the second level, a preemptive earliest deadline first (EDF) based scheduling algorithm with admission control has been designed. The processing time deduced by the baseband signaling has been taken into account in this level. The simulation results show that Wi-EDF has lower power consumption, higher CPU utilization and higher overall throughput comparing with LTE-A.