ID	原文	译文
14575	首先介绍了小蜂窝网络的基本概念,论述了其主要的技术特点和独特的应用优势,分析了其在未来无线通信系统中与其他各种技术可能的融合场景并讨论了对应的特点和存在的难题。	This paper introduces the basic concepts of small cell networks，discusses its main technicalcharacteristics and unique application advantages，analyzes the possible integration scenarios with other various technologies in future wireless communication systems，and discusses corresponding features and existing technologies challenge．
14576	然后概述了小蜂窝网络技术发展过程中所面临的主要挑战,分析了目前主要的研究成果和解决方案。	Also，it outlines the possible technical challenges which will meet in the practical communication scenes，and analyzes the current main research results and solutions．
14577	最后,结合下一代无线通信的关键技术,如大规模多输入多输出通信、毫米波通信、D2D(Device-to-Device)通信和认知无线电技术等,总结了小蜂窝网络的应用现状并分析了其市场和发展前景。	Finally，based onthe major wireless communication technologies，such as large－scale multiple－input multiple－output( MIMO) communication，millimeter－wave communication，device－to－device( D2D) communication，and cognitive radio technology，the application scenarios and the market and development prospects of small cellnetworks are summarized．
14578	社交物联网是社交网络概念在物联网中整合后兴起的一个蓬勃发展的研究领域。	The social Internet of Things (SIoT) is a booming research field after the integration of the concept of social network in the Internet of Things(IoT).
14579	提出了一种适用于社交物联网网络的改进型节点级信任模型,并通过与其他信任模型的对比仿真实验证明在恶意节点的攻击下,提出的模型拥有更好的稳定性和适用性,总体波动较小。	This paper proposes an improved node -level trust model suitable for the SIoT network and proves that the model has better stability and applicability under the malicious attack of malicious nodesthrough the simulation experiment compared with other trust models. The overall fluctuation of the results ofthe proposed model is significantly smaller.
14580	同时,针对实际社交物联网网络中新加入网络的陌生节点可能遇到的网络延迟影响信任值评估的问题,在改进型节点级信任模型的基础上进一步使用了深度学习模型对其进行信任值预测。	In view of the problem that network delay may affect the trustvalue assessment of new unfamiliar nodes in the actual SIoT,this paper uses the deep learning model topredict the trust value on the basis of the improved node-level trust model.
14581	仿真证明,使用深度学习预测后模型的系统性能明显优于不使用深度学习的模型,成功交互率提升约1.8%。	The simulation results show thatthe system performance of the model using deep learning is significantly better than that of the model without deep learning,and the successful interaction rate is increased by about 1. 8% .
14582	针对图像分割过程中三维Otsu算法运算时间长、计算量大的问题,提出了一种基于Levy-人工蜂群算法的三维Otsu阈值分割算法。	In view of the problem of long operation time and large amount of calculation of the three-dimensional (3D) Otsu algorithm in the image segmentation process,a 3D Otsu threshold segmentation algorithmbased on Levy-artificial bee colony algorithm is proposed.
14583	首先,以像素灰度值-邻域均值-邻域中值的三维类间方差作为人工蜂群算法的适应度函数。	First, the 3D inter-class variance of pixel grayvalue-neighborhood mean-neighborhood median is used as the fitness function of the artificial bee colonyalgorithm.
14584	其次,采用Levy飞行模式评价像素的适应度,对其种群更新及邻域搜索过程进行优化,以增强其全局搜索能力。	Second, the Levy flight model is used to evaluate the fitness of the pixel,and its population andneighborhood are updated. The search process is optimized to enhance its global search capabilities.