ID	原文	译文
3653	针对 6G 时代将会是移动通信与人工智能紧密结合的时代，产生数量庞大的边缘智能信号处理节点的趋势，提出了一种可部署于资源受限的边缘设备上的高效智能电磁信号识别模型。	In response to the trend that in the 6th generation wireless (6G) era, mobile communications and artificial in-telligence will be closely integrated, and a huge number of edge intelligent signal processing nodes will be deployed, anefficient and intelligent electromagnetic signal recognition model was proposed, which could be deployed on re-source-constrained edge devices.
3654	首先，通过绘制电磁信号的星座图将电磁信号具象为二维图像，并根据归一化点密度对星座图上色以实现特征增强；	The constellation diagram of electromagnetic signal was firstly drawn to visualize elec-tromagnetic signal as a two-dimensional image, and color the constellation diagram according to the normalized pointdensity to achieve feature enhancement.
3655	然后，使用二值化深度神经网络对其进行识别，在保证识别准确率的同时明显降低了模型存储开销以及计算开销。	Then, a binary deep neural network was adopted to recognize the colored con-stellation diagrams. It was shown that the approach can guarantee a high recognition accuracy, which significantly re-duced the model storage and calculation costs.
3656	采用电磁信号调制识别问题进行验证，实验选取常用的 8 种数字调制信号，选择加性高斯白噪声为信道环境。	For verification, the proposed approach was applied to the problem ofelectromagnetic signal modulation recognition. The experiment uses eight commonly used digital modulation signals andselects additive white Gaussian noise as the channel environment.
3657	实验结果表明，所提方案可以在信噪比为−6～6 dB 的噪声条件下获得 96.1%的综合识别率，网络模型大小仅为 166 KB，部署于树莓派 4B 的执行时间为 290 ms，相比于同规模的全精度网络，准确率提升了 0.6%，模型缩减到126.16，运行时间缩减到12.37。	The experimental results show that the scheme canachieve a comprehensive recognition rate of 96.1% under the noise condition of −6～6 dB, while the size of the networkmodel is only 166 KB. Further, the execution time, when executed on a Raspberry Pi 4B, is only 290 ms. Compared to afull-precision network of the same scale, the accuracy is increased by 0.6%, the model is reduced to126.16, and the run-ning time is reduced to12.37.
3658	针对无人机高移动性、飞行高度可控及视距信道概率高等使无人机无线网络的设计与部署存在多方面的挑战，	For the characteristics of unmanned aerial vehicle (UAV), such as high mobility, adjustable height, and highprobability of line-of-sight channels, which introduced multiple challenges to the design and deployment of current andfuture wireless network,
3659	且其组网性能是业界的焦点难点问题，分析了异构无人机无线网络的网络覆盖与切换性能，为无人机无线组网提供理论支撑。	and the networking performance of UAV was the focus and difficult issues in the industry, thenetwork coverage and handoff performance of heterogeneous UAV wireless network was analyzed, which provided in-sights into UAV wireless networking.
3660	具体地，考虑无人机移动性导致的信道状态信息过时问题，基于随机几何理论，推导了异构无人机无线网络的切换概率、切换误差概率和覆盖概率的解析表达式，探究了无人机移动性、飞行高度和地面基站密度对上述性能的影响。	Specifically, considering the outdated channel state information introduced by theUAV's mobility, analytical expressions of handover probability, handover error probability, and coverage probability forheterogeneous UAV wireless network were derived by utilizing the tools from stochastic geometry. In addition, the im-pacts of the UAV's mobility, height, as well as the density of terrestrial base stations on the aforementioned performancemetrics were investigated.
3661	研究表明，过时信道状态信息会造成切换误差，且随无人机移动速度和基站密度提升呈先增大后减小趋势。	It shows that the outdated channel state information caused handover error, which increasesfirst and then decreases with the increase of UAV's moving speed and density of base stations.
3662	同时，无人机飞行高度比移动速度对覆盖概率的影响更加显著。	Meanwhile, the impact ofUAV's flight height on coverage probability is more significant than that of moving speed.