一种改进的Sobel边缘检测算法的设计及其FPGA实现

为了快速得到边缘更清晰、敏感度更高的实时边缘图像数据,提出了一种改进的Sobel边缘检测算法.对原始图像的边缘进行平滑滤波后,利用2个3×3模板对选定的二维图像中同样大小区域内的数据进行卷积,合并水平及垂直方向的导数得到该区域的梯度值,再通过添加的梯度方向模块对该区域进行4个方向的判断以帮助确定边缘,最后通过非极大值抑制及二值化处理确定图像边缘.将改进后的Sobel边缘检测算法在FPGA上进行了实现并证明了其有效性.     

Mobility management architectures based on joint mobile IP and SIP protocols

In the all-IP wireless networks beyond the third generation, mobility management can be effectively achieved by applying mobile IP (MIP) and the session initiation protocol (SIP) jointly. Nevertheless, an efficient combination of both protocols remains an open research issue. Conventional hybrid MIP-SIP mobility architectures operate MIP and SIP almost independently, resulting in significant redundant costs. This article investigates the representative hybrid MIP-SIP architectures and explores the joint optimizations between MIP and SIP for a more cost-efficient mobility support whilst utilizing their complementary power. Two novel design approaches are presented. The first approach culminates in a tightly integrated architecture, which merges the redundant mobility entities in MIP and SIP to yield maximum system efficiency. The other approach leads to a loosely integrated architecture, where necessary interactions are introduced between MIP and SIP mobility servers while their physical entities are kept intact. Major mobility procedures, including location update, session setup and handoff, are discussed in these architectures. The analytical results demonstrate that both proposed architectures outperform typical hybrid MIP-SIP architectures in terms of clear-cut reduced signaling costs     

电子天平示值误差测量结果的不确定度评定

一切测量结果都不可避免地具有不确定度，结合实际应用的情况，详细介绍了电子天平示值误差测量结果的不确定度评定方法。     

集成CMOS四象限模拟乘法器

给出了一种CMOS型四象限模拟乘法器,该乘法器采用有源衰减器结合吉尔伯特单元结构.利用基于CSMC的0.6μm n阱2p2m工艺SPICE BSIM3V3 MOS模型(level=49)进行仿真,采用单电源5V电压供电.利用HSPICE仿真并给出了仿真的结果及版图实现.     

MEMS器件的腐蚀与释放

以硅微机械FP腔器件为代表,该器件采用了标准的硅表面加工工艺,分析了此类具有悬空结构的MEMS器件在进行牺牲层的腐蚀和最终的结构释放过程中的各种问题.根据所遇到问题的不同情况对器件的设计和工艺流程进行了改进,并通过实验验证了其可行性.     

Leveraging Bioinspired Structural Constraints for Tunable and Programmable Snapping Dynamics in High-Speed Soft Actuators

Creating high-speed soft actuators will have broad engineering and technological applications. Snapping provides a power-amplified mechanism to achieve rapid movements in soft actuators that typically show slow movements. However, precise control of snapping dynamics (e.g., speed and direction of launching or jumping) remains a daunting challenge. Here, a bioinspired design principle is presented that harnesses a reconfigurable constraint structure integrated into a photoactive liquid crystal elastomer actuator to enable tunable and programmable control over its snapping dynamics. By reconfiguring constrained fin-array-shaped structure, the snapping dynamics of the structured actuator, such as launching or jumping angle and height, motion speed, and release force can be on-demand tuned, thus enabling controllable catapult motion and programmable jumping. Moreover, the structured actuators exhibit a unique combination of ultrafast moving speed (up to 2.5 m s⁻¹ in launching and 0.22 m s⁻¹ in jumping), powerful ejection (long ejection distance of ≈20 cm, 35 mg ball), and high jumping height (≈8 cm, 40 times body lengths), which few other soft actuators can achieve. This study provides a new universal design paradigm for realizing controllable rapid movements and high-power motions in soft matter, which are useful for building high-performance soft robotics and actuation devices.     

云雾混合网络下基于多智能体架构的资源分配及卸载决策研究

针对D2D辅助的云雾混合架构下资源分配及任务卸载决策优化问题,该文提出一种基于多智能体架构深度强化学习的资源分配及卸载决策算法.首先,该算法考虑激励约束、能量约束以及网络资源约束,联合优化无线资源分配、计算资源分配以及卸载决策,建立了最大化系统总用户体验质量(QoE)的随机优化模型,并进一步将其转化为MDP问题.其次,...     

基于分数阶傅里叶变换的低信噪比线性调频信号参数快速估计算法

针对低信噪比线性调频信号参数估计精度低且运算量大的问题,该文提出一种基于高效分数阶傅里叶变换(FRFT)和分数阶频谱4阶原点矩的快速估计算法。该算法通过判断调频斜率的正负,以确定旋转阶次所在初始区间;进而应用高效FRFT获得初始旋转阶次;最终利用分数阶频谱4阶原点矩,进一步确定搜索区间和步长,实现精准搜索,从而满足参数精度的要求。实验结果表明,该算法尤其适合用于低信噪比情况下的线性调频(LFM)信号检测与参数的准确估计,而且运算量较低。     

基于球面Haar小波和卷积神经网络的飞行员虹膜识别

虹膜识别面临两个重要的问题：一是如何精细分解与重构虹膜球面图像;二是如何识别虹膜图特征。虹膜表面几何位置信息是一种重要的信号,传统的虹膜识别通常使用虹膜图像的平面特征,然而人的眼睛是一种球体,从平面图像难以提取到虹膜球体的几何特征。针对平面特征容易出现虹膜纹理的扭曲和失真等问题,该文建议一种正交对称的球面Haar小波(...     

Electrode Dependence of Tunneling Electroresistance and Switching Stability in Organic Ferroelectric P(VDF‐TrFE)‐Based Tunnel Junctions

Ferroelectric tunnel junctions (FTJs) are promising candidates for nonvolatile memories and memristor‐based computing circuits. Thus far, most research has focused on FTJs with a perovskite oxide ferroelectric tunnel barrier. As the need for high‐temperature epitaxial film growth challenges the technological application of such inorganic junctions, more easily processable organic ferroelectrics can serve as alternative if large tunneling electroresistance (TER) and good switching durability would persist. This study reports on the performance of FTJs with a spin‐coated ferroelectric P(VDF‐TrFE) copolymer tunnel barrier. The use of three different bottom electrodes, indium tin oxide (ITO), La_0.67Sr_0.33MnO₃, (LSMO), and Nb‐doped SrTiO₃ (STO) are compared and it is shown that the polarity and magnitude of the TER effect depend on their conductivity. The largest TER of up to 10⁷% at room temperature is measured on FTJs with a semiconducting Nb‐doped STO electrode. This large switching effect is attributed to the formation of an extra barrier over the space charge region in the substrate. The organic FTJs exhibit good resistance retention and switching endurance up to 380 K, which is just below the ferroelectric Curie temperature of the P(VDF‐TrFE) barrier.