基于K-距离拓扑的分布式数据存储方法

针对分布式数据存储算法通常需要较长的等待时间、且对海量数据加密时严重浪费计算资源和时间的问题,提出了一种基于K-距离拓扑的分布式数据存储方法.通过寻找K-距离拓扑子图来实现数据的安全放置,优先选择存取速度更快的节点和自身保护能力强的节点实现总体性能的提升.在Internet 2拓扑图与随机拓扑图下的仿真测试结果表明,所提出的方法能在满足安全距离约束的条件下选择到最优的数据存储节点,从而减小数据存取时间.     

Video multipath transmission mechanism with load balancing in the FiWi network

Aiming at the differentiated needs of users and the problem of video service interruption caused by the local overload of the converged network,a load-balanced video transmission mechanism is proposed.First,the split multipath routing protocol is improved to obtain the path selection model according to the front-end structural characteristics of the optical fiber wireless access network;second,we calculate the path differential delay and the video transmission delay on wireless side,and set their threshold as the quality of the experience constraint.Finally,while considering the quality of user experience,a particle swarm optimization algorithm with a multi-stage penalty function is used for video distribution.Simulation results show that the proposed mechanism is better than the throughput-aware load-sensing algorithms,enhances delay-control load distribution algorithms and delay-energy-quality aware multipath algorithms,and that it can effectively balance the network load and alleviate network congestion while avoiding user loss.     

Method for discrimination of false targets in multistation radar systems based on the deep neural network

For the existing jamming discrimination methods for multistation radar systems,only the single feature of target echo space correlation is utilized as the metric,which leads to insufficient comprehensiveness of feature extraction,so that effectiveness and universality are insufficient for the discrimination algorithm.In this paper,an identification method in multistatic radar systems based on the deep neural network is proposed.This method combines the characteristics of multistatic radar systems cooperative detection technology,which has many available resources and strong scheduling ability in space,time and frequency domain,with the strong model learning and feature representation ability in the process of information processing on the deep neural network,so that it can effectively apply to the field of anti-deception jamming.Full use is made of unknown information about echo data to obtain more multi-dimensional,more comprehensive,more complete and deeper feature differences besides correlation,so as to achieve a better jamming discrimination effect.Simulation results show that the proposed method can effectively reduce the influence of noise and pulse number on the jamming discrimination performance.At the same time,the limitation of the target echo correlation coefficient on anti-jamming technology under nonideal conditions is alleviated,which broadens the boundary conditions of the application process.     

T-NTRU IoT dynamic access authentication technology

The secure access authentication of the Internet of the things terminal is the key technology to ensure the large-scale construction of the power Internet of things.The elliptic curve cryptography (ECC) algorithm is usually used in the transmission authentication scheme,and it requires a large amount of calculation.Furthermore it is proved that its security against quantum attacks is poor.The number theory research unit (NTRU) algorithm can resist quantum attacks,and its computational speed is faster than that of the ECC.This paper proposes a dynamic T-NTRU secure access authentication algorithm based on time information in the Internet of things.It uses the dynamic changing time series as the secret key of the hash function to solve the internal attack security problem caused by the fixed hash function.Experiments are carried out on the computer and single chip microcomputer,respectively.Experimental results show that compared with the traditional ECC calculation,the T-NTRU algorithm proposed in this paper reduces about 97% of the calculational amount,which is equivalent to the typical NTRU algorithm,and is suitable for the need of the resource constrained power IOT network application.     

基于机器学习的网络拥塞控制研究

拥塞控制是网络研究的经典课题,可以避免网络因拥塞而性能下降。其在互联网的发展中扮演着重要的角色。近年来,随着机器学习、深度学习和强化学习的兴起,给拥塞控制提供了新的思路。对网络拥塞控制的机制进行了详细分析,阐述了国内外对于该领域的研究现状及进展,将有代表性的解决方案分为基于规则的解决方案、基于路由反馈的解决方案和智能解决方案3类,并详细分析了各方案的原理及优缺点。     

基于ArcReLU激活函数的优化研究

对常用的激活函数进行了研究比较。对曾经提出的一种新型激活函数ArcReLU进行了改进,使其具有更快的收敛速度和更低的计算消耗。实验证明,改进的ArcReLU函数既能显著加快反向传播神经网络的训练速度,又能有效降低训练误差,甚至能避免梯度消失的问题。     

基于层次分析法的电力终端安全评估方案

随着能源互联网中大量电力智能终端的部署,有效评估电力终端的安全性显得尤为重要。针对现有方案存在选取指标因素不全面、未对权重结果进行验证和评价结果主观性太强等问题,根据电力终端风险评估需求,提出了基于层次分析法的电力终端安全评估方案,模拟实验数据获取指标权重,引入模糊综合评价法来降低评估结果的片面性。实验结果表明,和已有方案相比,该方案更适用于电力终端安全评估,能有效发现存在的安全风险,且评估结果更接近实际情况。     

基于MODWT和BP神经网络的微电网故障诊断方法

近年来,随着微电网技术的持续发展,电力用户对其供电可靠性的要求也不断提高,因此微电网故障诊断研究也变得越来越重要。提出了一种基于极大重叠离散小波变换(MODWT)和反向传播(BP)神经网络的微电网故障诊断新方法,并通过仿真与算例进行了验证。结果表明:该方法能快速、准确地识别出故障类型,且不受故障初始相位角和过渡电阻等因素的影响;与现有的基于离散小波变换和反向传播神经网络的诊断方法相比所提出的方法可以提供更好的故障分类精度。     

基于多任务卷积神经网络的虹膜图像质量评估方法

为了自动、准确、高效地评估采集图像的质量,设计了一个名为MTIQA的卷积神经网络。该网络能够输出与主观评价指标保持较高一致性的客观评估结果。MTIQA采用多任务学习策略,包含网络训练质量评估和失真类型分类两个任务,将两个任务的损失融合并构成新的损失函数。为了评估算法所得到的客观指标的可靠性,建立了名为SIR2019的单眼虹膜质量评估数据集,并召集志愿者进行主观实验以得到主观评价指标。在SIR2019和CASIA-Iris-Distance-Lamp数据集上的实验结果表明,该网络在虹膜图像质量评估上具有较好的可行性、准确性和鲁棒性。     

D-2-DenseNet噪音鲁棒的城市音频分类模型

为了提高噪音环境下城市音频分类系统的鲁棒性,提出了一种双特征2阶密集卷积神经网络（D-2-DenseNet）噪音鲁棒的城市音频分类模型.首先介绍了噪音添加和噪音鲁棒处理,阐述了一种双特征互补偿的算法;然后结合2阶密集卷积神经网络与自适应机制提出了一种噪音鲁棒音频分类模型：双特征2阶密集卷积神经网络.模型采用双特征互补偿自适应算法,可在特征提取与模型训练中更有针对性地提取有效音频信息,降低噪音干扰,以提高噪音鲁棒性.最后,基于Dcase2016数据集开展噪音环境下城市音频分类测试.实验结果表明,模型分类准确率分别可达77.12%、75.52%,与基线模型相比,平均分类准确率分别提高了8.51%和10.38%,验证了模型良好的噪音鲁棒性.