基于LSTM的命名数据网络流量预测方法

网络流量预测是提高网络服务质量,提高拥塞控制效率的关键环节。命名数据网络作为新兴网络架构,其内容源与位置无关的特性使得TCP/IP网络中的拥塞控制方法不再适用,而且目前基于神经网络的流量预测方法只是研究预测模型性能的优化,使用的训练数据只考虑了单一的接口流量这一简单的维度,未考虑不同类型的数据包和出接口队列长度等参数对未来流量变化的影响,导致流量预测不够精准。文章提出一种应用于命名数据网络的基于长短期记忆网络的流量预测方法,将单一的流量特征值扩展到与链路流量相关的多维变量,实现了命名数据网络节点接收流量的预测,可用于命名数据网络流量的精准控制,避免网络拥塞。实验验证表明,与仅使用链路流量作为特征值的传统预测方法相比,预测精度有较大幅度的提高。     

一种基于用户行为预测的群移动性管理模型

移动性管理是未来5G移动的重要组成部分,如何均衡寻呼负载和位置更新开销,从而合理利用无线资源是当前的研究重点.提出了一种新的群移动性管理(GMM)方案,可以针对具有相同运动特征的用户进行集中式管理,从而可以减少单个用户移动性管理(SMM)时存在的重复开销.仿真结果表明,GMM比SMM可以获得更佳的运行开销.     

基于ELM的人体跌倒预测算法

为了应用于跌倒保护装置设计,需要设计一种跌倒保护预测算法,能够准确并快速的区分跌倒动作和正常行为动作,因此提出了基于ELM的人体跌倒预测算法.该算法通过六轴传感器芯片MPU6050提取人体各个姿态下的三相加速度和三相旋转角,通过多变量分析方法得到特征量,随后对提取的特征量进行预处理,通过滑动时间窗口对数据进行切割,对处...     

战术环境NDN基于判决选择的节点感知转发策略

传统互联网在多方面面临挑战,全新架构的命名数据网络NDN的特点使其适用于战术环境。目前无线移动NDN转发策略在提升网络效率同时会增加开销,也未考量战术环境服务发现、抗毁性和协议轻量化等关键特征。文中设计基于判决选择的节点感知转发策略(DSPAF),并用基于权重的优选内容提供者判决准则WFSRP选择中继,仿真验证该策略的优越性和适用性。     

一种基于位置和标识分离的移动性管理解决方案

基于位置和标识分离的解决方案不能很好地同时解决移动性支持和可扩展性差的问题,并且难于部署。文章提出一种新的基于位置和标识分离的网络架构,并给出与该架构相适应的移动性管理解决方案。该方案不仅能很好地解决网络的可扩展性问题,在移动性支持方面还具有如下的特点:不需要修改终端协议栈,易于部署;标识符含有一定的路由信息,能提供较好的与传统终端互通的能力;将映射系统分布于边缘网络内,增强了系统的鲁棒性并降低了切换时延。     

基于深度布隆过滤器的NDN网络三级名字查找方法

为提高命名数据网络(Name Data Networking, NDN)路由过程中内容名字查找的效率,该文提出一种基于深度布隆过滤器的3级名字查找方法。该方法使用长短记忆神经网络(Long Short Term Memory, LSTM)与标准布隆过滤器相结合的方法优化名字查找过程;采用3级结构优化内容名字在内容存储器(Content Store, CS)、待定请求表(Pending Interest Table, PIT)中的精确查找过程,提高查找精度并降低内存消耗。从理论上分析了3级名字查找方法的假阳性率,并通过实验验证了该方法能够有效节省内存、降低查找过程的假阳性。     

一种基于层次路由推送的CCN移动性支持方案

随着互联网的快速发展,用户对内容信息的需求日趋明显,当前网络架构的设计已经不能满足时下服务的要求,因此,以信息为中心的未来网络的研究得以迅速发展.内容中心网络（CCN）是未来网络的研究热点之一,然而现阶段CCN的设计未深入考虑用户移动性问题.本文针对CCN网络中的移动性管理问题,提出一种基于层次路由推送的解决方案.仿真结果表明,本文所提出的方法能有效降低用户获取内容的时间及网络中消息重传数目.     

基于动态层次位置管理的HIP移动性支持机制

针对现有HIP机制不支持节点微移动的问题,该文提出了基于动态层次位置管理的HIP移动性支持机制。在该机制中,网络划分成多个自治域,每个自治域划分成多个注册域。当节点在同一个注册域内移动时,在管理该注册域的本地集合服务点中进行位置更新;当节点在同一个自治域内移动时,在管理该自治域的网关集合服务点中进行位置更新。节点根据自己的移动速率以及呼叫到达率选取本地集合服务点并计算注册域的最佳范围。仿真结果表明,该机制能较好地降低节点移动时的信令开销,支持节点微移动。     

NDN中基于蚁群替换算法的邻居协作缓存管理策略

为提高NDN(命名数据网络)中的缓存利用率,提出了一种基于蚁群替换算法的邻居协作缓存管理(ACNCM)策略。首先将单节点的缓存替换问题,建模为0/1背包问题,并根据缓存数据的大小、使用频率以及邻居副本深度等信息定义本地存储内容的缓存价值,提出基于蚁群算法的缓存替换算法。然后利用邻域协作的思想,通过路由节点之间定期交换自身节点的缓存信息,对单个节点替换出去的缓存内容,选择邻居节点完成协作式缓存管理。实验结果表明,ACNCM策略在缓存命中率、网络开销和平均响应时延方面均优于现有方法。     

一种基于网络的全局移动性管理方法

MIPv6是由IETF标准化的全局移动性管理协议,它要求MN参与移动性管理,在MN频繁切换时会带来很大的信令开销。PMIPv6基于网络的特性使其避免了开销过大的问题,但它仅适用于本地移动性管理。为此,提出一种基于网络的全局移动性管理方法。该方法采用了级联隧道,避免了隧道嵌套的开销,同时完全基于网络的特点使其避免了无线链路上的隧道开销。仿真研究表明,相对于其他移动性管理协议,该机制在信令开销和数据传输开销方面都更有优势。     

Distributed mobility management in named data networking

The information‐centric networking concept was proposed to fulfill the scalability and efficiency requirements of the content‐centric Internet in the future. Among the multiple information‐centric networking proposals, Named Data Networking (NDN) is one of the most important representatives. NDN uses a hierarchical name to identify the data after which the on‐path cache can be deployed to improve the efficiency of data retrieval. However, with the development of mobile Internet, how to extend NDN in the mobile environment to enable efficient and scalable mobility management remains a challenge. We propose a distributed mobility management scheme for both the mobile receiver and the mobile publisher in NDN. Our proposed approach is based on the basic NDN naming and routing principles to select the branching node of the previous and new access locations of the mobile terminal after which the on‐path routing states are dynamically adjusted accordingly. Then we propose a novel analytical model to analyze the performance of the proposed scheme. The results demonstrate that the proposed scheme inherits the scalability and efficiency of NDN in the mobile Internet. Copyright © 2015 John Wiley & Sons, Ltd.     

Traffic scheduling method based on centralized control in named data networking

In order to address the global optimization problem for traffic scheduling in named data networking,related works were analyzed,a method based on centralized control was proposed.The proposed method took network performance and communication overhead into account.In the proposed scheme,appropriate nodes would be selected as E-NDN nodes,then the controller calculated the corresponding multi-path forwarding policies and sent them to E-NDN nodes according to the in-network cache,the aggregation of Interest packets,and the traffic demands of popular contents to achieve global optimization.The evaluation results indicate that the proposed method can significantly reduce the maximum link utilization and improve network performance.Simultaneously,the proposed method will not cause a large optimization cost,and communication overhead between the controller and nodes will increase slightly.     

Multipath inter-domain routing for named data networking

To facilitate content retrieval among NDN domains,a multipath inter-domain routing for named data networking(MIRNDN)was proposed.Under MIRNDN,an AS(autonomous system)maintained merely the routing state of content reachable from its own network or from its customers’ networks and aggregated routing information to mitigate routing scalability issue; “valley free”routing policy was applied to guide Interest packets that solicit content unreachable from the AS’s own network nor from its customers’ networks to explore matching data packets via multiple paths,and NDN’s requests aggregation,in-network caching and adaptive forwarding optimized such exploration; the reachability information of content from multiple paths was collected to support Interest multipath forwarding.The properties of FIB size and the convergence time and communication cost of routing update under MIRNDN were analyzed in theory.The experimental results in the AS level topology of current Internet validates that MIRNDN mitigates the scalability issue of NDN inter-domain routing,has relative short convergence time and moderate communication cost for routing update,and effectively reduces unnecessary Interest forwarding.     

Modeling denial‐of‐service against pending interest table in named data networking

Named data networking (NDN) has attracted much attention on the design for next generation Internet architecture. Although it embeds some security primitives in its original architecture, it may suffer from denial‐of‐service (DoS) attacks. In this paper, we model one representative type of NDN‐specific DoS attacks named DoS against pending interest table (PIT), or DoS‐PIT, which floods malicious Interests that request nonexistent content to bypass cached content at routers and to exhaust the memory resource for PIT, bringing in severe service degradation. In our proposed analytical model, the closed‐form expressions for the DoS probability for users suffering DoS‐PIT are derived, while considering several important factors of NDN networks such as PIT size, time‐to‐live of each PIT entry, popularity of content, and cache size. Moreover, extensive simulation experiments demonstrate the accuracy of the proposed model on evaluating the damage effect of DoS‐PIT. In addition, the proposed model can be chosen to guide designing effective countermeasures for DoS‐PIT (or attacks with similar way to harm NDN) by properly setting the values of some parameters (e.g., cache size) of each NDN router. Copyright © 2013 John Wiley & Sons, Ltd.     

Cooperative detection and protection for Interest flooding attacks in named data networking

Named data networking (NDN) is a new emerging architecture for future network, which may be a substitute of the current TCP/IP‐based network, for the content‐oriented data request mode becoming the future trend of development. The security of NDN has attracted much attention, as an implementation of next‐generation Internet architecture. Although NDN is immune to most current attack, it cannot resist the distributed denial of service like attack – Interest flooding attack (IFA) – effectively. IFA takes advantages of the forwarding mechanism of NDN, flooding a large number of malicious Interest packets at quite a high rate, and exploits the network resources, which may cause the paralysis of the network. Taking into account the severity of the destruction, we propose an algorithm to counter such new type of attack. We analyze three properties of IFA, and use them to judge and filter Interest packets. Vector space model and Markov model are used in our method to realize a cooperative detection. Meanwhile, we present the retransmission forwarding mechanism to ensure legitimate user request. The ndnSIM module of ns3 is used for the corresponding simulation, and results of the simulation will be given to show the effectiveness of our algorithm. Copyright © 2014 John Wiley & Sons, Ltd.     

基于ELM和MA的微型四频天线设计

提出一个基于极限学习机ELM（Extreme Learning Machine）和文化基因算法MA（Memetic Algorithm）的微型四频（0.92/2.4/3.5/5.8GHz）天线设计算法AntMA-ELM.为了提高天线的性能,算法在MA框架下引入基于综合学习粒子群优化算法CLPSO（Comprehensive Learning Particle Swarm Optimizer）全局搜索和DSCG（Davies,Swann,and Campey with Gram-schmidt）局部搜索,用于确定天线的几何参数.同时,建立ELM回归模型用于直接评估MA优化的适应值函数.实验结果表明,ELM回归模型能够根据输入参数正确估算天线的回波损耗,使MA算法有效提高设计性能和加速优化过程.天线在四个目标频段的回波损耗值均优于-10dB,满足设计要求.     

基于车联网的大数据应用研究

大数据技术具有数据处理量大、可处理非结构化数据、实时性高等特点.针对交通安全问题,提出将大数据技术应用于车联网中数据采集、数据分析和数据处理的问题.基于大数据技术,管理所有系统数据资源,并进行各种智能化数据分析处理.立足于为车联网提供智能支撑,让用户体验到交通管理智能化.     

基于网络社交数据分享的差分隐私保护

随着社大数据、数据挖掘的兴起和社交网络、移动设备的迅速发展,网络社交所产生的蕴涵巨额价值的海量数据在分享的过程中面临着隐私安全风险。差分隐私保护可以在保证数据可用的前提下通过添加噪声使网络社交数据失真。     

基于数据融合理论的多传感器几何定位算法

为解决多部(3部及以上)2D传感器网络对三维空间目标的定位估计和定位精度问题,克服地球曲率对观测模型的影响,建立了考虑实际地球曲率的等效地球模型和传感器观测模型,提出了此模型中基于二次数据融合的多传感器组网几何定位算法,该方法将几何定位与数据融合理论相结合,并对融合数据进行二次融合,充分利用了各传感器的量测数据。仿真实验证明了方法的有效性和实用性,在多部2D传感器组网的情况下可对三维空间内目标实现精确定位,定位误差趋近于克拉美-罗下界(CRLB),具有工程实用价值。