一种新的基于LRU的大流检测算法

针对LRU(Least Recently Used)算法大流漏检率过高的问题,提出了一种采用CBF(Counting Bloom Filter)和LRU两级结构的大流检测算法。该算法根据大流占用带宽较大、持续时间较长的特点在CBF中引入时间窗口机制来滤出可能的大流,然后将流信息记录在LRU中作进一步筛选。从理论上分析了影响该算法性能的因素,考虑了在存储资源一定的条件下,如何设置参数以发挥该算法最大效能的问题。最后基于实际的互联网数据进行了对比验证。结果表明,与同类算法相比,该算法在有效降低大流漏检率的同时,提高了大流流量的测量精度。     

一种基于大小流区分计数的公平抽样算法

针对一种草图指导公平抽样(SGS)算法对小流估计误差大的问题,该文提出一种基于大小流区分计数的包公平抽样算法(DCMFS),并给出哈希冲突对SGS算法估计误差影响的定量分析结果。DCMFS采用大小流区分计数器,对小流采用逐流精确计数,对大流采用哈希计数。理论分析及实际的数据仿真结果均表明,DCMFS算法对小流能够实现逐流精确统计,对大流的估计标准差接近公平抽样估计标准差理论值上限。算法采用不等长位宽计数器结构,保证其空间复杂度较SGS和自适应非线性抽样方法(ANLS)没有增加;引入计数器置换使得算法时间复杂度略有提高,但仍能满足10 Gbps线速处理要求。     

基于LRU_CBF的大流识别算法

对网络中的大流进行提取和分析对于网络管理和安全防御具有重要意义.文章通过把最近最久未使用(LRU)策略和计数型布鲁姆过滤器(CBF)两种结构结合起来,取其各自的优点,提出一种新的大流检测算法.该算法针对大流检测漏报率高的缺陷,将“大流过滤”和“大流判断”分离,提高了算法的准确性,降低了空间复杂度.最后通过理论分析和仿真实验进行了算法的验证.     

通信

TNgl 97020895一种改进的数宇式波束形成的快速自适应算法/王忠,张继宏,黄顺吉(电子科技大学)刀航天电子对抗一1996,(4)一37~42 提出了一种改进的数字式波束形成的快速自适应算法,即具有严格约束最小功率的抽样矩阵梯度算法(C SMG),它综合了传统抽样矩阵梯度算法和具有严格约束     

改进的否定选择算法在入侵检测系统中的应用

人工免疫系统中的一个重要的核心算法就是否定选择算法(Negative Select Algorithm),该算法用来生成成熟检测器,它是对免疫细胞成熟过程的模拟。否定选择算法的检测器是随机生成的,因此检测器集合冗余度高、对非自体空间的覆盖率低。针对这一问题,本文提出一种改进算法,通过对检测器进行二次筛选来降低检测器的重复率和冗余度。实验证明,在基于免疫原理的入侵检测模型中改进算法可以将模型的正确检测率提高10%,漏检率降低3%,该算法是有效的。     

改进YOLO V3遥感图像飞机识别应用

为了准确识别遥感图像中的飞机,基于YOLO V3算法,通过使用K-means算法对数据集进行聚类分析,借鉴Densenet网络的思想,将YOLO V3网络中的两个残差网络模块替换为两个密集网络模块,改进为一种Dense-YOLO深度卷积神经网络结构。对改进前与改进后的网络进行训练,分别选出使两个网络识别效果最好的权重文件,针对高质量遥感图像与过度曝光、云雾遮挡等低质量遥感图像分别进行测试与分析。实验结果表明,新改进的深度卷积神经网络应用在两种图像上效果均有提升。其中,改进的算法在高质量的遥感图像中准确率高达99.72%,比原始算法准确率提升了0.85%,召回率高达98.34%,召回率提升了1.94%。在低质量遥感图像中准确率高达96.12%,比原始算法准确率提升了5.07%,召回率高达93.10%,召回率提升了19.75%。     

基于LRU-BF策略的网络流量测量算法

针对传统大流检测算法漏检率高的缺陷,提出了一种基于LRU-BF(least recent used&Bloom filter)策略的流量测量算法。该算法使用LRU淘汰机制、Bloom filter快速表示方案,将"大流过滤"和"大流判断"分离,较大地提高了测量的准确性。基于"概率论"的相关知识,对算法进行了理论分析,建立了错误概率上界的解析表达式。仿真结果表明:与传统Na ve-LRU算法相比,LRU-BF具有较低错误概率和空间复杂度的同时,也能满足单线路10Gbit/s的线速报文处理能力。     

抽样子空间约束改进大数据谱聚类算法

在分析经典谱聚目标函数与加权核k-means目标函数等价基础上,设计了一种基于抽样子空间约束的改进大规模数据谱聚类算法,算法通过加权核k-means迭代优化避免矩阵特征分解的大量资源被占用,通过数据抽样及聚类中心的子空间约束,避免全部核矩阵都被使用,从而降低经典算法的时间空间复杂度。理论分析和实验结果表明,改进算法保持与经典算法相近聚类精度,提高了聚类效率,验证了改进算法的有效性。     

基于LRU的大流检测算法

高速网络中,检测大流是进行准确流量测量的一种重要可扩展解决方案。该文提出了一种新的基于LRU大流检测算法。它通过引入小流早期丢弃和大流预保护机制以提高测量准确性。算法分析表明:新算法具有10Gbps线速处理能力。该文基于实际互联网数据进行了实验比较,结果显示:与已有算法相比,新算法具有更高的测量准确性和实用性。     

基于流量负载的改进型几何抽样算法

为了解决在流量高峰时期网络节点由于超负荷工作导致报文丢失的问题,在几何抽样的基础上,设计了一种能自动适应流量负载变化的抽样算法.该算法在流量高峰时会根据负载情况动态地计算最佳的抽样概率来匹配节点的处理能力从而降低节点丢失的报文数,在流量负载轻时会自动地提高抽样概率以充分利用节点的处理性能.通过真实网络流量数据的实验分析表明,改进后的几何抽样算法不仅能有效降低节点丢弃的报文数,同时还提高了网络测量的精度,证明了该改进算法对流量负载具有良好的适应性.     

基于图像抽样重组的2维核鉴别分析

2维核鉴别分析(2DKDA)存在离散度量矩阵过大而无法计算的问题。该文通过将图像抽样重组与2DKDA的结合,提出了3种基于图像抽样重组的2DKDA(SR2DKDA),它们不仅克服了2DKDA在计算上的困难,识别性能也优于2维线性鉴别分析(2DLDA)。在ORL人脸库和UMIST人脸库的实验验证了SR2DKDA的有效性。     

Video super-resolution reconstruction based on deep convolutional neural network and spatio-temporal similarity

Existing learning-based super-resolution (SR) reconstruction algorithms are mainly designed for single image, which ignore the spatio-temporal relationship between video frames. Aiming at applying the advantages of learning-based algorithms to video SR field, a novel video SR reconstruction algorithm based on deep convolutional neural network (CNN) and spatio-temporal similarity (STCNN-SR) was proposed in this paper. It is a deep learning method for video SR reconstruction, which considers not only the mapping relationship among associated low-resolution (LR) and high-resolution (HR) image blocks, but also the spatio-temporal non-local complementary and redundant information between adjacent low-resolution video frames. The reconstruction speed can be improved obviously with the pre-trained end-to-end reconstructed coefficients. Moreover, the performance of video SR will be further improved by the optimization process with spatio-temporal similarity. Experimental results demonstrated that the proposed algorithm achieves a competitive SR quality on both subjective and objective evaluations, when compared to other state-of-the-art algorithms.     

A hybrid evolutionary algorithm for multiobjective sparse reconstruction

Sparse reconstruction (SR) algorithms are widely used in acquiring high-quality recovery results in compressed sensing. Existing algorithms solve SR problem by combining two contradictory objectives (measurement error and sparsity) using a regularizing coefficient. However, this coefficient is hard to determine and has a large impact on recovery quality. To address this concern, this paper converts the traditional SR problem to a multiobjective SR problem which tackles the two objectives simultaneously. A hybrid evolutionary paradigm is proposed, in which differential evolution is employed and adaptively configured for exploration and a local search operator is designed for exploitation. Another contribution is that the traditional linearized Bregman method is improved and used as the local search operator to increase the exploitation capability. Numerical simulations validate the effectiveness and competitiveness of the proposed hybrid evolutionary algorithm with LB-based local search in comparison with other algorithms.     

A nonlinear least square technique for simultaneous image registration and super-resolution.

This paper proposes a new algorithm to integrate image registration into image super-resolution (SR). Image SR is a process to reconstruct a high-resolution (HR) image by fusing multiple low-resolution (LR) images. A critical step in image SR is accurate registration of the LR images or, in other words, effective estimation of motion parameters. Conventional SR algorithms assume either the estimated motion parameters by existing registration methods to be error-free or the motion parameters are known a priori. This assumption, however, is impractical in many applications, as most existing registration algorithms still experience various degrees of errors, and the motion parameters among the LR images are generally unknown a priori. In view of this, this paper presents a new framework that performs simultaneous image registration and HR image reconstruction. As opposed to other current methods that treat image registration and HR reconstruction as disjoint processes, the new framework enables image registration and HR reconstruction to be estimated simultaneously and improved progressively. Further, unlike most algorithms that focus on the translational motion model, the proposed method adopts a more generic motion model that includes both translation as well as rotation. An iterative scheme is developed to solve the arising nonlinear least squares problem. Experimental results show that the proposed method is effective in performing image registration and SR for simulated as well as real-life images.     

用于ATM网络的拥塞避免算法（CAPCA）的稳定性分析

ＡＴＭ论坛流量管理规程４．０给出基于速率（Ｅｘｐｌｉｃｉｔ Ｒａｔｅ）的ＡＢＲ拥塞控制算法的三个例子（ＥＰＲＣＡ。ＣＡＰＣＡ,ＥＲＩＣＡＡ）。在文中,我们研究了ＣＡＰ（收敛性,稳定性,敏感性…）,我们证明了ＣＡＰＣＡ是边缘稳定的,并且对许多参数敏感,本文对ＣＡＰＣＡ作了改进,仿真结果表明,改进后的ＣＡＰＣＡ是稳定的。     

超分辨率技术在红外微扫描中的应用

超分辨率(SR)重建技术已经成为目前热门的数字图像处理技术之一,为了验证SR算法在红外成像领域中的应用效果,设计了相关红外长波微扫描实验平台。通过对微扫描获得的图像序列进行多种SR算法重建,然后对比平台光学参数所确定的实际光学衍射限,最后计算图像的调制传递函数(MTF)值来分析SR算法带来的图像空间分辨率提升。实验结果...     

SR3: secure resilient reputation-based routing

In this paper, we propose SR3 (which means secure resilient reputation-based routing), a secure and resilient algorithm for convergecast routing in wireless sensor networks. SR3 uses lightweight cryptographic primitives to achieve data confidentiality and unforgeability. Security of SR3 has been proven formally using two verification tools: CryptoVerif and Scyther. We made simulations to show the resiliency of SR3 against various scenarios, where we mixed selective forwarding, blackhole, wormhole, and Sybil attacks. We compared our solution to several routing algorithms of the literature. Our results show that the resiliency accomplished by SR3 is drastically better than the one achieved by those protocols, especially when the network is sparse. Moreover, unlike previous solutions, SR3 self-adapts after compromised nodes suddenly change their behavior.     

无线Mesh网络安全路由算法研究

从路由安全性的角度出发,对无线Mesh网络中的多径路由(MPR)算法和随机路由(SR)算法进行了研究.在分析这两种算法的思路、实现步骤和安全性的基础上,对MPR算法和SR算法进行了仿真.仿真结果表示SR算法在路由安全性能上优于MPR算法,但在网络时延性能上劣于MPR算法.     

Temporal Reference Algorithms versus Spatial Reference Algorithms for Smart Antennas

This paper compares algorithms from three different classes of adaptation schemes for smart antennas – the switched beam approach, a Temporal-Reference (TR) technique based on Direct Matrix Inversion or Least Squares adaptation, and a Spatial-Reference (SR) technique with direction finding by Unitary ESPRIT.The simulations are based on a channel model including directions of arrival (DOA) and flat Rayleigh fading. The fading signal is spread out in angle over several degrees dependent on the distance of the mobile from the base station.The results indicate that, in the uplink, TR algorithms and SR algorithms perform equally well, given perfect synchronization and successful user identification (for SR algorithms). TR algorithms are the most robust against close-by interference. For a Uniform Linear Array (ULA) with M=8 elements and an element spacing ofd equal to half a wavelength, they are able to separate two co-channel users that are as close as 5 degrees in angle. For an angular threshold as low as 10° an 8-element ULA is also sufficient to obtain nearly the same BER performance as for a single user.As concerns SR algorithms, we demonstrate their applicability to situations where no discrete DOAs exist, as it is the case in mobile communications.Mutual coupling of individual antenna elements increases the BER by an order of magnitude for the switched beam approach and for SR algorithms, in contrast to TR algorithms, where the influence of mutual coupling on the BER is negligible.     

Two ADPCM Algorithms with Widely Separated Error Recovery Times

Two adaptive differential pulse code modulation (ADPCM) algorithms are discussed that provide for transmitter-receiver resynchronization when used in a non-zero error rate digital transmission medium. Novel techniques are used in the algorithms (e.g., step-size bias) and in their extensive characterization (e.g., intermodulation distortion measurements).