Laurel:一种混合式数据分发覆盖网

覆盖网是各种数据分发应用的基础架构.在节点波动的网络环境中实现快速而准确的数据分发,对覆盖网提出了两个要求:高效的数据路由;较强的系统鲁棒性.已有的覆盖网构建方法多侧重于某个方面的优化,因而未能充分权衡数据路由效率与系统鲁棒性.提出了一种混合式数据分发覆盖网--Laurel.Laurel通过簇间多重结构化拓扑与簇内非结...     

输电线路在线监测WSNs优化部署研究

应用于输电线路在线监测的无线传感器网络(WSNs)通常呈长链型,存在跳数多、时延大的问题。引入具备无线公网通信模块的异构节点能够优化WSNs时延性能。考虑网络中传感器节点分布不均匀引起的数据分布不均匀这一普遍现象及其对时延的影响,建立了异构WSNs的最大时延模型;同时考虑网段划分和异构节点部署对时延的影响,提出了一种基于局部搜索思想的网络优化部署方法来优化网络时延。仿真结果表明:上述方法能有效降低网络最大时延,提高网络实时性。     

Stepwise fair-share buffering for gossip-based peer-to-peer data dissemination

We consider buffer management in support of large-scale gossip-based peer-to-peer data dissemination protocols. Coupled with an efficient buffering mechanism, system-wide buffer usage can be optimized while providing reliability and scalability in such protocols. We propose a novel approach, stepwise fair-share buffering, that provides uniform load distribution and reduces the overall buffer usage where every peer has a partial view of the system. We report and discuss the comparative performance results with existing buffering approaches as well as random buffering which serves as a benchmark. We present separate evaluations of bufferer selection and gossip-based data dissemination. Reliability, content dissemination time, message delay, buffering delay, and minimum buffer requirements are considered as the key metrics investigated through simulations. The performance of our approach in the case of multiple senders, link failures with multiple bufferers, and scalability to larger networks are investigated. Several power-law and hierarchical overlay topologies are considered. Analytical bounds for reliability of dissemination are also provided.     

高正确率的双语语块对齐算法研究

高质量的自动对齐双语语块,对于机器翻译系统,特别是计算机辅助翻译系统的性能提高有重要作用,而且对于人工翻译以及辞典编纂也都有巨大的应用价值。该文提出基于单词间粘合度与松弛度的语块划分评分方法以及双语语块划分的双向约束算法,使得源语言和目标语言的语块的划分与对齐能相互促进。与传统方法相比,因为无需事先进行双语语块划分,而是在搜索最佳对齐时动态地考察划分效果,故可以减少边界划分错误对对齐结果的影响。该算法获得了远超过传统算法的高正确率。     

Distributed optimization of P2P live streaming overlays

Peer-to-peer live media streaming over the Internet is becoming increasingly more popular, though it is still a challenging problem. Nodes should receive the stream with respect to intrinsic timing constraints, while the overlay should adapt to the changes in the network and the nodes should be incentivized to contribute their resources. In this work, we meet these contradictory requirements simultaneously, by introducing a distributed market model to build an efficient overlay for live media streaming. Using our market model, we construct two different overlay topologies, tree-based and mesh-based, which are the two dominant approaches to the media distribution. First, we build an approximately minimal height multiple-tree data dissemination overlay, called Sepidar. Next, we extend our model, in GLive, to make it more robust in dynamic networks by replacing the tree structure with a mesh. We show in simulation that the mesh-based overlay outperforms the multiple-tree overlay. We compare the performance of our two systems with the state-of-the-art NewCoolstreaming, and observe that they provide better playback continuity and lower playback latency than that of NewCoolstreaming under a variety of experimental scenarios. Although our distributed market model can be run against a random sample of nodes, we improve its convergence time by executing it against a sample of nodes taken from the Gradient overlay. The evaluations show that the streaming overlays converge faster when our market model works on top of the Gradient overlay.     

HiWalk: Learning node embeddings from heterogeneous networks

Heterogeneous networks, such as bibliographical networks and online business networks, are ubiquitous in everyday life. Nevertheless, analyzing them for high-level semantic understanding still poses a great challenge for modern information systems. In this paper, we propose HiWalk to learn distributed vector representations of the nodes in heterogeneous networks. HiWalk is inspired by the state-of-the-art representation learning algorithms employed in the context of both homogeneous networks and heterogeneous networks, based on word embedding learning models. Different from existing methods in the literature, the purpose of HiWalk is to learn vector representations of the targeted set of nodes by leveraging the other nodes as “background knowledge”, which maximizes the structural correlations of contiguous nodes. HiWalk decomposes the adjacent probabilities of the nodes and adopts a hierarchical random walk strategy, which makes it more effective, efficient and concentrated when applied to practical large-scale heterogeneous networks. HiWalk can be widely applied in heterogeneous networks environments to analyze targeted types of nodes. We further validate the effectiveness of the proposed HiWalk through multiple tasks conducted on two real-world datasets.     

异质信息网络中基于邻接熵的影响力最大化

影响力最大化是研究如何在社会网络中寻找小部分最具影响力的节点作为信息扩散源,使信息在网络中传播范围最大的问题.已有相关研究大多只是针对同质信息网络,但现实中的社会网络是包含了多种对象类型和对象之间多种关系类型的异质网络,因此提出了基于元路径的邻接信息熵(MPAIE)模型,以及基于元图的邻接信息熵(MGAIE)模型来模拟异质信息网络中的社会影响.通过设置元路径或元图,该方法可以灵活地整合异质网络中的结构和语义信息,对节点的影响力做出度量,并在两个真实数据集上验证了MPAIE及MGAIE模型的有效性.     

Dynamic Partitioning and Popularity based Caching for Optimized Performance in content-centric fog networks: DPPCOP

With the massive growth of information generation, processing, and distribution in the Internet of Things (IoT), the existing cloud architectures need to be designed more effectively using fog networks. The current IP-address-based Internet architecture is unable to deliver the desired Quality-of-Service (QoS) towards the increasing demands of fog networking-based applications. To this end, Content-Centric Networking (CCN) has been developed as the potential future Internet architecture. CCN provides name-based content delivery and is established as an architecture for next-generation fog applications. The CCN-based fog environment uses the cache of in-network fog nodes to place the contents near the end-user devices. Generally, the caching capacity of the fog nodes is very small as compared to the content catalog size. Therefore, efficient content placement decisions are vital for improving the network performance. To enhance the content retrieval performance for the end-users, a novel content caching scheme named “Dynamic Partitioning and Popularity based Caching for Optimized Performance (DPPCOP)” has been proposed in this paper. First, the proposed scheme partitions the fog network by grouping the fog nodes into non-overlapping partitions to improve content distributions in the network and to ensure efficient content placement decisions. During partitioning, the scheme uses the Elbow method to obtain the “good” number of partitions. Then, the DPPCOP scheme analyzes the partition’s information along with the content popularity and distance metrics to place the popular contents near the end-user devices. Extensive simulations on realistic network topologies demonstrate the superiority of the DPPCOP caching strategy on existing schemes over various performance measurement parameters such as cache hit ratio, delay, and average network traffic load. This makes the proposed scheme suitable for next-generation CCN-based fog networks and the futuristic Internet architectures for industry 4.0.     

基于HBase的多分类逻辑回归算法研究

为解决在大数据环境下,用于训练多分类逻辑回归模型的数据集可能会超过执行计算的客户端内存的问题,提出了块批量梯度下降算法,用于计算回归模型的系数。将训练数据集存入HBase后,通过设置表扫描对象的起始行键参数,可取出大小合适的含训练样本及结果值的数据块,同时为避免客户端到服务端频繁的RPC调用,取出的数据块可进行多次迭代计算,以加快系数的收敛。当取出的数据块达到指定的迭代次数后,再按行键次序取出下一个数据块,如此循环,直到系数收敛或达到指定的循环控制阈值。多分类的逻辑回归问题可转换为二分类来解决,因此需要为每一个分类在训练数据表中设定结果值列,结合训练样本列簇,按块批量梯度下降算法得到每个分类的回归系数。实验结果表明,得到的回归系数,能准确的对测试样本进行分类。     

基于SVM的组块识别及其错误驱动学习方法

给出了一种错误驱动学习机制与SVM相结合的汉语组块识别方法。该方法在SVM组块识别的基础上,对SVM识别结果中的错误词语序列的词性、组块标注信息等进行分析,获得候选校正规则集;之后按照阈值条件对候选集进行筛选,得到最终的校正规则集;最后应用该规则集对SVM的组块识别结果进行校正。实验结果表明,与单独采用SVM模型的组块识别相比,加入错误驱动学习方法后,组块识别的精确率、召回率和F值均得到了提高。     

Adding structure to unstructured peer-to-peer networks: the use of small-world graphs

The “small-world” graph structure is pervasive and is observed to arise “without-design” or “naturally” in many practical systems such as the World Wide Web. In contrast to natural systems, overlay networks provide an opportunity to design structure. We seek the advantages of designing overlay topologies with small-world properties to support file sharing in peer-to-peer networks. We focus on two metrics of performance: (a) search protocol performance, a local gain perceived directly by peer-to-peer network users and (b) network utilization, a global property that is of interest to network service providers. We propose a class of overlay topologies and show, by simulation, that a particular topology instance of this class where every node has many close neighbors and few random neighbors (i.e., a small-world graph) exhibits very good properties. In this overlay topology, the chances of locating files are high, and the nodes where these files are found are, on average, close to the query source. This improvement in search protocol performance is achieved while decreasing the traffic load on the links in the underlying network. We propose a simple greedy algorithm to construct such overlay topologies where each node operates independently and in a decentralized manner to select its neighbors.     

汉语基本块描述体系

块分析是自然语言处理研究中的重要技术,其处理基础是设计一套合理有效的块描述体系。本文在吸收和总结前人研究成果和经验的基础上,提出了一套基于拓扑结构的汉语基本块描述体系。它通过引入词汇关联信息确定基本拓扑结构,形成了很好的基本块内聚性判定准则,建立了句法形式与语义内容的有机联系桥梁。这套描述体系大大简化了从现有的句法树库TCT中自动提取基本块标注语料库和相关词汇关联知识库的处理过程,为进一步进行汉语基本块自动分析和词汇关联知识获取互动进化研究打下了很好的基础。     

A performance evaluation of scalable live video streaming with nano data centers

To improve the efficiency and the quality of a service, a network operator may consider deploying a peer-to-peer architecture among controlled peers, also called here nano data centers, which contrast with the churn and resource heterogeneity of peers in uncontrolled environments. In this paper, we consider a prevalent peer-to-peer application: live video streaming. We demonstrate how nano data centers can take advantage of the self-scaling property of a peer-to-peer architecture, while significantly improving the quality of a live video streaming service, allowing smaller delays and fast channel switching. We introduce the branching architecture for nano datacenters (BAND), where a user can “pull” content from a channel of interest, or content could be “pushed” to it for relaying to other interested users. We prove that there exists an optimal trade-off point between minimizing the number of push, or the number of relaying nodes, and maintaining a robust topology as the number of channels and users get large, which allows scalability. We analyze the performance of content dissemination as users switch between channels, creating migration of nodes in the tree, while flow control insures continuity of data transmission. We prove that this p2p architecture guarantees a throughput independently of the size of the group. Analysis and evaluation of the model demonstrate that pushing content to a small number of relay nodes can have significant performance gains in throughput, start-up time, playback lags and channel switching delays.     

一种异构环境下覆盖多播网络路由算法

由于IP多播在部署上的困难,覆盖多播网络(overlay multicast networks,简称OMN)作为构建通用的多播服务平台的另一可行途径正不断为人们所认可.针对实时多媒体应用对带宽需求的异构性,研讨了异构环境下OMN的路由问题.通过对度约束模型进行扩展,描述了一种新的适应异构环境的OMN网络模型.采用分层的带宽分配策略,提出了一种异构环境下构造OMN最小延时半径多播树的启发式算法--分层的压缩树(1ayered compact tree,简称LCT)算法,并对其性质进行了理论证明和分析.仿真实验结果表明,随着分配带宽的减少,LCT算法能够有效地降低多播树的高度和网络资源使用量,并保持较低的多播树延时半径增幅.     

一种移动容迟网络体系结构及其应用

移动容迟网络利用节点移动带来的通信机会,在非全连通的网络中提供低成本高可用性的传输服务.移动公交车组成的网络是典型的移动容迟网络,城市中公交车网络具有覆盖范围广、移动路线固定等优点,适合城市环境信息收集和分发等多种应用.结合移动容迟网络的特性,提出了层次化的移动容迟网络体系结构和3层网络架构,结合城市中公交车辆,研究并设计了若干关键技术,包括Internet快速接入机制、公平的数据收集机制、基于地理位置的数据分发机制和缓存管理与调度机制.通过真实场景下的实验和基于trace的仿真实验,验证了提出的算法和机制具有较好的性能和鲁棒性.