Optimized algorithms for predictive range and KNN queries on moving objects

There have been many studies on management of moving objects recently. Most of them try to optimize the performance of predictive window queries. However, not much attention is paid to two other important query types: the predictive range query and the predictive k nearest neighbor query. In this article, we focus on these two types of queries. The novelty of our work mainly lies in the introduction of the Transformed Minkowski Sum, which can be used to determine whether a moving bounding rectangle intersects a moving circular query region. This enables us to use the traditional tree traversal algorithms to perform range and kNN searches. We theoretically show that our algorithms based on the Transformed Minkowski Sum are optimal in terms of the number of tree node accesses. We also experimentally verify the effectiveness of our technique and show that our algorithms outperform alternative approaches.     

分布式超级节点选举算法

基于超级节点的分布式系统中,若超级节点失效或临时离开,希望系统能够自组织地选举出能力最强的节点作为新的超级节点。提出分布式超级节点选举算法,通过洪泛过程构造底层的生成树,叶子节点沿此树进行消息的传递,消息中包含着关于节点和边的信息,根节点根据这些信息构造最小生成树。根节点选出能力最强的节点作为超级节点,并沿着最小生成树广播选举结果。对算法性能从通信复杂度和时间复杂度两方面进行了分析和比较。     

Large-scale indexing of spatial data in distributed repositories: the SD-Rtree

We propose a scalable distributed data structure (SDDS) called SD-Rtree. We intend our structure for point, window and kNN queries over large spatial datasets distributed on clusters of interconnected servers. The structure balances the storage and processing load over the available resources, and aims at minimizing the size of the cluster. SD-Rtree generalizes the well-known Rtree structure. It uses a distributed balanced binary tree that scales with insertions to potentially any number of storage servers through splits of the overloaded ones. A user/application manipulates the structure from a client node. The client addresses the tree through its image that can be possibly outdated due to later split. This may generate addressing errors, solved by the forwarding among the servers. Specific messages towards the clients incrementally correct the outdated images. We present the building of an SD-Rtree through insertions, focusing on the split and rotation algorithms. We follow with the query algorithms. We describe then a flexible allocation protocol which allows to cope with a temporary shortage of storage resources through data storage balancing. Experiments show additional aspects of SD-Rtree and compare its behavior with a distributed quadtree. The results justify our various design choices and the overall utility of the structure.     

Efficient Collective Communications in Dual-Cube

The hypercube, or n-cube, has been widely used as the interconnection network in parallel computers. However, the major drawback of the hypercube is the increase in the number of communication links for each node with the increase in the total number of nodes in the system. This paper introduces a new interconnection network, namely dual-cube, for large-scale parallel computers and describes the algorithms for efficient collective communications in dual-cube. The dual-cube network mitigates the problem of increasing number of links in the large-scale hypercube network while retains hypercube's topological properties. Design of efficient routing algorithms for collective communications is the key issue for any interconnection network. In this paper, we show that the collective communications can be done in dual-cube with almost the same communication times as in hypercube.     

一种支持多维区间查询的云对等网络索引架构*

针对用户在大规模云对等网络环境下多维区间查询问题,将基于m叉平衡树的索引架构引入到云对等网络环境下,在该架构上实现集中式环境下支持多维数据索引的层次化树结构,例如R树,QR树等。多维区间查询算法保证查询从树的任意位置开始,避免了根节点引起的系统性能瓶颈问题。通过计算和实验验证,对于N个节点的网络,多维区间查询效率为O(logmN)(m>2)(m表示扇出),由此可见,查询效率和维数d无关,查询效率不会随着维数d的增加而降低。最后建立基于扇出m的代价模型,并且计算出了最优的m值。     

无线传感器网络不重复记录求和近似算法

针对现有的求和算法基本上都是对副本敏感的算法,提出一种对副本不敏感的求和近似算法FM-S。网络中各节点由FM-S和服从二项分布的随机数样本对节点记录进行哈希转换以填充一个长度为L的二进制求和序列,并且每个节点会把生成的序列转发给路由树中的父亲节点,根节点将接收到全网的求和序列,最终根据此序列可计算出网络中不重复记录求和的近似值。实验结果显示该算法是一种分布式、低功耗、容错性高、扩展性和健壮性强的聚集查询算法。     

Optimizing in-network aggregate queries in wireless sensor networks for energy saving

This study proposes a method of in-network aggregate query processing to reduce the number of messages incurred in a wireless sensor network. When aggregate queries are issued to the resource-constrained wireless sensor network, it is important to efficiently perform these queries. Given a set of multiple aggregate queries, the proposed approach shares intermediate results among queries to reduce the number of messages. When the sink receives multiple queries, it should be propagated these queries to a wireless sensor network via existing routing protocols. The sink could obtain the corresponding topology of queries and views each query as a query tree. With a set of query trees collected at the sink, it is necessary to determine a set of backbones that share intermediate results with other query trees (called non-backbones). First, it is necessary to formulate the objective cost function for backbones and non-backbones. Using this objective cost function, it is possible to derive a reduction graph that reveals possible cases of sharing intermediate results among query trees. Using the reduction graph, this study first proposes a heuristic algorithm BM (standing for Backbone Mapping). This study also develops algorithm OOB (standing for Obtaining Optimal Backbones) that exploits a branch-and-bound strategy to obtain the optimal solution efficiently. This study tests the performance of these algorithms on both synthesis and real datasets. Experimental results show that by sharing the intermediate results, the BM and OOB algorithms significantly reduce the total number of messages incurred by multiple aggregate queries, thereby extending the lifetime of sensor networks.     

A distributed multicast routing protocol for real-time multicast applications

Multicast routing is establishing a tree which is rooted from the source node and contains all the multicast destinations. A delay bounded routing tree is a tree in which the accumulated delay from the source node to any destination along the tree does not exceed a pre-specified bound. This paper presents a distributed routing protocol which constructs delay bounded routing trees for real-time multicast connections. A constructed routing tree has a near optimal network cost under the delay bound constraint. The proposed algorithm is fully distributed, efficient in terms of the number of messages required, and flexible in multicast membership changes. A large number of simulations have been done to show the network cost of the routing trees generated by our method is better than the other major existing algorithms.     

网格环境下基于流水线的多重相似查询优化

提出一种网格环境下基于流水线技术的分布式多重相似查询的优化算法(pipeline-based distributed similarity query processing,简称pGMSQ).首先,当用户提交若干个查询请求时,采用基于代价的动态层次聚类策略(dynamic query clustering,简称DQC)对其进行合并.然后在数据结点层,采用索引支持的向量集缩减方法快速过滤无关向量.最后,在执行结点层对候选向量执行求精操作返回结果向量.由于本查询采用了流水线技术,实验结果表明,该方法在提高查询性能的同时也提高了系统的吞吐量.     

基于子树间快捷连接的非结构化P2P资源搜索方法

通过对非结构化P2P网络资源搜索方法的研究,提出了一种P2P覆盖网络。网络采用多路平衡树形拓扑结构,当新节点加入时在其与根节点的各子树间创建仅参与查询消息第一跳转发的快捷连接,搜索过程中利用快捷连接实现查询消息在根的各子树间并行转发并且不依赖根节点。仿真比较了所提网络同Gnutella和随机漫步网络的性能,所提网络具有低消息冗余率、高搜索成功率和较低的平均搜索时间复杂度。分析结果表明所提网络在资源搜索方面是高效可行的。     

A Distributed Algorithm for Computing the Node Search Number in Trees

We present a distributed algorithm to compute the node search number in trees. This algorithm extends the centralized algorithm proposed by Ellis et al. (Inf. Comput. 113(1):50–79, 1994). It can be executed in an asynchronous environment, requires an overall computation time of O(nlog?n), and n messages of log?₃ n+4 bits each. The main contribution of this work lies in the data structure proposed to design our algorithm, called hierarchical decomposition. This simple and flexible data structure is used for four operations: updating the node search number after addition or deletion of any tree-edges in a distributed fashion; computing it in a tree whose edges are added sequentially and in any order; computing other graph invariants such as the process number and the edge search number, by changing only initialization rules; extending our algorithms for trees and forests of unknown size (using messages of up to 2log?₃ n+5 bits).     

Hierarchical Hypercube Networks (HHN) for Massively Parallel Computers

The hypercube is one of the most widely used topologies because it provides small diameter and embedding of various interconnection networks. For very large systems, however, the number of links needed with the hypercube may become prohibitively large. In this paper, we propose a hierarchical interconnection network based on hypercubes called hierarchical hypercube network (HHN) for massively parallel computers. The HHN has a smaller number of links than the comparable hypercube and in particular, when we construct networks with 2^Knodes, the node degree of HHN with the minimum node degree isO([formula]) while that of hypercube isO(K). Regardless of its smaller node degree, many parallel algorithms can be executed in HHN with the same time complexity as in the hypercube.     

Burst Mode Bandwidth Allocation for Real-Time Messages in IEEE 802.12 Networks

In IEEE 802.12 network, a repeater arbitrates among transmission requests on a round robin basis. A node can transmit a packet only when it is granted by the repeater. Recently, the IEEE 802.12 Committee approved a burst mode in which a node can transmit multiple packets per grant. In burst mode a bandwidth, or the number of packets per grant should be allocated taking into account the timing constraints of real-time messages. This paper proposes bandwidth allocation algorithms based on the burst mode to guarantee the deadlines of periodic real-time messages. Once the bandwidth of each node is derived from the lengths and deadlines of periodic messages, a node is allowed to transmit packets up to the allocated bandwidth per grant. Experimental results show that the proposed algorithms providemuch higher guarantee ratio and graceful degradation for heavy network load, compared with the existing approaches.     

对等网络中基于节点相似性的改进破圈搜索算法

对等网络中资源搜索算法的研究是近十几年的一个研究热点.当前无结构对等网中的搜索算法很多是基于泛洪算法的改进,这些算法在网络中查全率和查准率都比较高,但由于在搜索过程中产生大量的冗余信息导致对等网的扩展性很差,不能满足大范围组网的要求.本文提出一种对等网中基于稳定性和相似性的改进破圈路由搜索算法,在此算法中,如果节点在路由中发现与另外几个节点形成一个圈,则节点根据与圈中邻居节点之间的稳定性和内容相似性等因素选择其中的一个邻居节点转发信息.在这种破圈法构造的拓扑结构中采用泛洪协议去搜索资源,转发信息数将从指数级增长有效地减少到常数值,并且资源搜索成功率始终在一个稳定的范围.仿真实验表明所提算法在保证一定搜索成功率的前提下大大减少了搜索中产生的冗余信息.     

Bandwidth-constrained queries in sensor networks

Sensor networks consist of battery-powered wireless devices that are required to operate unattended for long periods of time. Thus, reducing energy drain is of utmost importance when designing algorithms and applications for such networks. Aggregate queries are often used by monitoring applications to assess the status of the network and detect abnormal behavior. Since radio transmission often constitutes the biggest factor of energy drain in a node, in this paper we propose novel algorithms for the evaluation of bandwidth- constrained queries over sensor networks. The goal of our techniques is, given a target bandwidth utilization factor, to program the sensor nodes in a way that seeks to maximize the accuracy of the produced query results at the monitoring node, while always providing strong error guarantees to the monitoring application. This is a distinct difference of our framework from previous techniques that only provide probabilistic guarantees on the accuracy of the query result. Our algorithms are equally applicable when the nodes have ample power resources, but bandwidth consumption needs to be minimized, for instance in densely distributed networks, to ensure proper operation of the nodes. Our experiments with real sensor data show that bandwidth-constrained queries can substantially reduce the number of messages in the network while providing very tight error bounds on the query result.     

PeerCluster: A Cluster-Based Peer-to-Peer System

This paper proposes a cluster-based peer-to-peer system, called PeerCluster, for sharing data over the Internet. In PeerCluster, all participant computers are grouped into various interest clusters, each of which contains computers that have the same interests. The intuition behind the system design is that by logically grouping users interested in similar topics together, we can improve query efficiency. To efficiently route and broadcast messages across/within interest clusters, a hypercube topology is employed. In addition, to ensure that the structure of the interest clusters is not altered by arbitrary node insertions/deletions, we have devised corresponding JOIN and LEAVE protocols. The complexities of these protocols are analyzed. Moreover, we augment PeerCluster with a system recovery mechanism to make it robust against unpredictable computer/network failures. Using an event-driven simulation, we evaluate the performance of our approach by varying several system parameters. The experimental results show that PeerCluster outperforms previous approaches in terms of query efficiency, while still providing the desired functionality of keyword-based search.     

基于CURE聚类算法的静态R树构建方法

R树索引结构在空间对象查询和复杂空间关系查询方面具有重要作用。传统空间索引结构R树是动态生成的,树的结构是根据连续插入算法实现的,通过分裂子节点直至生成R树的根节点。动态生成算法会导致R树节点最小外包矩形之间的大量重叠,影响空间查询效率,且空间利用率不高。为了弥补动态生成R树的不足,提出了基于CURE算法的静态R树生成方法,给出CU_RHbuilt建树算法,该算法不仅能有效地处理海量数据,识别任何形状的簇,减少矩形重叠度,而且采用划分技术可较大程度地减小计算代价,空间利用率较高。进一步提出了基于CURE算法的R树节点分裂方法。理论研究与实验表明,所提方法具有较高的查询效率。     

ETI: an efficient index for set similarity queries

Set queries are an important topic and have attracted a lot of attention. Earlier research mainly concentrated on set containment queries. In this paper we focus on the T-Overlap query which is the foundation of the set similarity query. To address this issue, unlike traditional algorithms that are based on an inverted index, we design a new paradigm based on the prefix tree (trie) called the expanded trie index (ETI) which expands the trie node structure by adding some new properties. Based on ETI, we convert the TOverlap problem to finding query nodes with specific query depth equaling to T and propose a new algorithm called TSimilarity to solve T-Overlap efficiently. Then we carry out a three-step framework to extend T-Overlap to other similarity predicates. Extensive experiments are carried out to compare T-Similarity with other inverted index based algorithms from cardinality of query, overlap threshold, dataset size, the number of distinct elements and so on. Results show that T-Similarity outperforms the state-of-the-art algorithms in many aspects.     

External segment trees

The segment tree is a well-known internal data structure with numerous applications in computational geometry. It allows the dynamical maintenance of a set of intervals such that the intervals enclosing a query point can be found efficiently (point enclosure search).In this paper we transfer the underlying principle of the segment tree in a nontrivial way to secondary storage and arrive at the EST-an external file structure with the same functionality and the following properties: (1) Point enclosure searches are very efficient—only very few pages are accessed that are not filled to more than 50% with result intervals. (2) A page filling of 50% is guaranteed—on the average it will be around 70%. Although the segment tree represents, in the worst case, each interval by a logarithmic number offragments, in practical cases fragmentation remains low and the storage requirements about linear. (3) The EST is balanced and the update algorithms are efficient. (4) Unlike many other file structures for spatial objects the EST has no problems with an arbitrarydensity, that is, an arbitrarily large number of intervals covering any point of the line.Furthermore, the EST can be used as a file structureconstructor in the following sense: Let there be a file structureX supporting searches for objects with propertyx and suppose it is necessary to maintain a collection of objects with associated (e.g., time) intervals. Then an EST-X structure that supports searches for objects with propertyx present at timet can be built. This suggests using the EST as a building block in the implementation of temporal database systems. Other applications include the one-dimensional indexing of collections of spatial objects in two or more dimensions.More generally, this paper shows techniques for mapping internal tree structures with node lists (other examples: range tree, interval tree) to secondary memory. In this context an intriguing theoretical problem, thecover-balancing problem, is solved: Given a tree whose nodes have associatedweights partitioned into subtrees whose weights must lie in a certain range, maintain this partition under weight changes at arbitrary nodes. This is in contrast to classical balancing problems where updates occur only at the leaves.This work was supported by the DFG (Deutsche Forschungsgemeinschaft) under Grant Cr 65/2-5.     

Distributed mutual exclusion algorithms for grid applications: A hierarchical approach

The majority of current distributed mutual exclusion algorithms are not suited for parallel or distributed applications on a Grid as they do not consider the heterogeneity of latency on Grids. We propose two distributed mutual exclusion algorithms, based on Naimi–Trehel's token-based algorithm, which take into account latency gaps, especially those between local and remote clusters of machines. Our first algorithm exploits cluster locality by giving higher priority to critical section requests issued from nodes of the same cluster when compared to those from remote nodes. Our second algorithm adds a router layer to the first algorithm, bringing it closer to Grid network topology. Viewing each cluster as a single node, the Naimi–Trehel algorithm is applied to this router layer. Redirection of inter-cluster messages to cluster's nodes is then minimized.