G-Tree: An Efficient and Scalable Index for Spatial Search on Road Networks

Zhong, Ruicheng; Li, Guoliang; Tan, Kian-Lee; Zhou, Lizhu; Gong, Zhiguo

doi:10.1109/tkde.2015.2399306

Cited by 148 publications

(205 citation statements)

References 19 publications

(58 reference statements)

Supporting

Mentioning

203

Contrasting

Order By: Relevance

“…Both ER and NE use the R-tree to speed up query processing. Zhong et al proposed the G-tree, a hierarchy structure, for processing nearest neighbor queries in the road network [21]. Assuming only the broadcast communication is available, Sun et al proposed the Network Partition Index (NPI) for processing range queries and nearest neighbor queries in the road network [22].…”

Section: Related Workmentioning

confidence: 99%

The SSP-Tree: A Method for Distributed Processing of Range Monitoring Queries in Road Networks

Jung

Kim

2017

IJGI

View full text Add to dashboard Cite

This paper addresses the problem of processing range monitoring queries, each of which continuously retrieves moving objects that are currently located within a given query range. In particular, this paper focuses on processing range monitoring queries in the road network, where movements of the objects are constrained by a predefined set of paths. One of the most important challenges of processing range monitoring queries is how to minimize the wireless communication cost and the server computation cost, both of which are heavily dependent on the amount of location-update stream generated by moving objects. The traditional centralized methods for range monitoring queries assume that moving objects periodically send location-updates to the server. However, when the number of moving objects becomes increasingly large, such an assumption may no longer be acceptable because the amount of location-update stream becomes enormous. Recently, some distributed methods have been proposed, where moving objects utilize their available computational capabilities for sending location-updates to the server only when necessary. Unfortunately, the existing distributed methods only deal with the objects moving in Euclidean space, and thus they cannot be extended to processing range monitoring queries over the objects moving along the road network. In this paper, we propose the distributed method for processing range monitoring queries in the road network. To utilize the computational capabilities of moving objects, we introduce the concept of vicinity region. A vicinity region, assigned to each moving object o, makes o monitor whether or not it should be included in the results of nearby queries. The proposed method includes (i) a new spatial index structure, called the Segment-based Space Partitioning tree (SSP-tree) whose role is to efficiently search the appropriate vicinity regions for moving objects based on their heterogeneous computational capabilities and (ii) the details of the communication strategy between the server and moving objects, which significantly reduce the wireless communication cost as well as the server computation cost. Through simulations, we verify the effectiveness for processing range monitoring queries over a large number of moving objects (up to 100,000) in the road network (modeled as an undirected graph).

show abstract

Section: Related Workmentioning

confidence: 99%

The SSP-Tree: A Method for Distributed Processing of Range Monitoring Queries in Road Networks

Jung

Kim

2017

IJGI

View full text Add to dashboard Cite

show abstract

“…On the other hand, several KNN graph construction approaches were based on a research index. In this direction, Zhong et al [26] make use of a balanced search tree index to address the KNN search problem for a specific context in which the objects are locations in a road network. Paredes et al [23] proposed a KNN graph construction algorithm for general metric space.…”

Section: Related Workmentioning

confidence: 99%

Nearest Neighbors Graph Construction: Peer Sampling to the Rescue

2016

View full text Add to dashboard Cite

Abstract. In this paper, we propose an efficient KNN service, called KPS (KNN-Peer-Sampling). The KPS service can be used in various contexts e.g. recommendation systems, information retrieval and data mining. KPS borrows concepts from P2P gossip-based clustering protocols to provide a localized and efficient KNN computation in large-scale systems. KPS is a sampling-based iterative approach, combining randomness, to provide serendipity and avoid local minimum, and clustering, to ensure fast convergence. We compare KPS against the state of the art KNN centralized computation algorithm NNDescent, on multiple datasets. The experiments confirm the efficiency of KPS over NNDescent: KPS improves significantly on the computational cost while converging quickly to a close to optimal KNN graph. For instance, the cost, expressed in number of pairwise similarity computations, is reduced by ≈ 23% and ≈ 49% to construct high quality KNN graphs for Jester and MovieLens datasets, respectively. In addition, the randomized nature of KPS ensures eventual convergence, not always achieved with NNDescent.

show abstract

“…Given a query point q on a road network G, a kNN query finds k nearest neighbors (kNN) to q [1,2], or on the contrary, finds one nearest neighbor v for k query points, such that the sum of their distances to v is minimal [3][4][5][6], which is a hot research issue during the past years [1][2][3][4][5][6][7][8][9][10][11][12][13], due to its numerous applications in practice. For example, a tourist may want to search for k nearest hotels while walking in a city, a driver may want to find out k nearest gas stations during driving.…”

Section: Introductionmentioning

confidence: 99%