## Querying about the past, the present, and the future in spatio-temporal databases (2004)

### Cached

### Download Links

Venue: | In ICDE |

Citations: | 18 - 0 self |

### BibTeX

@INPROCEEDINGS{Sun04queryingabout,

author = {Jimeng Sun and Dimitris Papadias and Yufei Tao and Bin Liu},

title = {Querying about the past, the present, and the future in spatio-temporal databases},

booktitle = {In ICDE},

year = {2004},

pages = {202--213}

}

### OpenURL

### Abstract

Moving objects (e.g., vehicles in road networks) continuously generate large amounts of spatio-temporal information in the form of data streams. Efficient management of such streams is a challenging goal due to the highly dynamic nature of the data and the need for fast, on-line computations. In this paper we present a novel approach for approximate query processing about the present, past, or the future in spatio-temporal databases. In particular, we first propose an incrementally updateable, multi-dimensional histogram for present-time queries. Second, we develop a general architecture for maintaining and querying historical data. Third, we implement a stochastic approach for predicting the results of queries that refer to the future. Finally, we experimentally prove the effectiveness and efficiency of our techniques using a realistic simulation. 1.

### Citations

981 | The r*-tree: an efficient and robust access method for points and rectangles
- Beckmann, Kriegel, et al.
- 1990
(Show Context)
Citation Context ...ucture that takes advantage of spatial conditions to accelerate query processing. 5.3 Implementation using a 3D R-tree Our second implementation is based on a main-memory adaptation of the 3D R*-tree =-=[BKSS90]-=-. In this structure, each intermediate entry contains a 3D box which encloses the extents and lifespans of all the buckets in its sub-tree. Given a HT q (which can also be regarded as a 2D rectangle d... |

318 | Indexing the Positions of Continuously Moving Objects
- Saltenis, Jensen, et al.
- 2000
(Show Context)
Citation Context ...n their present motion vectors (e.g., "based on the current information, find the cars that will be in the city center 10 minutes from now"). The only practical index in this category is the TPR-tree =-=[SJLL00]-=- and its variations [SJ02, TPS03] (also based on R-trees). Other, mostly theoretical, predictive indexes have been proposed in [KGT99] (applicable only to 1D space) and [AAE00] (with good asymptotical... |

210 | Wavelet-based histograms for selectivity estimation
- Matias, Vitter, et al.
- 1998
(Show Context)
Citation Context ...ide a compact data representation. The DCTbased histogram [LKC99] maintains a large number of small buckets using discrete cosine transform, which cannot be incrementally maintained. The histogram of =-=[MVW98]-=- extracts the most descriptive wavelet coefficients. Although it can be incrementally maintained [MVW00], its performance degrades with the number of updates, so that eventually it has to be re-built.... |

205 | Generating network-based moving objects
- Brinkhoff
- 2002
(Show Context)
Citation Context ...e endpoints of each trip are different, the overall distribution after the completion of a trip is either MG or LB. The second dataset, denoted as road, is created by the spatio-temporal generator of =-=[B02]-=-. The input of the generator is the road map of Oldenburg (a city in Germany), containing 6105 nodes and 7035 edges, and the output is a set of point objects moving on this network. Each point is repr... |

205 | New sampling-based summary statistics for improving approximate query answers - Gibbons, Matias - 1998 |

201 | On indexing mobile objects
- Kollios, Gunopulos, et al.
- 1999
(Show Context)
Citation Context ...now"). The only practical index in this category is the TPR-tree [SJLL00] and its variations [SJ02, TPS03] (also based on R-trees). Other, mostly theoretical, predictive indexes have been proposed in =-=[KGT99]-=- (applicable only to 1D space) and [AAE00] (with good asymptotical performance, but inapplicable in practice due to the large hidden constants). † Department of Computer Science Hong Kong University o... |

198 | Selectivity estimation without the attribute value independence assumption
- POOSALA, E
- 1997
(Show Context)
Citation Context ...histogram of [MD88] first partitions along one dimension so that each bucket has the same number (i.e., frequency) of objects. Then, these buckets are partitioned again along another dimension. Mhist =-=[PI97]-=- splits on the most “critical” dimension according to a partitioning metric (e.g., variance). Minskew [APR99] partitions the space into a regular grid. Neighboring cells are then grouped into rectangu... |

183 |
Equi-depth histograms for estimating selectivity factors for multidimensional queries
- Muralikrishna, DeWitt
- 1988
(Show Context)
Citation Context ...plit the data space into buckets, usually based on the assumption that data within a small region are (almost) uniform. The various methods differ on the partition policy. The equi-depth histogram of =-=[MD88]-=- first partitions along one dimension so that each bucket has the same number (i.e., frequency) of objects. Then, these buckets are partitioned again along another dimension. Mhist [PI97] splits on th... |

169 | Indexing moving points
- Agarwal, Arge, et al.
- 2000
(Show Context)
Citation Context ...tegory is the TPR-tree [SJLL00] and its variations [SJ02, TPS03] (also based on R-trees). Other, mostly theoretical, predictive indexes have been proposed in [KGT99] (applicable only to 1D space) and =-=[AAE00]-=- (with good asymptotical performance, but inapplicable in practice due to the large hidden constants). † Department of Computer Science Hong Kong University of Science and Technology Clear Water Bay, ... |

157 | Fast incremental maintenance of approximate histograms
- Gibbons, Matias, et al.
- 2002
(Show Context)
Citation Context ...h; when the query arrives, the histogram is extracted from the sketch. The main drawback of this approach is that the extracting process is expensive and, therefore, not suitable for on-line queries. =-=[GMP02]-=- incrementally maintains a small sample set (backing sample) of underlying data and histograms are then constructed/maintained by the backing sample. However, it requires scanning the entire data to r... |

146 | The TPR*-tree: An optimized spatio-temporal access method for predictive queries - Tao, Papadias, et al. - 2003 |

131 | Self-tuning histograms: Building histograms without looking at data
- Aboulnaga, Chaudhuri
- 1999
(Show Context)
Citation Context ... to the changes that occurred during its re-building. 2.2 Query-adaptive multi-dimensional histograms Query adaptive histograms avoid re-building by using query feedback to refine the buckets. STGrid =-=[AC99]-=-, based on the heuristic that “buckets with higher frequencies contribute more to the estimation error”, splits buckets with high frequencies and merges buckets with low frequencies. Restructuring is ... |

125 |
MV3R-Tree: A SpatioTemporal Access Method for Timestamp and Interval Queries,” VLDB
- Tao, Papadias
- 2001
(Show Context)
Citation Context ...ediction estimate the number of objects that will satisfy some spatial condition during a future interval, based on the current location and velocity information. Choi and Chung [CC02] and Tao et al. =-=[TSP03]-=- present probabilistic models for uniform data, which are then applied to non-uniform distributions using some conventional multi-dimensional histogram. An experimental comparison of several technique... |

113 | Novel approaches in query processing for moving object trajectories - Pfoser, Jensen, et al. - 2000 |

105 | STHoles: a multidimensional workloadaware histogram
- Bruno, Chaudhuri, et al.
(Show Context)
Citation Context ...parameter of the histogram) for entire rows or columns, selected by a split or a merge threshold (also a parameter). Its structure, however, cannot accurately capture arbitrary distributions. STHoles =-=[BGC01]-=- alleviates this problem by allowing nesting of buckets. If a query identifies large frequency variance within a bucket, a "hole", i.e., a new bucket, is created inside the original one. This "drillin... |

90 | Indexing of Moving Objects for Location-Based Services - Saltenis, Jensen - 2002 |

83 | Dynamic maintenance of wavelet-based histograms
- Matias, Vitter, et al.
- 2000
(Show Context)
Citation Context ...ets using discrete cosine transform, which cannot be incrementally maintained. The histogram of [MVW98] extracts the most descriptive wavelet coefficients. Although it can be incrementally maintained =-=[MVW00]-=-, its performance degrades with the number of updates, so that eventually it has to be re-built. Another technique [TGIK02] compresses the data distribution into a multi-dimensional sketch; when the q... |

73 |
Exponential smoothing: The state of the art
- Gardner
- 1985
(Show Context)
Citation Context ...tput of q can be represented as a function of the result of a PT and a set of recent HT queries with the same region qR. Exponential smoothing, a well-known forecasting method in time series analysis =-=[G85]-=-, is based on the same reasoning. According to this method, the estimated result S'1 of the query (qR,1) can be modeled as a time series such that: S'1= αSo+ α(1- α)S-1 + α(1- α) 2 S-2 + …+ α(1- α) n ... |

73 | Approximating multidimensional aggregate range queries over real attributes
- Gunopulos, Kollios, et al.
- 2000
(Show Context)
Citation Context ...hboring cells are then grouped into rectangular, non-overlapping buckets by a greedy algorithm that tries to minimize the spatial-skew (i.e., the variance of the cell density in each bucket). Genhist =-=[GKTD00]-=- allows overlapping buckets. As with Minskew, it starts with a regular grid and creates buckets for the cells with high density. Then it removes a percentage of data from these cells to make the area ... |

60 | Efficient Indexing of Spatiotemporal Objects - Hadjieleftheriou, Kollios, et al. - 2002 |

52 | Multi-dimensional selectivity estimation using compressed histogram information
- Lee, Kim, et al.
- 1999
(Show Context)
Citation Context ...al approximation methods Several multi-dimensional methods embed the data into another domain and summarize the embedded data in order to provide a compact data representation. The DCTbased histogram =-=[LKC99]-=- maintains a large number of small buckets using discrete cosine transform, which cannot be incrementally maintained. The histogram of [MVW98] extracts the most descriptive wavelet coefficients. Altho... |

37 |
The exponentially weighted moving average
- Hunter
- 1986
(Show Context)
Citation Context ...or the current and recent timestamps, since the actual results are known), we chose to fix global values for all queries in order to avoid the overhead of tuning. In particular, it has been suggested =-=[H86]-=- that a = 0.2~0.3 gives accurate results for a variety of problems, while n = 6 represents a good trade-off between accuracy and query cost (the longer we go into the past the higher the cost). Algori... |

33 | Efficient aggregation over objects with extent - Zhang, Tsotras, et al. - 2002 |

32 |
Selectivity estimation for spatio-temporal queries to moving objects
- Choi, Chung
(Show Context)
Citation Context ...for spatio-temporal prediction estimate the number of objects that will satisfy some spatial condition during a future interval, based on the current location and velocity information. Choi and Chung =-=[CC02]-=- and Tao et al. [TSP03] present probabilistic models for uniform data, which are then applied to non-uniform distributions using some conventional multi-dimensional histogram. An experimental comparis... |

27 | Accurate Estimation of the Cost of Spatial Selections
- Aboulnaga, Naughton
(Show Context)
Citation Context ...e cells with high density. Then it removes a percentage of data from these cells to make the area around them smoother. Finally it decreases the resolution and repeats the process recursively. The SQ =-=[AN00]-=- and Euler histograms [SAA02], in addition to locations, take into account objects extents, and are more accurate for nonpoint data. All the above histograms apply to spatial or multidimensional datab... |

27 |
Selectivity estimation for predictive spatio-temporal queries
- Tao, Sun, et al.
(Show Context)
Citation Context ...ediction estimate the number of objects that will satisfy some spatial condition during a future interval, based on the current location and velocity information. Choi and Chung [CC02] and Tao et al. =-=[TSP03]-=- present probabilistic models for uniform data, which are then applied to non-uniform distributions using some conventional multi-dimensional histogram. An experimental comparison of several technique... |

16 | Performance Evaluation of Spatio-Temporal Selectivity Estimation Techniques
- HADJIELEFTHERIOU, KOLLIOS, et al.
- 2003
(Show Context)
Citation Context ...stic models for uniform data, which are then applied to non-uniform distributions using some conventional multi-dimensional histogram. An experimental comparison of several techniques can be found in =-=[HKT03]-=-. All existing methods assume linear movement and that the velocities of all objects are known (the same assumptions that hold for the TPR-tree-based indexes [SJLL00, TPS03]). This restricts their app... |

16 | Applying the golden rule of sampling for query estimation - Wu, Agrawal, et al. - 2001 |

15 | Abbadi. Exploring Spatial Datasets with Histograms
- Sun, Agrawal, et al.
(Show Context)
Citation Context ...hen it removes a percentage of data from these cells to make the area around them smoother. Finally it decreases the resolution and repeats the process recursively. The SQ [AN00] and Euler histograms =-=[SAA02]-=-, in addition to locations, take into account objects extents, and are more accurate for nonpoint data. All the above histograms apply to spatial or multidimensional databases, where the data are (alm... |

14 | SASH: A SelfAdaptive Histogram Set for Dynamically Changing Workloads. VLDB
- Lim, Wang, et al.
- 2003
(Show Context)
Citation Context ...query identifies large frequency variance within a bucket, a "hole", i.e., a new bucket, is created inside the original one. This "drilling" operation replaces the splitting mechanism of STGrid. SASH =-=[LWV03]-=- decomposes the multi-dimensional space into sub-spaces of lower dimensionality. For each subspace a separate histogram is built and maintained using query feedback mechanisms.sQuery-adaptive histogra... |

6 | Using the Golden Rule of Sampling for Query Estimation - Wu, Agrawal, et al. - 2001 |