Distributed trajectory similarity search

Abstract

Mobile and sensing devices have already become ubiquitous. They have made tracking moving objects an easy task. As a result, mobile applications like Uber and many IoT projects have generated massive amounts of trajectory data that can no longer be processed by a single machine efficiently. Among the typical query operations over trajectories, similarity search is a common yet expensive operator in querying trajectory data. It is useful for applications in different domains such as traffic and transportation optimizations, weather forecast and modeling, and sports analytics. It is also a fundamental operator for many important mining operations such as clustering and classification of trajectories. In this paper, we propose a distributed query framework to process trajectory similarity search over a large set of trajectories. We have implemented the proposed framework in Spark, a popular distributed data processing engine, by carefully considering different design choices. Our query framework supports both the Hausdorff distance the Fréchet distance. Extensive experiments have demonstrated the excellent scalability and query efficiency achieved by our design, compared to other methods and design alternatives.

The maximum trajectory coverage query in spatial databases

“…They also support distributed computation by using an abstraction called Tra-jDataset, which is compatible with MapReduce and RDDs. These studies [36,35,10] are orthogonal to ours. Interestingly, our proposed TQ-tree based index can be easily adapted for distributed frameworks where the first level TQ-tree as a global index and the second level TQ-trees as local indexes can be used.…”

The maximum trajectory coverage query in spatial databases

“…They also support distributed computation by using an abstraction called Tra-jDataset, which is compatible with MapReduce and RDDs. These studies [36,35,10] are orthogonal to ours. Interestingly, our proposed TQ-tree based index can be easily adapted for distributed frameworks where the first level TQ-tree as a global index and the second level TQ-trees as local indexes can be used.…”

Querying Similar Multi-Dimensional Time Series with a Spatial Database

“…However, such information is often discarded to find trajectories with similar spatial characteristics. In this regard, the similarity between spatial trajectories is usually measured by either the warp-based distance (including DTW, LCSS, and EDR distances) [77,78], or/and the shape-based distance (including Hausdorff and Fréchet distances) [79], which are consistent with our proposed method. Additional distance measures for trajectory data include the integral of the distance from points on one trajectory to another trajectory (such as the one-way distance, and the symmetrized segment-path distance [80]), and the overlay of two trajectories in the 2D plane (such as the locality in-between polylines distance [81]), which might be also applicable to our analytical framework.…”

“…A similar trajectory search is often implemented as a dynamic programming process, with a computational complexity of O(mn). Similar to time series search, the pruning and local filtering strategies are often applied to boost the speed of similar trajectory search [79]. Moreover, specific storage and indexing methods for trajectory data have also been proposed, such as the segment-based partition and indexing [83], and the pivot-based metric indexing [84].…”

Simple Distances for Trajectories via Landmarks

“…And because they have a familiar Euclidean structure, we can immediately invoke favorite algorithms in this space, from Lloyds for k-means clustering, linear and kernel SVM, and highlyengineered approximate nearest neighbor libraries. In comparison to recent trajectory similarity search systems [19,23], we show using d Q is much simpler and several orders of magnitude faster.…”

The maximum trajectory coverage query in spatial databases

“…They also support distributed computation by using an abstraction called Tra-jDataset, which is compatible with MapReduce and RDDs. These studies [36,35,10] are orthogonal to ours. Interestingly, our proposed TQ-tree based index can be easily adapted for distributed frameworks where the first level TQ-tree as a global index and the second level TQ-trees as local indexes can be used.…”

Querying Similar Multi-Dimensional Time Series with a Spatial Database

“…However, such information is often discarded to find trajectories with similar spatial characteristics. In this regard, the similarity between spatial trajectories is usually measured by either the warp-based distance (including DTW, LCSS, and EDR distances) [77,78], or/and the shape-based distance (including Hausdorff and Fréchet distances) [79], which are consistent with our proposed method. Additional distance measures for trajectory data include the integral of the distance from points on one trajectory to another trajectory (such as the one-way distance, and the symmetrized segment-path distance [80]), and the overlay of two trajectories in the 2D plane (such as the locality in-between polylines distance [81]), which might be also applicable to our analytical framework.…”

“…A similar trajectory search is often implemented as a dynamic programming process, with a computational complexity of O(mn). Similar to time series search, the pruning and local filtering strategies are often applied to boost the speed of similar trajectory search [79]. Moreover, specific storage and indexing methods for trajectory data have also been proposed, such as the segment-based partition and indexing [83], and the pivot-based metric indexing [84].…”

Simple Distances for Trajectories via Landmarks

“…And because they have a familiar Euclidean structure, we can immediately invoke favorite algorithms in this space, from Lloyds for k-means clustering, linear and kernel SVM, and highlyengineered approximate nearest neighbor libraries. In comparison to recent trajectory similarity search systems [19,23], we show using d Q is much simpler and several orders of magnitude faster.…”

The maximum trajectory coverage query in spatial databases

“…They also support distributed computation by using an abstraction called Tra-jDataset, which is compatible with MapReduce and RDDs. These studies [36,35,10] are orthogonal to ours. Interestingly, our proposed TQ-tree based index can be easily adapted for distributed frameworks where the first level TQ-tree as a global index and the second level TQ-trees as local indexes can be used.…”

Querying Similar Multi-Dimensional Time Series with a Spatial Database

“…However, such information is often discarded to find trajectories with similar spatial characteristics. In this regard, the similarity between spatial trajectories is usually measured by either the warp-based distance (including DTW, LCSS, and EDR distances) [77,78], or/and the shape-based distance (including Hausdorff and Fréchet distances) [79], which are consistent with our proposed method. Additional distance measures for trajectory data include the integral of the distance from points on one trajectory to another trajectory (such as the one-way distance, and the symmetrized segment-path distance [80]), and the overlay of two trajectories in the 2D plane (such as the locality in-between polylines distance [81]), which might be also applicable to our analytical framework.…”

“…A similar trajectory search is often implemented as a dynamic programming process, with a computational complexity of O(mn). Similar to time series search, the pruning and local filtering strategies are often applied to boost the speed of similar trajectory search [79]. Moreover, specific storage and indexing methods for trajectory data have also been proposed, such as the segment-based partition and indexing [83], and the pivot-based metric indexing [84].…”

Simple Distances for Trajectories via Landmarks

“…And because they have a familiar Euclidean structure, we can immediately invoke favorite algorithms in this space, from Lloyds for k-means clustering, linear and kernel SVM, and highlyengineered approximate nearest neighbor libraries. In comparison to recent trajectory similarity search systems [19,23], we show using d Q is much simpler and several orders of magnitude faster.…”