Erik G. Hoel

Advances in Spatial and Temporal Databases

Spatiotemporal reachability queries arise naturally when determining how diseases, information, physical items can propagate through a collection of moving objects; such queries are significant for many important domains like epidemiology, public health, security monitoring, surveillance, and social networks. While traditional reachability queries have been studied in graphs extensively, what makes spatiotemporal reachability queries different and challenging is that the associated graph is dynamic and space-time dependent. As the spatiotemporal dataset becomes very large over time, a solution needs to be I/O-efficient. Previous work assumes an ‘instant exchange’ scenario (where information can be instantly transferred and retransmitted between objects), which may not be the case in many real world applications. In this paper we propose the RICC (Reachability Index Construction by Contraction) approach for processing spatiotemporal reachability queries without the instant exchange assumption. We tested our algorithm on two types of realistic datasets using queries of various temporal lengths and different types (with single and multiple sources and targets). The results of our experiments show that RICC can be efficiently used for answering a wide range of spatiotemporal reachability queries on disk-resident datasets.

A qualitative comparison study of data structures for large line segment databases

A qualitative comparative study is performed of the performance of three popular spatial indexing methods - the R-tree, R+-tree, and the PMR quadtree-in the context of processing spatial queries in large line segment databases. The data is drawn from the TIGER/Line files used by the Bureau of the Census to deal with the road networks in the US. The goal is not to find the best data structure as this is not generally possible. Instead, their comparability is demonstrated and an indication is given as to when and why their performance differs. Tests are conducted with a number of large datasets and performance is tabulated in terms of the complexity of the disk activity in building them, their storage requirements, and the complexity of the disk activity for a number of tasks that include point and window queries, as well as finding the nearest line segment to a given point and an enclosing polygon.

Spatial, temporal and spatio-temporal databases - hot issues and directions for phd research

Spatial and temporal database systems, both in theory and in practice, have developed dramatically over the past two decades to the point where usable commercial systems, underpinned by a robust theoretical foundation, are now starting to appear. While much remains to be done, topics for research must be chosen carefully to avoid embarking on impractical or unprofitable areas. This is particularly true for doctoral research where the candidate must build a tangible contribution in a relatively short time.The panel session at the Eighth International Symposium on Spatial and Temporal Databases (SSTD 2003) held on Santorini Island, Greece [7] in July 2003 thus took as its focus the question What to focus on (and what to avoid) in Spatial and Temporal Databases: recommendations for doctoral research. This short paper, authored by the panel members, summarizes these discussions.

SaIL: A Spatial Index Library for Efficient Application Integration

With the proliferation of spatial and spatio-temporal data that are produced everyday by a wide range of applications, Geographic Information Systems (GIS) have to cope with millions of objects with diverse spatial characteristics. Clearly, under these circumstances, substantial performance speed up can be achieved with the use of spatial, spatio-temporal and other multi-dimensional indexing techniques. Due to the increasing research effort on developing new indexing methods, the number of available alternatives is becoming overwhelming, making the task of selecting the most appropriate method for indexing the data according to application needs rather challenging. Therefore, developing a library that can combine a variety of indexing techniques under a common application programming interface can prove to be a valuable tool. In this paper we present SaIL (SpAtial Index Library), an extensible framework that enables easy integration of spatial and spatio-temporal index structures into existing applications. We focus on design issues and elaborate on techniques for making the framework generic enough, so that it can support user defined data types, customizable spatial queries, and a broad range of spatial (and spatio-temporal) index structures, in a way that does not compromise functionality, extensibility and, primarily, ease of use. SaIL is publicly available and has already been successfully utilized for research and commercial applications.

Building a Robust Relational Implementation of Topology

Topologically structured data models often form the core of many users’ spatial databases. Topological structuring is primarily used to ensure data integrity; it describes how spatial objects share geometry. Supporting topology within the context of a relational database imposes additional requirements – the complex topological model must retain integrity across transactional boundaries. This can be a problematic requirement given the complexities associated with implementing safe referential integrity structures in relational databases (e.g., bulk data loading into a topologically structured model) [19, 5]. Common implementation techniques such as allowing dangling pointers (i.e., null foreign keys) complicates the issues for client applications that consume these models. In this paper, we revisit the problem of building a robust and scalable relational implementation of a topologically structured data model. We propose a different approach to representing such models that avoids many of the traditional relational database problems associated with maintaining complex semantic models.

Data-Parallel Spatial Join Algorithms

Efficient data-parallel spatial join algorithms for bucket PMR quadtrees and R-trees, common spatial data structures, are given. The domain consists of planar line segment data (i.e., Bureau of the Census TIGER/Line files). Parallel algorithms for map intersection and a spatial range query are described. The algorithms are implemented using the scan model of parallel computation on the hypercube architecture of the Connection Machine.

Spatial indexing and analytics on Hadoop

Effective processing of extremely large volumes of spatial data has led to many organizations employing distributed processing frameworks. Hadoop is one such open-source framework that is enjoying widespread adoption. In this paper, we detail an approach to indexing and performing key analytics on spatial data that is persisted in HDFS. Our technique differs from other approaches in that it combines spatial indexing, data load balancing, and data clustering in order to optimize performance across the cluster. In addition, our index supports efficient, random-access queries without requiring a MapReduce job; neither a full table scan, nor any MapReduce overhead is incurred when searching. This facilitates large numbers of concurrent query executions. We will also demonstrate how indexing and clustering positively impacts the performance of range and k-NN queries on large real-world datasets. The performance analysis will enable a number of interesting observations to be made on the behavior of spatial indexes and spatial queries in this distributed processing environment.

Data-Parallel R-Tree Algorithms

Data-parallel algorithms for R-trees, a common spatial data structure are presented, in the domain of planar line segment data (e.g., Bureau of the Census TIGER/Line files). Parallel algorithms for both building the data-parallel R-tree, as well as determining the closed polygons formed by the line segments, are described and implemented using the SAM (Scan-And-Monotonic-mapping) model of parallel computation on the hypercube architecture of the Connection Machine.

Maintaining Connectivity in Dynamic Multimodal Network Models

Network data models are frequently used as a mechanism to describe the connectivity between spatial features in many existing and emerging GIS applications (location- based services, transportation design, navigational systems, etc.). Connectivity information is required for solving a wide range of location-based queries like finding the shortest path, service areas discovery, allocation, and distance matrix computation. Nevertheless, real-life networks are dynamic in nature since spatial features can be periodically modified. Such updates may change the connectivity relations with the other features and connectivity must be reestablished. Existing approaches are not suitable for a dynamic environment, since whenever a feature change occurs, the whole network connectivity has to be reconstructed from scratch. In this paper, we propose an efficient algorithm that incrementally maintains connectivity within a dynamic network. Our solution is based on the existing functionality (tables, joins, sorting algorithms) provided by a standard relational DBMS and has been implemented and tested and will be shipped with an upcoming release of the ESRI ArcGIS product.

Versioning of Network Models in a Multiuser Environment

The standard database mechanisms for concurrency control, which include transactions and locking protocols, do not provide the support needed for updating complex geographic data in a multiuser environment. The preferred method to resolve conflicts in GIS systems is to encapsulate the modifications generated by the end users through the use of multiple versions. Multiuser (or versioned) geographic databases allow users to operate as though they have full access to the entire dataset. Instead of relying upon row locking, versioned databases allow multiple users to simultaneously edit the same row. They implement a model for conflict detection and resolution where the first to commit the change wins by default (though clients can manually intervene and select the latter change as the winner). Network models are frequently used as a mechanism to describe the connectivity information between spatial features in many emerging GIS applications. Supporting networks within the context of a versioned database imposes additional requirements --- the complex network model must retain integrity irrespective of the sequence of simultaneous edits by various clients. In this paper, we review our network model and discuss the enhancements necessary to maintaining topological network integrity in this complex environment. Our solution is based on the notion of dirty areas and dirty objects (i.e., regions or elements that contain edits that have not been reflected in the network connectivity index). The dirty areas and objects are identified and marked during editing of the network feature data. They are then subsequently cleaned as a byproduct of the incremental update of the connectivity network.