Recently, researchers start to explore the area of moving objects with multiple transportation modes due to novel applications on detecting transportation modes and advanced trip plannings or recommendations. Some work (e.g., Geolife Project in Microsoft) tries to detect outdoor transportation modes from raw GPS data in order to capture important characteristics and features of mobile users, as these pieces of information can help fully understand the mobility. From the viewpoint of the database community, the complete trip of humans should be managed by a database system, enabling novel queries associated with transportation modes and underlying environments. In spite of massive research work having been conducted on the area of moving objects databases, previous works mainly deal with one environment such as free space or road network. Evidently, a person's trip can cover different environments such as road network, bus network and indoor.
We investigate the following fundamental problems of managing and querying generic moving objects in a database system: (1) data model and data management; (2) data generator and query algorithms; (3) benchmark. A short description of each topic is listed below and the corresponding materials are provided.
Ph.d Thesis Jianqiu Xu (supervised by Prof. Güting), Moving Objects with Multiple Transportation Modes, July, 2012, with the score "magna cum laude" (from Latin, meaning with great honor).
The current data models for moving objects focus on the location representation in a single environment and are not able to represent moving objects in different environments. The major difficulty is to have a framework which applies to all cases as real world environments have different features such as free and constraint movement, 2D and 3D. We propose a method that is able to represent the location of moving objects in different environments including road network, region-based outdoor, bus network, metro network and indoor.
[1] J. Xu and R.H. Güting. A Generic Data Model for Moving Objects, Geoinformatica, 2012.
[2] J. Xu. Research on Moving Objects with Multimodal Transportation Modes, ACM SIGMOD/POSD IDAR, 2011.
Due to the difficulty of getting a large amount of real data including both moving objects and infrastructure data such as roads, bus routes and buildings, a tool is developed to create infrastructure data based on roads and public floor plans. There are a lot of public resources for such kinds of input data. To create generic moving objects, a navigation algorithm through all available environments is developed to find a path like from home to the office room. The complete path located in several environments is used to produce a moving object with different transportation modes.
[3] J. Xu and R.H. Güting. MWGen: A Mini World Generator, 13th International Conference on Mobile Data Management (IEEE MDM), pages 258-267, 2012. ppt
[4] J. Xu and R.H. Güting. Infrastructures for Research on Multimodal Moving Objects, 12th International Conference on Mobile Data Management (IEEE MDM), Demo, pages 329-332, 2011.
To evaluate the performance of a database system managing moving objects with transportation modes, we propose a benchmark that consists of a comprehensive set of datasets and a wide range of well-defined queries. Benchmark data are created in a realistic scenario in order to model human movement in practice. Optimization techniques are developed to reduce the query cost.
[5] J. Xu and R.H. Güting. GMOBench: A Benchmark for Generic Moving Objects, short paper, to appear in ACM SIGSPATIAL, 2012.
[6] J. Xu and R.H. Güting. Manage and Query Generic Moving Objects in SECONDO, Demo, PVLDB 5(12):2002-2005, 2012. poster
Given a polygon with obstacles and a query location, VPS (visible points searching) returns all obstacle vertices that are visible to the query location, i.e., the line between the query location and the obstacle vertex does not cross any obstacle. This is a fundamental problem in computational geometry and geographical databases in obstructed space. We propose a novel and fast algorithm that can find the results in O(N) where N is the total number of vertices, improving the existing method O(N+NlogN). This algorithm is used to find the shortest path for pedestrians where the whole walking area inside a city is modeled as a large polygon with many obstacles (buildings and streets without crossings).
[7] J. Xu and R.H. Güting. Visible Points Searching in Large Obstructed Space , Fernuniversität in Hagen, Technical Report, 2012.
Experimental Materials: (1) Datasets; (2) VPS-Scripts ; (3) Readme
Jianqiu Xu, e-mail: xjq_cs@yahoo.com.cn
Ralf Hartmut Güting, e-mail: rhg@FernUni-Hagen.de