Research On Data Acquisition And Generation For Indoor Space

With the development of the Internet of Things and indoor positioning technologies, indoor moving-object data management as well as indoor location-based services (LBS) has become a hot research topic. Although there are a number of studies on algorithms, systems, and trajectory data management towards outdoor spaces, these previous work are not suitable for indoor environments because of the differences between indoor and outdoor spaces including space elements, positioning methods, etc. Therefore, it is important to study new theories and methods for indoor spaces.In this paper, we focus on a fundamental issue in indoor moving-object management, namely indoor data acquisition and generation. In outdoor spaces, people can obtain digital maps, such as Google Maps and city road-network maps, and GPS-based trajectories. On the contrast, there are few indoor maps and indoor moving-object trajectories, which becomes an obstacle of the research on indoor moving-object databases.This paper aims to provide possible solutions to the above problems in terms of two techniques, which are construction of indoor-space maps and generation of indoor moving-object trajectories. The main contributions of this paper can be summarized as follows:(1) We study the issue of constructing indoor-space maps, and present a database approach to automatically extract indoor spatial objects and generate indoor maps from CAD (Computer-Aided Design) models. We also design and implement a Web-based prototype system to demonstrate the feasibility and effectiveness of our proposal. CAD files are usually available, which contain the information of indoor environments. Therefore, we propose to obtain the geometry and topology of indoor entities to construct indoor-space maps. Firstly, we extract line set from CAD files, and cluster lines based on near-line. Secondly, we merge lines in the same cluster into a single line. Thirdly we grow each line until generating a polygon, which we take as indoor cell candidates. At last, we use post-processing rules to clean redundant and wrong-detected ones. With this mechanism, our methods are more robust than existing work for CAD files. The experimental results demonstrate the effectiveness of our algorithom. In addition, we also develop a Web-based interactive interface to delete wrong-detected and add new indoor cells. Finally, we take the CAD model of our experiment building as example. With our method, indoor spatial objects in the CAD model are automatically extracted and transformed into an indoor moving-object database. And the Web-based prototype system provides the demonstration of evaluating indoor spatial queries, like indoor navigation, indoor hot spot query, etc.(2) We investigate the generation of indoor moving-object trajectories, and propose a semantic-rule-based method. As it is complex, unscalable, and costly to deploy positioning devices, such as RFID and Bluetooth devices, to get indoor moving-object trajectories, researchers proposed to develop simulation tools to generate indoor trajectory data. Differing from previous simulation approaches, we propose to use semantic rules to generate trajectory data. Particularly, we first classify indoor moving objects into three types, i.e., regular objects, interested objects, and random objects, and then define the correlations between indoor locations and objects. Regular objects are likely to visit their primary locations, secondary locations, and service locations while interested objects are likely to visit their interesting locations. Random objects visit indoor locations randomly. Based on these rules, we design and implement a data generation tool named IndoorSTG. Users can manually draw the structure of an indoor environment using this tool. It also supports importing space structures from CAD models. In addition, we can also label indoor cells with semantics. Consequently, IndoorSTG can generate trajectories with semantic labels, which provides data support to trajectory analysis and applications in indoor spaces, such as trajectory similarity search and personalized recommendation.