DEVELOPMENT AND QUALITY CONTROL OF THE SPATIAL DATABASE FOR TELECOM NETWORK MANAGEMENT GIS

Transcription

1 DEVELOPMENT AND QUALITY CONTROL OF THE SPATIAL DATABASE FOR TELECOM NETWORK MANAGEMENT GIS Summary Dragan Stojanovi`, Slobodanka'MorGMevi`-Kajan, Milan Petkovi`, Leonid Stoimenov Faculty of Electronic Engineering, University of Ni^ CG&GIS Lab, Computer Science Department Beogradska 14, Ni^, Yugoslavia In this paper the development process of spatial database for Telecom network management GIS is presented. It has been developed in accordance with methodology and approach proposed at the Computer Graphics and GIS Lab at the University at Ni^. The spatial database design and development have been performed under permanent quality control and estimation of quality criteria and characteristics using visual and statistical sampling methods. The formed spatial database is the basis of the GeoTT, geographic information system developed for the purpose of TT (Telegraph Telephone) traffic department of public PTT Serbia company at Ni^. 1. Introduction In accordance with the great expansion of GIS applications around the world, the conversion of the existing conventional geographic data into digital form is under special attention and consideration. This is more significant knowing the fact that about eighty percent of all data have a spatial component and understanding and maintaining of this component of data brings vast opportunities. The full implementation of specific GIS must be preceded by the geographic data acquisition and conversion that can take a great part of complete system budget and time. Such digital geographic data organized and stored in the form of spatial database is the foundation of specific GIS implementation. The spatial database enables storing, processing querying, analyzing and displaying large amount of information about spatial properties of geographic entities (geometry, topology) along with their non-spatial, thematic attributes in the alphanumeric and multimedia (sound, image, video) form. Many countries world wide are investing in their national projects with the intention to transform conventional geographic data into digital form, i.e. in the form of specific spatial databases. With the advent of computer mapping and geographic information technology the establishment of spatial data quality standards by the government or professional institutions becomes more important [1, 2]. All these activities are prepared and performed in Yugoslavia

2 mainly in public utilities and government institutions [3]. Computer Graphics and GIS Laboratory providing complete solutions in modeling and development of dedicated spatial databases and GIS based on them, for government and public institutions gives a great contribution to this process. The methodology for spatial database development supports the latest techniques and up-to-date information technology and perfectly meets modern trends of geographic data capture, conversion and quality control. The methodology successfully implemented in forming the 018 TT network area spatial database is described further in this paper. This spatial database consists of hybrid, both raster and vector data in addition to multimedia type attributes about all entities representing telephone cable network and cable equipment, along with appropriate (adequate) information about surrounding objects of interest. The quality and completeness of the developed spatial database have been proved during the function of GeoTT, telecom network management GIS for evidencing, monitoring, maintaining and analyzing TT cable lines of 018 network area, for last several months [4, 5]. 2. Spatial database development methodology Computer Graphics and GIS Laboratory has established highly productive, effective and modern approach and methodology for development digital geographic data in the form of specific spatial databases with the intention to satisfy specific requests and demands of our clients, who need digital geographic data and GIS based on them [6]. This methodology is based on the best technology available for geographic data capture and conversion to specific digital format. The spatial database development is supported by a wide variety of software used (covered) in all phases of the conversion process and specific quality control tasks. Dedicated software systems have been developed by CG&GIS team of professionals. The spatial components of digital geographic data consist of two kinds of spatial information: raster and vector. These geographic data completely describe spatial properties of geographic entities of interest including their geometry (shape, size, location) and topology (relationships with other entities). Also geographic entity is described by their non-spatial, thematic attributes describing their quantitative and/or qualitative properties and traits in the form of text, numbers, graphic symbols, sounds, images, video records and so on. In order to fully represent some important aspects of the geographic world, the spatial database must be modeled and developed to include, store and appropriate organize all these data about geographic entities constituting this geographical aspect. Proposed methodology covers all these activities in spatial database developments. The usage of raster geographic data in the form of raster maps is very significant aspect of GIS, because they can be completely used instead of classical paper

3 maps, providing users with all data necessary to spatial reference and locate themselves depending on the surrounding spatial objects existing on the raster map. Such raster maps present a basic layer (base map) for all other geographic information which exist mostly in the vector form. Fig 1. Raster maps at scales 1:1000, 1:5000, 1:25000 in multiple windows The starting point of the phase for creating raster map database is preparing available paper maps at necessary scales. Each paper sheet is scanned all at once or in parts, depending on the scanner and sheets dimensions. Obtained set of files presenting the scanned parts of paper maps represent input for dedicated software systems for creating raster map database. MapEdit software [7] enables filtering operations that eliminate and correct scanning errors, processing, organizing and storing raster data into meaningful units appropriate annotated. Produced raster data is georeferenced on the basis of exact geographical coordinates of control points obtained (supplied) in this process. Input of thematic data about raster maps (map s legend), further postprocessing and database installation are done by MapBase and MapInstall software [6]. Raster map database enables fast retrieval of raster geographic data forming continuous georeferenced raster map at the time of database exploitation. Raster maps at scales 1:1000, 1:5000 and 1:25000 in separate windows are shown on Figure 1. The next phase of this methodology is the production of the vector and raster geographic data comprising spatial components of geographic entities information in the form of points, lines, polylines, polygons, and other graphic structures, along with thematic attributes in appropriate forms (alphanumeric, sound, image, video records). The collection and acquisition of all this relevant

4 and important information are performed using various available sources and different methods of data capturing and conversion: Using all possible information about geometry and topology (spatial relations) and attributes of spatial entities placed on the geographic area of interests, already existing as databases and files by conversion into specific formats, or in a paper form as technical documentation. Digitizing manual digitizing using high resolution, precision digitizing boards along with high performance graphic workstations; semiautomatic and/or automatic vectorizing based on the formed and edited raster map, specific knowledge-based software, with or without operator interaction; on screen (heads-up) digitizing based on formed raster map; GPS (Global Positioning System) surveying and measurements. The developed methodology covers all these forms of spatial data acquisition and conversion. Quality control and quality estimation on specific criteria are important and integral parts of each of spatial database development phases. Generated spatial data is stored as objects in an object-oriented spatial database. This object-oriented spatial database is the basis for the GIS developed around (on top of) the GinisNT - a scaleable, object-oriented GIS framework, specially developed software system [8]. The geographic data from raster map database form continuos base map enabling users to continuously move, pan, zoom, and reference themselves on (over) it. The geographic data presenting spatial domain of interest, stored and organized in an object-oriented spatial database is displayed on the raster layer with their exact geometry, topology and attributes. The values of these attributes that describe some quantitative and/or qualitative characteristics of the spatial entities are displayed next to corresponding geographic entities in form of annotations and symbols, or inside the forms and dialog boxes shown on the user request. Logically connected vector data and annotations form one vector layer. At the same time few different, physically independent vector data layers can be displayed on the basic raster layer (gas, electricity, telecommunications, streets, land use), enabling users to get their own specific views on stored geographic data. These layers can be shown or hidden on the user demand, providing him all necessary information and meeting his requirements while exploring GIS application. 3. The spatial database for Telecom network management GIS Establishment and development of the geographic information system in the PTT Serbia department at Ni^, are needed for the purpose of evidencing, monitoring, analyzing and maintaining TT cable lines of the 018 TT network area. The first activity was to collect, analyze and convert all important and necessary geographic information and to form raster map database of this

5 network area. Paper maps at scales 1:300000, 1:25000, 1:5000 and 1:1000, were the basis for the phase of the continuos raster map development: Geographic map at a scale 1: (4 sheets) covers the area of the Republic of Serbia; Topographic map at a scale 1:25000 (35 sheets) covers the whole 018 TT cable network area; Geographic map at a scale 1:5000 (26 sheets) covers Ni^ and its wide area; Cadastral geographic map at a scale 1:1000 (33 sheets) covers the area of the Nikola Tesla telephone exchange needed for the purpose of pilot project development. Paper maps at scales 1: and 1:25000 were scanned by the 256 colours A3 scanner, while paper maps at scales 1:5000 and 1:1000 were scanned by the black&white A0 scanner all in 100 dpi resolution. After processing the scanned maps and performing all necessary operations described in previous chapter, the raster map database was formed. For the process of generating TT network database it was necessary to collect all information about geometrical, topological and descriptive characteristics of geographic objects that make TT cable network such as: telephone exchanges, cable segments, cable conduits, junctions, junction boxes, manholes, connection boxes, terminal blocks, etc. Technical documentation of the local TT cable network, according to the regulations of PTT Serbia, consists of: General cable plans of the 018 TT network area, Sketch of cable in conduits, Cabling drawing in manholes, Cabling diagrams, Cable scheme, Technical documentation of cable junctions and connection boxes, Network overview, Inventory. The whole technical documentation data existing on the paper in tabular and schema (sketch) form (in the alphanumeric form) has been manually entered by a particular software. This software provides interactive GUI and by using the formed raster map ensures visual verification of input data related to their position on the raster map. The whole amount of data presenting subscribers and faults documentation existing mostly in specific files has been converted into the specific spatial database format. The two-dimensional CAD system, GeoCAD [6] has been used for geometric entities drawing, dimensioning and editing (for example scheme of cables in manholes and conduits)and have been directly entered into spatial database. Data that make the cable scheme has been entered by semiautomatic vectorizing of the scanned plans by a dedicated software. This method is still under development and promises good and accurate results [9]. A part of spatial data that does not exist in technical documentation has been collected and entered using field survey and up-to-date GPS technology. This method of collecting spatial data by the use of GPS is

6 under further development and seems to be very appropriate and attractive in the near future [10]. Some parts of the spatial data have been collected in the process of on-screen (heads-up) digitizing, using PC based workstations with high graphics capabilities, which is known to be the easiest and fastest way of entering data into a GIS, with precision being satisfactory in most cases. Fig 2. TT network map at scale 1:1000 This has been done by the use of the formed raster maps at scales 1:1000 and 1:5000. In the process of TT network area digital map production, manual digitizing has not been used mostly because of its inherently labour-intensive character and the lack of necessary equipment. Multimedia information about important parts of TT network and TT equipment is often necessary for TT network maintenance and monitoring. Such information like photos, audio and video records (for example photos of manholes in natural environment, manual for specific interventions, warnings or critical actions, etc.) in appropriate format have been entered and organized into spatial database Graphic displays of various types of telecom network data in appropriate windows are shown in Figure 2 and 3.

7 Fig 3. Various types of information of telecom network and cable equipment All spatial and non-spatial data has been stored in RDBMS MS Access format, and in some of the formats that make these data accessible by the use of ODBC standard and available ODBC drivers (Oracle). The object-oriented data model and object-oriented environment defined by the GinisNT framework are used at the application level, while the underlying database is actually stored and maintained by the RDBMS. The relational implementation level is transparent for the end user (developer of GIS) by existence of intermediary software components in the system which automatically performs mappings between the object-oriented model and relational one [11]. The application developer is freed from worrying about data storage details and can concentrate on the GIS application that appears to him to be completely object-oriented. The GinisNT system provides integration of both raster and vector forms of digital geographic data, allowing development of hybrid GIS application. 4. Quality control and quality estimation of spatial database Wide acceptance and application of geographic information systems in government institutions, public and private companies, make geographic data quality standards very important. These standards must be established and produced according to the international, national or professional geographic

8 information standards [1, 2]. Such standards must provide data quality characteristics, their estimation criteria, checking methods and acceptance principles of a final spatial database product. That the geographic database product should meet quality requirements the quality criteria must be considered and accepted by both producers and consumers. Firstly, all possible sources of bad geographic data quality must be eliminated. Paper maps must be made with high accuracy and quality, and all data sources that constitute technical documentation must be checked for correctness, completeness and accuracy. Such activities should be done by the customers who deliver and supply all data needed for spatial database development. Parameters which determine quality estimation criteria, in conformance with [12, 13], can be divided into: Geographic accuracy - discretization accuracy - raster map layer correctness, - position accuracy, - displayed vector entity precision - distance precision Attribute accuracy - attribute precision, - attribute completeness, - attribute (uniformity). Logical consistency - geographic consistency - correction of topological relation Graphic s quality - raster map graphic quality - vector data quality - symbol and annotation quality For each phase of the spatial database development process a quality control and quality estimation procedure are defined together with the customer, and applied along with established quality control process. Geographic accuracy is controlled by choosing a pattern of at least 50 checked points with known exact geographic coordinates, for each scale, which are well distinguished and well distributed. Along with it, 50 line or polyline features representing spatial entities or measured distances for each scale were examined, too. By the use of the checked points, position accuracy and raster map accuracy have been measured evidencing deflections of the checked points. All these measured data have been used to calculate mean square error of point, as well as mean square error of line and polyline features by means of geographic accuracy estimation. Attribute accuracy, logical consistency and graphic quality are checked and established according to the specific customers needs, with the intention to meet all their requirements. This was done by the use of the combination of visual methods and the statistic method of population on the 100 attributes pattern. The described quality control and estimation process ensure making decisions either to come back to some of the previous phases of production process where detected errors must be corrected or to validate spatial database quality and put it in use. High quality and accuracy of the spatial database for telecom network management GIS was established and verified on that way.

9 5. Conclusions The approach and methodology for spatial database development proposed by Computer Graphics and GIS Laboratory at the Faculty of Electronic Engineering at Ni^ is presented. The methodology perfectly meets modern trends and achievements of geographic data capture and GIS development in the information world, and is based on the dedicated software systems developed on the basis of the object-oriented and open systems principles and paradigms. It has been successfully implemented in the production and quality control process of the TT network spatial database. The telecom network spatial database is permanently evolving: the new telephone cables are laid, new subscribers telephone numbers are added, new connection and junction boxes are added and old ones are changed. All these changes and additions are evidenced and entered into the spatial database in appropriate ways. From time of its development, the telecom network database comprises various types of geographic information about domain of interest including spatial, alphanumeric and full multimedia data. Further directions of our activities are to extend this telecom network spatial database with full temporal capabilities in order to enable maintenance and analyzing the history states of the telecom network as they evolve and change over time. These would also provide simulation, planing and designing all future telecom network developments, maintaining various alternatives and possibilities and support decision making based on comparing important characteristics of these alternatives. In the course of time this spatial database will become the complete encyclopedia of geographic space that is 018 TT network area, that is in authority of the TT department at NI^. References [1] ISO/TC 211 Geographic information/geomatics [2] H. Tom, GIS Standards - The Time Has Come, GISAsiaPacific, Vol. 2, No. 2, April 1996, pp [3] YUGIS - The First Yugoslav Conference on GIS Technologies, GIS - Situation and Perspectives, Beograd, Yugoslavia, mart (in Serbian) [4] S.?or_evi`-Kajan, S. Mladenovi`, M. To^i`, Geographic Information System for TT Traffic Department at Ni^, YUGIS, Beograd, Yugoslavia, March 1996., pp (in Serbian) [5] S.?or_evi`-Kajan, M. Petkovi`, L. Stoimenov, A. Mitrovi`, S. Mladenovi`, M. To^i`, GIS in Ni^ PTT, GIS/LIS, Budapest, Hungary, Jun 1996.

10 [6] The Metodology, Procedures and Software for Production of Digital Geographic Maps, Technical Report, CG & GIS Lab, Faculty of Electronic Engineering, Ni^, Yugoslavia, January (in Serbian). [7] S.?or_evi`-Kajan, D. Mitrovi`, D. Ranai`, M.Petkovi`, A. Mitrovi`, MapEdit - Software for creating a continous raster map base, Eurocarto XIII, Ispra, Italy, October [8] A. Mitrovi`, D. Mitrovi`, S.?or_evi`-Kajan, M. Petkovi`, A Scaleable Object-Oriented GIS Framework, ISPRS Workshop on New Development in GIS, Milan, Italy, 6-8. March 1996, pp [9] M.Smiljani`, M. Petkovi`, D. Ranai`, D. Stojanovi`, One Algoritm for Automatic Vectorizing Raster Images in GIS, YU INFO, Brezovica, Yugoslavia, 2-5. April (in Serbian) [10] D.Mitrovi`, A. Mitrovi`, D. Ranai`, Global Positioning System as Method for Spatial Data Acquisition, YUGIS, Beograd, Yugoslavia, March 1996., pp (in Serbian) [11] L.Stoimenov, A. Mitrovi`, D. Stojanovi`, Coupling OO GIS Application with Relational Databases, YUGIS, Beograd, Yugoslavia, March 1996, pp (in Serbian) [12] D.Daosheng, Introduction to Data Quality Standard for Digital Topographic Map Product, 17th Intern. Cartographic Conference and 10th General Assembly of the ICA/ACI, 3-9. September 1995, Barcelona, pp [13] G. Hunter, Accuracy and GIS - Why Worry About It?, GIS AsiaPacific, Vol. 2, No. 2, April 1996, pp contact address: Dragan Stojanovi` Computer Graphics & GIS Laboratory Faculty of Electronic Engineering, Computer Science Department University of Ni^ Beogradska 14, Ni^, Yugoslavia Phone: +(381) Fax: +(381) dragans@europa.elfak.ni.ac.yu