Istanbul 2004 
  
  
  
r data 
e increase in 
delity of the 
/S the number 
ee stages of 
N p 
km2 
Areas of 
2 
A, a, polygons 
A 
2 2 0 
areas at three 
e: 
HA tetes» 
a te TT 
M 
j 
1 
] 
1 
3 
3 
1 
1 
{ 
| 
| 
9 
20 21 22 23 24 
The ID of polygon 
transmission 
olygons in the 
and Level, 3), 
t vary greatly. 
the numbers of 
? are 506, 438, 
edure of the 
the polygons. 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
Figure 9 illustrates the area a sample of 24 polygons at different 
transmission stages. It is clear that there is not great variation in 
the areas of these polygons. 
Response time is a key criterion to evaluate server performance. 
The response time is defined here as the time span from the user 
submitting a query to the client side until a result is visualised 
by the client. It can be represented as 
T (response) = Ty (queryfromdatabase) + T; (transmission) 
* T; (rendering) 
To(queryfromdatabse) is the time span from submitting SQL 
statements to getting the result; 7,(transmission) is the data 
transmission time from the server to the client: and 7, 
(rendering) is the time cost of rendering the results on the client 
side. T(response) is the time cost without simplification 
operations. Suppose the simplification operation is performed 
on the server side, the response time 7 (response) will be 
represented as 
T (response) = Ty (queryfromdatabase) + T (simplification) 
* T, (transmission) * T, (rendering) 
1, (simplification) is the running time of the algorithm to 
sunplify the data set to a certain level. The algorithm has been 
used to simplify several data sets to different levels. Moreover, 
the running time of the algorithm and the response time are 
compared to evaluate performance, and the response time 
with/without simplification operations is also compared. The 
experiments were undertaken on an intranet with a Tomcat 
server running under Linux. The client side was a Java applet- 
based interface. 
Resolution(Kb) 
1200 
1000 
800 
600 
400 
200 
3.345 4.235 4.432 4.803 5.156 5.931 Time(Seconds) 
Figure 10. Running time of simplification at different levels 
í gopResolution(Kb) 
1600 
600 
400 
200 
1953 ^ 2142 2437 2579 2621. 2773 29.4 Vime(Seonds) 
Figure 11. The response time at different simplification levels 
29 
Figures 10 and Figure 11 show the running time of the 
algorithm and the response time of different levels of data. A 
key advantage of the progressive transmission algorithm is that 
it is able to reduce the response time of the server greatly. 
Figure 10 shows that it takes the algorithm about 3.3 seconds to 
simplify the data set from 1,800Kb to 1,100Kb, and about 6.1 
seconds to simplify the data set to 710Kb. It demonstrates thc 
algorithm performs well in simplifying data sets which is 
important especially when the data set is complex and network 
bandwidth is narrow. Figure 11 illustrates the response time at 
different data levels. The response times of level 1 and level 2 
are about 19.5 seconds and 21.42 seconds respectively. 
Compared with the response time of the original, unsimplified 
data — 29.4 seconds, the algorithm improves the response time. 
The data volume to be transmitted is reduced after 
simplification and, as less data volume needs to be transmitted, 
the response time is improved. 
6. CONCLUSIONS 
With the wide popularity of web-based applications and 
services, spatial data delivery over the internet is becoming the 
‘bottleneck’ to rapid data transmission. Progressive 
transmission of vector map data is a promising solution to 
overcome this ‘bottleneck’. The precondition of progressive 
transmission extraction of a coarse level data set from the 
original data set. Extracting multiple-level representations of 
spatial entities on-line is an efficient solution to implement 
progressive vector data transmission over the internet. 
In this paper a methodology for delivering large volumes of 
vector data via progressive transmission is presented and 
implemented. The method represents the original data as 
coarser level data plus a series of removed vertices through 
vertex removal. Moreover rules for vertex removal, that 
maintain the shape fidelity of vector objects and retain correct 
topology, are proposed in this paper. Secondly, the 
corresponding simplification algorithm and recovery algorithm, 
respectively, have been developed in Java, and a client-server 
system architecture has been designed to test the performance 
of the progressive transmission algorithm. Thirdly, the 
experimental results demonstrate that the methodology can 
simplify original data sets to a low level of detail efficiently and 
maintain the shapes of geometry objects as well. Because of 
ability to simplify data the methodology decreases the 
transmission time considerably. The experimental results 
demonstrate that the methodology can provide a viable and 
efficient solution for the progressive transmission of vector map 
data. 
Further work is required to investigate the applicability of the 
method to large datasets with complex geometries and many 
layers. 
ACKNOWLEDGEMENTS 
This research is supported by the EU-IST Project No. IST- 
2001-35047 (SPIRIT). 
REFERENCES 
Burrough, P., A. and McDonnell, R., 1998. Principles of 
Geographical Information Systems. Oxford University Press.