c. IF details matched and parsing result indicate 
valid .wsdl, then put this web service in a 
discovered list. 
3) Sends the discovered list to the decisionMakerAgent 
3.2.6 decisionMakerAgent 
Input: Discovered web service list. 
Output: Ranked web service list. 
Procedure: 
1) decisionMakerAgent gets the discovered web service 
list from the discoveryAgent. 
2) Loop (For each and every service listed in discovered 
list) 
a. Fetch the QoS (availability and throughput) 
results of service. 
b. Calculate input membership levels. 
c. Find the fuzzy rules which can be applied to the 
fuzzy inputs. Evaluate the rules by using center of 
gravity method (COG). These rules are as per 
Table 1. 
  
  
  
  
  
  
  
  
  
  
d. Defuzzify the fuzzy output to system output. 
Rule | Availability | Throughput QoS 
1 Bad Bad Bad 
2 Bad Good Bad 
3 Bad Best Good 
4 Good Bad Bad 
5 Good Good Good 
6 Good Best Best 
7 Best Bad Good 
8 Best Good Best 
9 Best Best Best 
  
  
  
  
  
  
Table 1. Fuzzy Rules 
3) Sorts the list according to final numeric value, which 
gets from defuzzification step and rank it accordingly. 
4) Sends the ranked web service list to the service 
requestor. 
4. SIMULATION OF GEO WEATHER WEB SERVICE 
For validating the developed system, we have created an 
environment for simulating the geo-weather web services as 
described below: 
1. Web Service Creation: Three geo-weather web services 
were developed that provide weather information when a 
corresponding city and country are provided as input. 
These SOAP based web services were deployed over Web 
Server. 
2. Registration to UDDI: An interface to UDDI Server is 
developed. The developed web services were registered to 
UDDI Server. 
3. WSDL Parsing- The wsdl parsing of the sample web 
services was done and the results were stored. 
4. Discovery & Selection of Web Service- Based on 
keyword entered by the service requester the list of best 
fit web services was generated. For our case, we have 
entered ‘weather’ as keyword and the three sample web 
services were fetched as shown in Figure 3. 
On selection of wsdl results option, two web services 
were listed that filters the web service that doesn’t have 
syntactically correct wsdl. The fuzzy details related to 
each web service can be displayed. 
5. Ranking- Based on the fuzzy calculations as described in 
section 3.2.6, the web services were ranked and displayed. 
For simulation, we have considered two input QoS 
parameters namely (Availability and Throughput). 
The input fuzzy set membership function (Bad, Good, 
Best) is a triangle form. The range of the input is given 
below: 
Bad: = {(0, 1) (60, 1) (75, 0)}; 
Good: = {(60, 0) (75, 1) (90, 0)}; 
Best: = {(75, 0) (100, 1)}; 
For output, the same range has been applied. For 
defuzzification, we have used “Center of Gravity” method. 
The overall QoS results, depicted in Figure 4 (a-c) and 
Service requester feedbacks (in context to good, bad) were 
listed. 
6. Web Service Feedback- Service requester can provide 
feedback for the respective web services. 
7. Binding- We have developed a web application to bind 
the highest ranked weather web service for fetching the 
weather information based on city and country. 
5. CONCLUSION & FUTURE SCOPE 
We have developed an automated end-to-end solution to be 
deployed at broker side that provides a common framework for 
service registration, wsdl validation, QoS measurement, 
discovery and selecting the highest ranked web service using 
multi-agent system. The solution can be used for intranet as 
well as internet environment. 
Currently, we have considered only two QoS parameters 
namely availability and throughput for implementation. In 
future the other domain specific Geospatial QoS parameters 
namely: accuracy of geospatial data, resolution, completeness, 
and data types also be considered. 
References from Books: 
[1] Web services: Principles and technology - by Papazoglou, 
Michael P.; Harlow, Pearson Education. 
232 
2 
Sa 
tm 
* 3 
Cl OS