Wireless Web Services using Mobile Agents and Ontologies

1. Wireless Web Services using Mobile Agents and Ontologies V. Baousis, E. Zavitsanos, V. Spiliopoulos, S. Hadjiefthymiades, L. Merakos, G. Veronis Pervasive Computing Research Group Communication Networks Laboratory Dept. of Informatics & Telecommunications, University of Athens, Greece. ICPS 2006 Lyon, France, June 26 – 29, 2006

2. Outline • Introduction • Platform overview • Performance evaluation • Conclusions -Future work

3. Introduction • We propose the integration of Mobile Agents (MA) and Semantically enriched Web Services (SWS) for wireless users who access SWS in the fixed network. • MA transport themselves from one system to another and invoke SWS while being in the same host or network with the SWS. • The system uses an enhanced WS registry that provides semantic matching to incoming service queries and the published SWS descriptions.

4. Platform overview • Service scenario • A user accesses the system and places service requests specifying some criteria. • Creation of a MA that migrates to the registry to find the WS that best meets the user requirements. • Service registry allows for a capability search to be performed, since it is enriched with semantic information. • The MA, after acquiring the WS listing and technical details, migrates to service provider (s), invokes the WS and collects the results • MA delivers the results to the user. • The MA route may vary, depending on the user’s preferences and the network topology. • WS parallel or serial execution by MA • WS execution locally or remotely • USR :User Service Requestor • MA :Mobile Agent • RSA :Registry Stationary Agent • SWSR :Semantic Web Services Registry • WSP : Web Service Provider • PSA : Provider Stationary Agent

5. Mobile Agent (MA) • User representative in the fixed network • Moves, finds, executes SWS and delivers results to the user. • May spawn clones to execute the selected WS in parallel. • Important components are: • The data state contains the information carried by the MA during migrations. • The policies specifies the autonomous behaviour of the MA (migration, cloning). • The matching engine is responsible for post-processing the service registry query results.

6. MA- Policy Management component • Communication Service enables the MA to interact with other network entities. • Monitoring Service filters incoming system messages. • Event Service handles events concerning policy changes. • Trigger Service updates the policy repository when a policy change occurs. • Specification service is responsible for fulfilling this task.

7. SW Services Registry (SWSR) • Consists of the RSA, the matchmaking tool and the UDDI registry. • The Matchmaker enhances the UDDI by adding capability-based discovery and in combination with Racer, processes the OWL ontologies. • Semantic information in SWS descriptions are passed to the OWL-S matchmaking engine. • The Matchmaker maps the OWL-S service description to UDDI.

8. Registry Stationary Agent (RSA) • Acts as a broker between the MA and the service registry. • Implements part of the registry’s functionality and serves MA’s requests. • MA does not have to be aware of the registry implementation details.

9. Provider Stationary Agent (PSA) • Resides in the host offering a SWS. • Wraps the SWS functionality and delivers service results to MA • Communicates with the SWS through SOAP. • MA communicates with PSA with RMI. • PSA existence is indicated in the WS description • MA need not be SOAP fluent • PSA methods are multi-threaded

10. Web Service Provider (WSP) • Provides the SWS • SWS expressed in WSDL and OWL-S. • WSDL (Service grounding) • OWL-S (Service profile and model). • SWS can expose a PSA to act as delegate and interact with the user’s MA

11. Performance Evaluation We have developed and evaluated the following system variants: • A WS system implemented with the “Conventional WS Business Model” (WSBM). • Our MA framework with stationary agents in Service registry and Service providers - (WITH PSA.) • Our MA framework without stationary agents in Service registry and Service providers - (NO PSA.) • A hybrid system, where some Service Providers accommodate a Stationary agent, while others do not (Hybrid.) Terms RIT : Registry Interaction Time MSPTi: Migration to the ith Service Provider Time ITSPi : Interaction Time with the ith Service Provider

12. Performance Evaluation test bed

13. Migration Time • “With PSA” system exhibits constantly less migration time from the hybrid approach and the latter exhibits constantly less from the “No PSA”. • “With PSA” system the MA agent does not have to be SOAP fluent i.e., it does not have to carry extra code in order to support such communication.

14. Interaction Time • Better performance sequence: WSBM, “With PSA”, “Hybrid” and “No PSA” system. The same performance sequence is observed when CL is enabled (lines 5,7,6). • Systems with PSA communicate faster than systems communicating with SOAP. • The better inter-agent communication is attributed to the Mobile Agent Platform used, where agents communicate with synchronous inter-agent message passing.

15. Total Service Time (TST) • Besides WSBM, the system with the lowest service time is that having PSA and MA cloning enabled. • MA cloning increases the Interaction Timebetween the MA and the WS but, eventually, entails considerable improvement to the system (due to WS parallel execution).

16. Conclusions-Future work • A framework for wireless access to semantically enriched WS using MA. • The main advantages of the system are : • Users invoke a set of services with only one interaction with the fixed network • User off-line operation, • Better resource utilization • MA dynamic behaviour improves system robustness and fault tolerance, • New services, agents, users and service registries can be easily integrated to the framework thus providing an expandable, open system. Future work • Study of agent mobility for efficient composition of SWS invocation (implement routing algorithms and considering network status and topology). • Integration with SNMP agents for network performance monitoring.

17. Thank you for your attention… Questions? http://p-comp.di.uoa.gr, http://cnl.di.uoa.gr ICPS 2006 Lyon, France, June 26 – 29, 2006