THE CASCOM ABSTRACT ARCHITECTURE FOR SEMANTIC

SERVICE DISCOVERY AND COORDINATION IN IP2P

ENVIRONMENTS

Cesar Caceres, Alberto Fernandez, Sascha Ossowski, Matteo Vasirani

Artificial Intelligence Group, University Rey Juan Carlos, Madrid, Spain

Keywords: Service coordination, intelligent agents, peer-to-peer computing, mobile computing.

Abstract: Intelligent agent-based peer-to-peer (IP2P) environments provide a means for pervasively providing and

flexibly co-ordinating ubiquitous business application services to the mobile users and workers in the

dynamically changing contexts of open, large-scale, and pervasive settings. In this paper, we present an

abstract architecture for service delivery and coordination in IP2P environments that has been developed

within the CASCOM project. Furthermore, we outline the potential benefits of a role-based interaction

modelling approach for a concrete application of this abstract architecture based on a real-world scenario for

emergency assistance in the healthcare domain.

1 INTRODUCTION

The ever-growing number of services on the WWW

provides enormous business opportunities. In

particular, there is a huge potential for creating

added value through service coordination. For this to

happen, technology must be developed capable of

pervasively providing and flexibly co-ordinating

ubiquitous business application services to mobile

users and workers in the dynamically changing

contexts of open, large-scale and pervasive

application domains.

One step toward the realization of this vision is

the development of an intelligent agent-based peer-

to-peer (IP2P) environment. IP2P environments are

extensions to conventional P2P architectures with

components for mobile and ad hoc computing,

wireless communications, and a broad range of

pervasive devices. A major challenge in IP2P

environments is to guarantee a secure spread of

(personal) service requests across multiple

transmission infrastructures and ensure the

trustworthiness of services that may involve a broad

variety of providers.

In this paper, we present an abstract architecture

for service delivery and coordination in IP2P

environments that has been developed within the

CASCOM project (CASCOM 2005) and discuss the

benefits of role based interaction modelling for its

instantiation to a particular application domain.

This architecture aims at providing support for

business services for mobile workers and users

across mobile and fixed networks. For end users, the

architecture provides easy and seamless access to

semantic Web services anytime, anywhere, and

using any device. For network operators, it aims

towards a vision of seamless service experience

providing better customer satisfaction, which in turn

helps to retain current customer relations as well as

attract new customers. To service providers the

architecture offers an innovative platform for

various mobile business application services.

2 CASCOM ABSTRACT

ARCHITECTURE

The CASCOM abstract architecture is based on a

layered approach that combines innovative network

aspects, such as peer-to-peer and mobile networks,

with modern software technologies, like semantic

web services and intelligent agents (see Figure 1).

The four main components of this architecture act as

a mediator between the applications and the

underlying networks.

205

Caceres C., Fernandez A., Ossowski S. and Vasirani M. (2006).

THE CASCOM ABSTRACT ARCHITECTURE FOR SEMANTIC SERVICE DISCOVERY AND COORDINATION IN IP2P ENVIRONMENTS.

In Proceedings of the Eighth International Conference on Enterprise Information Systems - SAIC, pages 205-208

DOI: 10.5220/0002459402050208

 SciTePress

The Networking Layer is located just over the

networks. Its main functionalities are enabling an

efficient and reliable agent message transport

communication over wireless (and wireline)

communication paths, providing network-related

context information (e.g. QoS of a connection,

network availability, etc.), to the context subsystem,

and providing an agent execution environment for

resource-constrained mobile devices. Requesters

search dynamically for services published by

different providers in a directory service (DS),

which can be centralized or distributed. In the latter

case, a federation of DSs could be built, where each

DS registers itself in other DSs as a service. Thus, a

DS can be found by querying entries in a DS. Such

an approach was used in the Agentcities project

(Constantinescu, 2003), where the directories were

federated accordingly to “application domains”.

The Service Coordination Layer is located

between the Networking Layer and the Application

Layer, and uses the services offered by both the

Context and the Security & Privacy Subsystems.

This layer has two main functionalities, the semantic

service discovery (service discovery + semantic

matchmaking) and the service coordination (service

composition + execution monitoring and

replanning). In the CASCOM abstract architecture,

the semantic service discovery functionality is

realised by two different types of agents: Service

Discovery Agents (SDA) and Service Matchmaking

Agents (SMA). In much the same way, the service

coordination functionality is realised by Service

Composition Agents (SCA) and Service Execution

Agents (SEA). Usually, SDAs receive service

specifications from the users’ Personal Agents (PA)

and acquire relevant contextual information from the

context subsystem. With that information, they use

the service discovery functionality of the networking

layer and the semantic matching functionality of

SMAs, to determine services that fulfil the received

service discovery request. SDAs return then a set of

descriptions/providers and their correspondent

service process model and/or grounding.

SCAs are capable of creating value-added

composite services that match specific service

specifications. Once SCAs receive service

specifications, they contact SDAs to discover

existing services in a given domain constrained to

the current context, and plan a composite

value-added service matching the received service

specification.

SEAs manage the execution of composite

value-added service descriptions generated by

SCAs. Since the received compound service

description relies on simpler services, the execution

will also coordinate the execution of these simpler

services. Whenever necessary, SEAs will use SDAs

to discover appropriate available service providers

for each of the simpler services invoked from the

compound service description.

The Context Subsystem (Lopes, 2005) is

accessed by both the Network Layer and the Service

Coordination Layer. Its main functionality is to

discover, acquire and store useful context

information (e.g. the geographical position or the

user preferences). This kind of information can be

accessed by both layers either by explicitly querying

the context subsystem (“pulling”) or by subscribing

a listener that is in charge of notifying changes and

events occurring in the environment (“pushing”).

The Security and Privacy Subsystem is also

orthogonal to the first two layers. It is responsible

for ensuring security and privacy of information

throughout the different components of the

CASCOM infrastructure. Every node of the network

should keep its data confidentiality, integrity, and

availability (CIA) -- see SecPedia website. But not

only data is a security concern, also the software that

deals with it, especially for network-centric systems,

where the misuse, theft, and unauthorized usage of

computing resources is well studied. The security

and privacy requirements identified are:

identification, authentication, authorisation, single

sign-on, and local and network security. We also

need to guarantee the integrity of transmitted data,

non-repudiability, traceability, privacy, delegation,

and nationalization.

Security & Privacy

subsystem

Service Coordination Layer

Applications

Networks

Security & Privacy

subsystem

Networking

Layer

Service Coordination Layer

Service

Coordination

Layer

Applications

Context

subsystem

Context

subsystem

Networks

Service

SDA

SEA

SCA

SMA

Security & Privacy

subsystem

Service Coordination Layer

Applications

Networks

Security & Privacy

subsystem

Networking

Layer

Service Coordination Layer

Service

Coordination

Layer

Applications

Context

subsystem

Context

subsystem

Context

subsystem

Context

subsystem

Networks

Service

SDA

SEA

SCA

SMA

Figure 1: CASCOM abstract architecture and its elements.

ICEIS 2006 - SOFTWARE AGENTS AND INTERNET COMPUTING

206

3 ROLE-BASED INTERACTION

MODELLING

In order to improve both the efficiency and the

usability of the CASCOM architecture, we claim

that it is necessary to account for the types of

interaction that certain semantic services can be used

in. In this section we outline how we will apply and

extend our role-based and interaction-centric

modelling framework (Serrano et al. 2002) within

the CASCOM project.

The abstract architecture conceives services to

be delivered essentially by agents. In such an

approach the agents usually act as mere wrappers for

web services. The difference between a web service

and a service provided by an agent boils down to a

matter of interface: an agent can provide an

implemented web service by a process of wrapping

the service within an ACL interface in such a way

that any agent can invoke its execution by sending

the adequate (cg. request) message.

However, agents are not only able to execute the

service but also they can engage in different

communicative interactions around that service.

That means the provider must be able to engage in

several different interactions during the provision of

a service

To better explain our role-based approach, let’s

analyze a use case story of a healthcare scenario. In

this use case a physician of a local emergency centre

(or the patient) can ask a health professional for a

diagnosis on the basis of the symptoms and the

medical records of the patient, like exams and past

diseases.

The “conversation” between the physician of the

local emergency centre and the health professional

can be modelled with a sequence of speech acts

between the two agents involved, as depicted in

Figure 2. The physician requests to the health

professional the status of his availability, that is if he

can provide a second opinion over a certain matter.

If so, he asks him for an opinion, providing the

symptoms and the medical records. If there is no

sufficient information, the health professional

requests (possibly more times) additional

information and finally gives his advisement. If the

provided diagnosis is doubtful or not clear, the

physician can also solicit an explanation.

Starting from this conversation we can isolate

four different interactions: (1) the first is checking

the availability of the health professional, then (2)

the real second opinion exchange starts. It can

comprise (3) a detailed information exchange. When

the second opinion exchange finishes, an

explanation (4) could occur.

Having individuated the basic interactions, now

we can model the second opinion use case in a more

precise way, like depicted in Figure 3, and

furthermore define the roles involved in the

Figure 2: Second Opinion conversation.

Figure 4: Roles and interactions taxonomy.

Figure 3: Second Opinion interactions.

THE CASCOM ABSTRACT ARCHITECTURE FOR SEMANTIC SERVICE DISCOVERY AND COORDINATION IN

IP2P ENVIRONMENTS

207

interactions. The product of this analysis should be a

basic ontology of roles (Figure 4), which is possible

to refine and make more general. For example, the

role SecondOpinionAdvisor could be generalized in

an Advisor role, as well the ExchangeSecondOpinion

interaction could be generalized in an Advisement

interaction, in which the Advisor informs the

Advisee about his beliefs with the aim of persuading

the Advisee of the goodness of these beliefs.

Role and organizational concepts are abstraction

mechanisms that have been used for many years in

object oriented methodologies and are usually

present in most agent-oriented methodologies like

GAIA (Wooldridge, 2000), MASE (DeLoach,

2001), etc. However, at least at the moment, there

are no approaches that use them to describe what the

services do. Furthermore, most current OWL-S

matchmakers (Paolucci, 2002) (Li, 2003) only

consider service inputs and outputs in their matching

algorithm. We are exploring means of taking into

account the roles that agents can play and the

interactions in which they can engage in the

publication of services in the service directory in the

CASCOM project. Roles and interactions will be

defined in an ontology that is being constructed.

The efficiency of the matchmaking process can

be improved by previously filtering those services

that are compatible in the terms of roles and

interactions they can participate in. The precision of

the matchmaking process can also be improved by

including information regarding the roles and

interactions. For instance, a service may have the

symptoms and medical records as inputs and the

report as output. However, the service can consist of

(i) actually generating the report, (ii) retrieving a

previously done or (iii) a brokering service to

contact other (external) healthcare experts. The

inputs and outputs are the same but the role the

service plays is different (advisor, informer or

broker, respectively). The idea in our approach is

that, in the previous second opinion service example,

we only make a detailed search for service provider

agents that are able to play the advisor role.

4 CONCLUSIONS

In this paper, we have described the CASCOM

abstract architecture that smoothly integrates

intelligent agent technology, semantic Web services,

peer-to-peer, and mobile computing, for intelligent

peer-to-peer mobile service environments. The

potential benefits of a role-based interaction

modelling approach have been illustrated based on a

real-world use case scenario for emergency

assistance in the healthcare domain.

The adequacy of the above CASCOM

architecture has been analysed for three different

scenarios in the healthcare, patient telemonitoring &

e-inclusion and shopping mall assistance domain.

We are currently initiating the implementation of

the presented architecture. In particular, the interest

of the authors will focus on the instrumentation of

role-based mechanisms for service discovery and

coordination. At the end of the project, we will come

up with a fully fledge demonstrator for a concrete

real-world business application scenario.

ACKNOWLEDGEMENTS

This work has been supported in part by the

European Commission under the project grant FP6-

IST-511632-CASCOM. The authors would like to

thanks all project partners for their contributions.

REFERENCES

Cáceres, C., Fernández, A., Ossowski, S. CASCOM –

Context-aware Health-Care Service Coordination in

Mobile Computing Environments. ERCIM News 60,

pp. 77-78

CASCOM, 2005. http://www.ist-cascom.org

DeLoach, S.A., Wood, M. F., and Sparkman, C. H.,

(2001), “Multiagent systems engineering”.

International Journal of Software Engineering and

Knowledge Engineering, 11(3):231-258

Li, L., and Horrock, I. (2003). “A software framework for

matchmaking based on semantic web technology”. In

Proc. 12th Int World Wide Web Conference Workshop

on E-Services and the Semantic Web (ESSW 2003).

Lopes, A., and Botelho, L. (2005) "SEA: a Semantic Web

Services Contextaware Execution Agent".

Proceedings 1st International AAAI Fall Symposium

on Agents and the Semantic Web, Arlington VA, USA

Paolucci, M., Kawamura, T., Payne, T., and Sycara, K.,

(2002). “Semantic matching of web services

capabilities”. In Proceedings of the First International

Semantic Web Conference on The Semantic Web,

pages 333-347. Springer-Verlag

Serrano, J.M.; Ossowski, S.; Fernández, A. (2002), “The

Pragmatics of Software Agents – Analysis and Design

of Agent Communication Languages”, Intelligent

Information Agents – The European AgentLink

(Klusch et al. ed.), Springer, pp. 234−274

Wooldridge, M., Jennings, N. R., and Kinny, D. (2000),

“The Gaia Methodology for Agent-Oriented Analysis

and Design”. Journal of Autonomous Agents and

Multi-Agent Systems 3(3) 285-312.

ICEIS 2006 - SOFTWARE AGENTS AND INTERNET COMPUTING

208