Extending UML State Diagrams to Model Agent Mobility

ACEEE Int. J. on Communication, Vol. 02, No. 03, Nov 2011

Extending UML State Diagrams to Model Agent
Mobility
Feras Ahmad Hanandeh1 and Majdi Yousef Al-Shannag2
1
Department of Computer Information Systems, Faculty of Prince Al-Hussein Bin Abdallah II For Information Technology,
Hashemite University, Jordan.
Email:Feras@hu.edu.jo
2
Department of Computer Information Science, Faculty of Information Technology
Yarmou University, Jordan.
Email:majdis@yu.edu.jo

Abstract— This paper presents a simplified form of UML state as UML 2.0. The UML is gaining wide acceptance for the
diagrams for modeling agent mobility. Mobile agent has gained representation of engineering artifacts in object-oriented
more importance technology. The notations used to model software. Our view of agents as the next step beyond objects
agent mobility are focused on capturing agent creation, leads us to explore extensions to UML and idioms within
mobility paths and current agent location. In this paper, we
UML to accommodate the distinctive requirements of agents.
demonstrate how the simplification of the state UML 2.0
Activity Diagrams can be used for modeling mobile agent
The result is an AUML [5]. Agent interaction protocols are
applications. The paper concludes with the appropriateness used to model the interaction between agents. It is important
of the presented approach for modeling agent mobility with to model the complex logic, including data flow, within a
UML state diagrams as well as with sequence diagrams of the software agent [6]. To overcome these limitations, we propose
mobile agent system. using the UML 2.0 [8] activity diagram to specify action plans.
This diagram models the system behavior by the state
Index Terms— UML State Diagrams, Agent Mobility, UML diagram. Actions are considered the basic units of the system
Sequence Diagrams behavior. The activity diagram is the most noticeable change
in UML 2.0 [6]. The paper is organized as follows: Section 2
I. INTRODUCTION presents related work. Section 3 elaborates modeling agent
This research extends UML 2.0 states diagrams to the mobility using UML sequence diagrams. In section 4 an
one of the most important new concepts which is the mobile approach for modeling agent mobility using state diagrams
computing which gains more and more interest. The objective is presented. Section 5 concludes the paper.
behind this research is that agent concepts and mobile
software agents have become part of the system and service II. RELATED WORK
architecture of new generation networks, systems and Mario Kusek and Gordan Jezic in [1] presented a proposal
services. The main goal of the presented approach is to for modeling agent mobility with UML sequence diagrams.
successfully model such a dynamic environment with a good They described and compared four approaches according to
representation of agent mobility and execution paths. The their clarity, the space needed for graphics and their expression
advent of network technology has prompted new computing of mobility. The stereotyped mobility diagram gives an overall
and communication paradigm in which codes and processes view of nodes and agent mobility paths. The swimlaned
can move over the network. Mobile agents are autonomous mobility diagram gives a clear representation of agent mobility,
computing entities (codes/processes), which can current locations and creation. Both stereotyped and
autonomously decide where to move over the network and swimlaned mobility diagrams clearly represent agent
what tasks to perform at different hosts [3]. A large number of execution and mobility paths, but they are not suitable for
programming languages, systems and APIs have emerged to modeling the systems with a large number of nodes and
support this paradigm; so there has been an emerging trend agents. In state representation mobility diagrams, mobility is
to use Unified Modeling Language (UML) for modeling represented by changing the state of a moving agent. In
agent-based systems in general. The most notable work in frame fragment mobility diagrams, each frame fragment of a
this arena is Agent UML (AUML) [7] in which UML is sequence diagram represents the execution at a node. The
extensively used to model all aspects of agent. Agent state representation and frame fragment mobility diagrams
technology enables the realization of complex software have poorer representations of migration and execution paths.
systems characterized by situation awareness and intelligent These diagrams are suitable for modeling multi–agent systems
behavior, and can open the way to new application domains with mobile agents and a large number of nodes. An advantage
while supporting the integration of existing and new software, of these approaches is a better overall view of agent roaming
and make the development process for such applications and current agent location, but in some cases it is not pos-
easier and more flexible. However, deploying agent technology sible to order agents in such a way that one frame fragment
successfully in industrial applications requires industrial- can represent all the agents at a certain node [1]. Hubert
quality software methods and explicit engineering tools, such Baumeister and et al. in [2] presented an extension to UML
© 2011 ACEEE 12
DOI: 01.IJCOM.02.03. 521


class and activity diagrams to model mobile systems and with plans, to check the information in the agent mental state
they assumed that mobile objects can migrate from one loca- by using guard conditions, to describe messages, to repre-
tion to another. Locations can be nested and mobile too. sent agents changing their roles, to describe the actions that
They introduced stereotypes to model mobile objects, loca- are duties and rights, to model agents moving from an orga-
tions, and activities like moving or cloning, and they intro- nization to another and to model agents moving from an en-
duced two notational variants of activity diagrams for mod- vironment to another [6]. Dianxiang Xu and Yi Deng in [9]
eling mobility. They found that locations, mobile objects, proposed a two-layer approach for the formal modeling of
and the atLoc relation could be modeled explicitly as abstract Logical Agent Mobility (LAM) using predicate/transition
classes and associations in class diagrams and mobile ob- (PrT) nets. They viewed a mobile agent system as a set of
jects, like Passenger and Planes, would inherit from these agent spaces and agents could migrate from one space to
classes [2]. Miao Kang and et al. in [3] proposed an exten- another. Each agent space is explicitly abstracted to be a
sion of activity diagrams in UML 1.5 for modeling mobile component, consisting of an environmental part and an in-
agent applications and found that the latest version of UML ternal connector dynamically binding agents with their envi-
2.0, which provides better model elements in activity dia- ronment. They used a system net, agent nets, and a connec-
grams, is more effective and flexible in modeling mobile agent tor net to model the environment, agents, and the connector,
applications. This observation prompted them to upgrade respectively. They presented a case study of modeling and
their work in UML 1.5 to UML 2.0, and they demonstrated analyzing an information retrieval system with mobile agents.
how UML 2.0 activity diagrams can be used for modeling They obtained that the modeling of migration process has
mobile agent applications and discussed their underlying led to some insights into logical code mobility of software
computational models which capture mobility of agents. They agents as well as the differences from physical mobility in an
used multidimensional hierarchical partitions to model loca- ad hoc networks. In terms of migration request, two styles of
tions and agents respectively, which help to visualize the migration, autonomous and passive, are formally identified.
algorithmic behavior of multi agent systems with locality [3]. Strong mobility is made explicit by ensuring state preserva-
Bernhard Bauer and James Odell in [4] shown how UML 2.0 tion during migration. Agents are disabled when deactivated
can be applied for the specification of agent-based systems, upon migration and disconnected from environment. Migra-
and they given a short overview on existing agent method- tion as atomic change of location attribute is inadequate for
ologies to have a reference what has to be specified in such logical code mobility, though location is a critical concept.
systems. They found that many of the errors and inconsis- Managing locations by environments, rather than by agents
tencies of the original submission have been rectified, and themselves, facilitates the investigation of agent transfer [9].
more than 3000 issues were files and resolved by the UML Jacques Ferber and Olivier Gutknecht in [10] presented a
2.0 Finalization Task Force. As such software vendors can generic meta-model of multi-agent systems based on organi-
begin to build software tools that support the UML 2.0 Su- zational concepts such as groups, roles and structures, which
perstructure and Infrastructure [4]. James Odell and et al. in called AALAADIN, that defines a very simple description of
[5] addressed two requirements (relating the use of agents to coordination and negotiation schemes through multi-agent
the nearest antecedent technology “object-oriented software systems. This meta-model allows for agent heterogeneity in
development” and using artifacts to support the develop- languages, applications and architectures. They introduce
ment environment throughout the full system lifecycle) by the concept of organizational reflection which uses the same
describing some of the most common requirements for mod- conceptual model to describe system level tasks such as
eling agents and agent-based systems using a set of UML remote communication and migration of agents. AALAADIN
idioms and extensions. They illustrated the approach by pre- is able to overcome these problems by allowing designers of
senting a three-layer AUML representation for agent inter- multi agent systems to describe any kind of organization
action protocols and conclude by including other useful using only the core concepts of groups, agents and roles.
agent-based extensions to UML. They suggested several They also presented a development platform, called MadKit,
straightforward UML extensions that support the additional in which all these concepts have been implemented [10].
functionality that agents offer over the current UML version Marcus Huber in [11] presented a hybrid intelligent agent
1.3. Many of these proposed extensions are already being architecture (JAM) that draws upon the theories and ideas
considered by the OO community as useful extensions to of the Procedural Reasoning System (PRS), Structured Cir-
OO development on UML version 2.0 [5]. Viviane Torres da cuit Semantics (SCS), and Act plan interlhtgua. JAM draws
Silva and et al. in [6] proposed the use of UML 2.0 activity upon the implementation pragmatics of the University of
diagrams to model agent plans and actions. They considered Michigan’s and SRI International’s implementation of PRS
a plan to be an activity. Both plans and activities are com- (UMPRS and PRS-CL, respectively). The JAM agent archi-
posed of actions and define the action execution sequence. tecture also provides an agentGo primitive function utilizing
They demonstrated how these diagrams can be applied to Java’s object serialization mechanism to provide widely-sup-
model agent plans and actions by using some features avail- ported mobility capabilities. Huber [11] extended the JAM
able in the UML 2.0 activity diagrams and defining some new plan representation to include declarative representations for
ones. They seen that it is possible to describe actions using individual primitive actions and implementing partial order
a domain-independent notation, to associate goals and roles planning algorithms based upon the new representations.
© 2011 ACEEE 13
DOI: 01.IJCOM.02.03.521


III. MODELING AGENT MOBILITY WITH SEQUENCE DIAGRAMS IV. MODELING AGENT MOBILITY WITH STATE DIAGRAMS
A sequence diagram in UML is a kind of interaction diagram A state diagram is a type of diagram used in computer
that shows how processes operate with one another and in science and related fields to describe the behavior of systems.
what order. It is a construct of a Message Sequence Chart. State diagrams are used to give an abstract description of the
Sequence diagrams are sometimes called event diagrams, behavior of a system. This behavior is analyzed and
event scenarios, and timing diagrams. represented in series of events that could occur in one or
more possible states. State diagrams require that the system
A. The Sequence Diagram of the UML 2 in Model Agent
described is composed of a finite number of states;
Mobility
sometimes, this is indeed the case, while at other times this is
- Stereotyped Mobility Diagram: a reasonable abstraction.
The stereotyped mobility diagram [1] introduced three
stereotypes (Agent <<agent>>, Location<<at>> and Move A. The State Diagram of the UML 2 in Model Agent
<<move>>) as presented in Figure 1. The diagram initiated Mobility
by locating the agent at node n1 which is indicated with a Representing agent mobility by state diagram is a
message with stereotype <<at>>. After that, agent moves complicated process when there are many agents with many
from this location to location at node n2, and that is locations. Figure 3 represents agent mobility state diagram
represented with message and stereotype <<move>>. When for 2 agents and 2 locations.
the agent creates a new agent, it is indirectly done by sending
message to node where the new agent should be created.

Figure 3.Agent Mobility State Diagram for 2 Agents and 2
Loca tions
Figure 1.The Stereotyped Mobility Diagram representing 2 Agents The diagram shows that Agent1 move from Loaction1 to
and 2 Locations
Location2 (change its state) by the action Move and creates
B. Case study: Simple price searcher diagram a new agent (Agent2) by the action New. Adding a new
This diagram as presented in [1] represents three network location (Location3) means that extracting the previous
nodes: Home, Host1 and Host2 (store agent), which is diagram to the diagram in Fig. 4:
responsible for providing pricelist. The Searcher agent is
created at the Home node. The Searcher agent move from
Home node to Host1 node and requests Store1 agent to be
provided by the price list. The Store1 agent responds with
the whole pricelist. The Searcher extracts the price for the item
and move to the next node. After visiting all nodes the Searcher
agent move back to the Home node and informs the user where
and how much the price of the specified item.

Loca tions
Again adding a new agent (Agent 3) means that extracting
the previous diagram to diagram in Fig. 5:

Figure 2. The Stereotyped Mobility Diagram for the Scenario
© 2011 ACEEE 14
DOI: 01.IJCOM.02.03. 521


The diagram initiated by locating the created agent; A[1] at
location; L[0]. After that, once the agent moves from this
location to another location by the action Move, j is increased
by one and the agent will be located at location L[j+1]. When
the agent creates a new agent, it is indirectly done by the
action New which increases i by one such that the new agent
is A[i+1].

V. CONCLUSION
As the mobile agent system is an important technology,
we demonstrate how state UML 2.0 Activity Diagrams can
be used for modeling it. We use notations for the model agent
mobility to capturing agent creation, mobility paths and
current agent location. The presented approach extracts the
State Diagram that can be used for multi locations and agents.
The simplified state diagram presented in this research
competes with the sequence diagrams for modeling mobile
agent applications.
Loca tions
REFERENCES
As the number of agent’s and locations increases, the agent
mobility state diagram becomes huge and reading is difficult [1] M. Kusek and G. Jezic. “Extending UML Sequence Diagrams
because of many relating parameters. Within such an to Model Agent Mobility”, AOSE’06 Proceedings of the 7th
environment, agents repeatedly interact with one another, international conference on Agent-oriented software engineering
VII, 2007.
and each agent’s actions affect the joint state of all agents,
[2] H. Baumeister, N. Koch, P. Kosiuczenko and M. Wirsing.
which in turn make the agent mobility state diagram complex. “Extending Activity Diagrams to Model Mobile Systems”, In
B. Simplified State Diagram of the UML 2 in Model Agent Proceedings of NODe ’02 Revised Papers from the International
Mobility Conference NetObjectDays on Objects, Components,
Architectures, Services, and Applications for a Networked World,
As a solution of the above state diagram drawbacks, the 2003.
suggested approach of this article use two arrays, one for [3] M. Kang, L. Wang and K. Taguchi. “Modelling Mobile Agent
Agents, A[i] and the other for Locations, L[j]. Along with, Applications in UML2.0 Activity Diagrams”, Proceedings of the
adding a new computation notation to increase the index of 3rd International Workshop on Software Engineering for Large-
the agents array and/or of the locations array as presented Scale Multi-Agent Systems, IEE, Edinburgh, United Kingdom, 2004,
in Figure 6. pp 104 - 111.
[4] B. Bauer and J. Odell. “UML 2.0 and agents: how to build
agent-based systems with the new UML standard”, Engineering
Applications of Artificial Intelligence 18, 2005, 141–157.
[5] J. Odell, H. Parunak and B. Bauer. “Extending UML for
Agents”, Proc. of the Agent-Oriented Information Systems
Workshop at the 17th National conference on Artificial Intelligence,
2000.
[6] V. da Silva, R. Noya and C. de Lucena. “Using the UML 2.0
Activity Diagram to Model Agent Plans and Actions”, AAMAS
‘05 Proceedings of the fourth international joint conference on
Autonomous agents and multiagent systems, 2005.
[7] B. Bauer, J. P. Muller, and J. Odell. “Agent UML: A Formalism
for Specifying Multi agent Software Systems”, In Proceedings of
the First International Workshop on Agent-Oriented Software
Engineering AOSE’00, Limerick, Ireland, LNCS 1957 Springer, 2001,
pp. 91-103.
Figure 6.Agent Mobility State Diagram for n Agents and m [8] UML: Unified Modeling Language Specification, version 2.0,
Loca tions
OMG. Available at: <http://www.uml.org>. Accessed in: May 10,
2011.

© 2011 ACEEE 15
DOI: 01.IJCOM.02.03.521

Extending UML State Diagrams to Model Agent Mobility

More Related Content

What's hot (19)

Viewers also liked (8)

Similar to Extending UML State Diagrams to Model Agent Mobility (20)

More from IDES Editor (20)

Recently uploaded (20)

Extending UML State Diagrams to Model Agent Mobility