2009 International Conference on New Trends in Information and Service Science

The Development of Multi-Agent System using Finite State Machine

Yazed Al-Saawy, Ajlan Al-Ajlan, Khalid Aldrawiesh and Abdullah Bajahzer
Software Technology Research Laboratory (STRL)
De Montfort University
The Gateway, Leicester, LE1 9BH, UK.
yazed,ajlan,khalid,[email protected]
and also chooses the preferred host from which the
agent begins the search. Also, the model will be
created using the JFLAP software.
After the modelling will be the analysis using UML
which analyses the requirements both functional and
non-functional using the class diagram, class and
sequence diagrams will demonstrate the best software
design which helps to create and develop the software
system. Finally, will be the design of the software
application using JADE and JAVA which will create
the type of the Agent Handler, Agent Interface and
Book Agent System which will search according to the
users search criteria. The system will have two
databases. However, the design will be expandable,
allowing more databases to be added.
This paper will examine how to use agent and
Multi-Agent System (MAS) in a search system. MAS
have provoked more interest in research because of the
advantages found within such systems, for example
they are able to deal with problems that are too large
for a single-agent system are faster and more reliable.
Additionally, they are able to cope with uncertain
knowledge and data. Key areas of the research into the
ability of MAS to solve problems focus on
communication,
coordination
and
negotiation.
Examining how multi agents function in a virtual
system. The virtual system exposes how behaviour of
the components affects the system as a whole.
This paper is organized as follows: Section 2
provides MAS. FSM is described in Section 3. Section
4 presents requirements and Analysis for SAS. Section
5 focuses on the implementation of SAS. Testing &
evaluation are described in Section 6. Finally,
conclusion and future work are presented in Section 7.

Abstract
This paper sets out to develop and create a search
agent that can retrieve requested information from two
databases; the user will be able to search for a book by
inputting variables such as title, author, ISBN and
more significantly for this study the price. The system
will identify the books requested, where they can be
found and will show the cheapest available. SAS will
work from the Application software and will link to the
two databases, the design will allow more databases to
be incorporated in the future. Each of the two
databases has the same names of the titles however
each database may contain a different price. The agent
will get the required information quickly, helping the
user to find the cheapest and possibly nearest. The
agent will connect to the database using the Agent
Interface; there are two agents in the system, known as
MAS. These agents collect the required data and send
it to the Agent Handler.
Keywords: Finite State Machine, Multi-Agent System
JFLAP, Java, and UML.

1. Introduction
The main contribution of this paper is to create
Search Agent System (SAS), which helps the user to
perform a search for a book title quickly; the
information about the books will be stored on two
different databases. The user can search for the book
by name; author or ISBN number and the system will
present all options as well as identify the cheapest price
for the user. The system uses a multi-agent
configuration that works together with a special
platform, the JADE program.
In order to model and demonstrate the software the
Finite State Machine (FSM) is used.
It will
demonstrate how the software will run with multi
agents to search for the requested book and give the
cheapest price. The user can start the search by
selecting either the book name, ISBN, author or price
978-0-7695-3687-3/09 $25.00 2009 IEEE
DOI 10.1109/NISS.2009.162

2. Multi-Agent System
An agent is a software program that communicates
with other software programs, interacting and
responding to behaviour, acting and linking with
available resources on demand. Each agent has size,
capability and speed; all these are attributes of each
203

Multi-Agent
Simulation
Collective
robotics

Multi-Agent
system

Host 2

Application Agent

Application Agent

Application Agent

Host 3
Application Agent

Application Agent

Application Agent

Application Agent

Application Agent

Host 2

Application Agent

agent. The environment of each agent has to be

considered in MAS in order to coordinate the activities
between a numbers of agents. An example of this
could be robots. the agents, (A) are the robots, the
geometrical space (E) is where the robots move, and O
is obviously made up of agents, but also of physical
objects placed here and there which the robots have to
avoid, pick up or manipulate. The operations, Op, are
the actions that the robots can take in moving
themselves,
moving
the
other
objects
or
communicating, and R is the assembly of relations that
link certain agents to others [1, 2, 3].
The domains of application for multi-agent systems
are particularly numerous. We shall refer only to the
main trends, as an attempt to address a wider list could
only end in stagnation; the research of domains is
evolving all the time. Multi-agent systems can be
divided into four main categories: multi-agent
simulation, building artificial worlds, collective
robotics and program design as in Figure 1.

JADE distributed Agent Platform

Jade Front-end

JRE 1.2

Jade Agent Container

Jade Agent Container

JRE 1.2

JRE 1.2

Network Protocol stack

Figure 2. Software Architecture of JADE Agent [17]

The idea behind JADE communication architecture

is that it attempts to offer efficient and flexible
messaging; it openly chooses the best transport and
leverages distributed object technology which is found
within the Java runtime environment. JADE recognises
the difference between inter-platform messaging where
the sender and the receiver agents operate within
different platforms and intra-platform messaging where
the two agents are on the same platform. Inter-platform
messaging has to comply with FIPA requirements;
however intra-platform message delivery is solely an
issue for JADE, so a more convenient transport can be
utilised. JADE uses Java RMI in intra-platform
communication [6, 8]. Java and JADE have attributes
that support multi-agent applications: 1) Autonomy 2)
Intelligence 3 Mobility)

Building artificial
Worlds
Program

Figure 1: Application domains in MAS

In computer science, there is a branch known as

simulation, which consists of analysing the properties
of theoretical models. Simulation is currently an active
area of computer science which is concerned with the
analysis of the properties of theoretical models. The
models are constructed of the mathematical
relationships between the variables which represent the
physical values which can then be measured. The
commonly used models are transition matrices and
differential equations [4, 2].
Brahms is a set of software tools used to develop
and simulate MAS of human and machine behaviour.
In addition, it is rule-based and multi-agent activity
based on programming language. The Brahms
language shows how the situated activities of agents
are represented in a geographical relationship to the
world. Situated activities are actions that occur in a
specific situation; therefore their function is not only
constrained by the reasoning abilities of the agent, but
also by the agents perception of the world [5].
Java Agent Development framework (JADE) agent
can be split over several hosts with one of them acting
as the front end for inter-platform IIOP
communication. JADE is made up of one or more
Agent. Each Jade lives in a separate Java Virtual
Machine and communicates using java RMI. The
Figure 2 shows the architecture of this agent [6, 7].

3. Finite State Machine

This section will examine modelling of the software
using FSM and how the Search Agent will search for
name, ISBN, price and author or maybe just two of
these elements or search all elements together
depending on what the user wants [9]. FSM comprises
of a data structure used to show actions with a
sequence of events. In both Java and JADE
applications, FSM is able to activate and deactivate
certain behaviours in time. This section will define
how to describe FSM using JFLAP as in Figure 3, and
then the coding using JADE. JFLAP is instructional
software used to experiment with grammars automata.
However it also allows experimentation with
applications and proofs. The main feature of JFLAP is
that it can experiment with grammars and theoretical
machines. It allows the building and running of userdefined input on pushdown automata, finite automata,
regular grammars [10].
204

As in Figure 4, the problem analysis is important for

analyzing the user's requirements and to set up the
system boundary. End users of a system frequently
find it difficult to express abstract requirements for the
system, a goal oriented approach to the requirements
analysis is an effective way of improving the elicitation
process. The agents in the system need to be identified
with the relationships between them by agent
modelling to be modelled, according to the relationship
between goals and agents. Therefore, we have
developed a set of heuristic guidelines to create and
search agents to identified major goal, create a
corresponding agent when necessary. A sequence
diagram shows the interactions between objects in a
time sequence. More specifically, it shows the objects
and their interactions with the sequence of messages
exchanged. Sequence diagrams are different from
collaboration diagrams, as they show time sequences
but do not show the relationships between objects. A
sequence diagram can be in generic form, showing all
possible scenarios or in instance form, focusing on one
specific scenario [11].

Here, we will examine the design of FSM, which

has eight states; the first is to start the search using the
name, ISBN, price and author. Afterwards the search
will compare and find the cheapest; this is one of the
final states. The other final state is the normal state
which gives all possible prices of the book. This covers
all possibilities of the search as in Figure 3.

Figure 3: Search Agent Model of Finite State Machine

5. Implementation

4. Requirements and Analysis

This section examines the implementation of the

SAS whilst running, and the functions that a tool built
by the study can perform. The system needs to work on
any PC that have java environment; therefore there are
some requirements to insulate the software, as follows,
JADE 3.4 and JDK. Before running the SAS of Book
Agent has to be put into the JADE folder in C drive so
it is easy to read that class. After that, inside the folder
Book Agent there is a command, the command is
opened and writes Javac the BookAgentSys.java to
compile the class in folder. After compilation JADE
will be run using the JADE tag which is java jade.Boot
t1: RunAgent. t1 is the name of the agent in the pc.
After running the system, it will show the user
interface so the user can start the search.

This section will examine the requirements and

analysis before creating the application software for the
Search Agent. These requirements have to be
identified, giving clear guidelines as to what is required
in the finished, usable software that links multi-agents.
For our framework, the three stages shown in Figure 4
will be followed. The properties of MAS will be
incorporated and a suggestion of required architecture
which reflect the properties will be made. The first
phase in the approach is known as the Problem
Analysis. This phase attempts to gain an
understanding of how the system works in the abstract,
this will be a starting point for the development
process of the architecture. After gaining an
understanding we then move to the next phase Agent
Modelling, here agents will be identified to satisfy the
goals and the goals relationships. Finally the MAS
architecting phase, here the focus in on the
architecture of the agents and the setting up of a
federation of agents that are able to cooperate with
each other, at the same time, to a large extent,
maintaining autonomy [11].
Problem Analysis
System Goal
Elicitation
Goal Analysis
Goal Hierarchy
diagram

Agent Modelling
Agent Search
Agent Class
diagram
Belief and Plan
modelling

MAS Architecting

Figure 5: search By Book Name

Agent Organisation
Considering Agent
Coordination
Agent Internal
Architecting

Figure 5 shows the Book Name as 'java'. The search

will begin in the Host 1, set as the default Host,
however the search can change looking at Host 2 if the
user requires. Any type of search from Book Name,

Figure 4: Architecture development process for MAS

205

The system was to enable users to search for book

titles using various search criteria giving the result
according to book found, price and location. It was
evaluated using user-friendly functionality testing and
performed well against desired objectives. The paper
demonstrated how multi-agents functioned well in
making decisions.
For future development, the design of the system
will be able to handle more than two hosts, a function
that is necessary for overall implementation in a reallife system. This system can be further improved by
incorporation of a facility to show the user the location
of the book stockiest that have the requested title.
Further usability testing can be carried out in
collaboration with a wider spectrum of typical users, as
many users may highlight a multitude of desires and
expectations.

Author, Price, and ISBN can be performed

simultaneously. Figure 5 explains all the books with
'java' in the title in a list that shows the host, price,
ISBN and author. The next label shows the user which
is the cheapest book and where to find the book, this
saves time looking for it from the entire list.

6. Discussion
This paper used two types of testing, which are
Black Box and White Box testing tools. The first
testing is the functional testing, that examines whether
the system behaved according to the set-out objectives
of the search system, for example whether or not the
system was successful in identifying the cheapest book
as in Table 1.
Table 1. Black Test
Book Name Author Host Price Result

Cheapest

JAVA
JAVA
JAVA

The
cheapest
is 22

SUN
SUN
SUN

1
1
1

25
27
22

Found list
Found list
Found list

8. Acknowledgment
The authors wish to acknowledge contributions to
Hussein Zedan, Hongji Yang and Antonio Cau.

The second testing is the usability testing that test

examined if the system met the requirements set out for
'user friendliness', it examines specifically the user
interface and whether or not it is suited to the type of
user i.e. a person searching for a book name and being
given the result including the price and location.
Finally, user acceptance testing was carried out by
giving the system to a user to see if the system met
their expectations and behaved according to what was
expected. Overall, the system was deemed a success in
that in functioned according to the desired objectives.
The functionality, usability testing concluded that the
user was able to input the search variables and was
given the desired result i.e. the book they were looking
for, where it was located and identification of the
cheapest book on offer. The mobility agent performed
very well and it is able to search in two hosts at the
same time.
The system was able to search more than one search
variable simultaneously i.e. book name and host and
still identify the cheapest book. A slight criticism, from
the tested usability perspective (Black Box testing), is
the terminology used in the interface; for example the
term 'host' is a technical term used by the developer
and the term 'location' would be a more user-friendly,
understandable term. FSM helps to modelling in
useful for programmer to understand the requirement.
However, FSM give the mobility of each state how to
be worked and create each of state as class.

9. References
[1] J. Ferber, Multi agent systems, an introduction to
distributed artificial intelligence, Addison Wesley/ Pearson,
London, 1999.
[2] A. Roberto, Multi agent, available from:
http://www.acm.org/crossroads/xrds5-4/multiagent.html
[3]
H.
Fran
Available
from:
http://www.itdirector.com/article.php?articleid=11774, 2004.
[4] M. Dastani, and J. Gomez-sanz, Programming multiagent systems. Knowl. Eng. Rev., Cambridge University
Press, vol. 20(2), pp. 151-164, 2005.
[5]

Agent

Solution,

retrieve

on

2/1/2009

from:

http://www.agentisolutions.com/brahms.htm.
[6] F Bellifemine, et al., JADE: a FIPA2000 compliant agent
development environment, In Proc. of the Fifth on
Autonomous Agents, ACM, New York, pp: 216-217, 2001.
[7] F. Bellifemine, et al., JADE, White Paper, V.3(3), 2003.
[8] Bellifemine, et al., Developing Multi-agent Systems with
JADE. In Proc. of the 7th international Workshop on
intelligent Agents,. Springer, London, pp: 89-103,2001
[9] B. Lee and E. Lee, Interaction of Finite State Machines
and Concurrency Models, in Proc. of Thirty Second Annual
Asilomar Conference on Signals, Systems, and Computers,
Pacific Grove, California, 1998.
[10] S. Rodger, et al., Increasing interaction and support in
the formal languages and automata theory course, SIGCSE
Bull, vol.39(3) ACM New York, NY, USA, pp:58-62, 2007

7. Conclusion and Future Work

[11] S. Parka and V. Sugumaranb, Designing multi-agent

systems: a framework and application, Expert Systems with
Applications, Seoul, Korea vol. 28(2), pp: 259271, 2005.

The overall aim of the paper was to create and

develop a Search Agent which uses DAI using MAS.
206