Revolutionary Architecture for Autonomous Software
Agents
Mr. Gopal Sakarkar
S. G. B. Amravati University
G.H.R.I.I.T., MCA Department,
Nagpur ,India
g.sakarkar@gmail.com
Abstract— The field of software agents is an expansive and
rapidly developing area of research, which encompasses a diverse
range of topics and interests. In order to study the various
Autonomous Software Agent Architectures for agent design, a
comprehensive classification of architecture is required. This
paper identifies the key aspects of different architectures of
software agents and then provides an overview of design issues
regarding the Autonomous agent. Finally focus on different
Modeling Norms for Autonomous Agents. These research results
are helpful to clarify which architecture will be suitable for
today’s growing information eras.
Keywords-Software agent, Agent Classification, Autonomous
software agent architectures, Modeling Norms, Agent design
issues, Agent Technology.
I. INTRODUCTION
Agent technology is a rapidly expanding area that
encompasses many disparate areas of research and offers
several fundamentally different architecture design
approaches. The rapid growth of this field in the past decade
has occurred in parallel with the evolution of the World Wide
Web as Autonomous software agent systems show great
promise for exploiting massively distributed systems such as
the Internet. Despite the popularity of software agent systems,
there is very little consensus about what exactly constitutes a
software agent. Research into agent-based systems has been
quite diverse, which makes it difficult to form a
comprehensive definition for a software agent.
While a range of different approaches has been taken for agent
design, several key features that are shared among all
approaches to agents. The fundamental feature of software
agents is autonomy. Like their human counterparts, software
agents must be able to act on behalf of some other party, be it
a person or another agent. To do this effectively, some degree
of autonomy is required. Hence, agents must be able to take
action when necessary without human interaction. As societies
are regulated by norms and, consequently, autonomous agents
that want to be part of such societies must be able to reason
about norms. However, no reasoning can be done if agents
lack a model of norms that enables them to know how, by
complying with norms, some of their goals might be affected,
this paper try to focuses on various modeling norms for
autonomous agents. A number of existing software
architectures survey and classifications of software agents
have been presented previously [1]. However, these are
Dr. V. M. Thakare
S. G. B. Amravati University
Amravati, India
vilthakare@yahoo.co.in
generally focused on a specific subset of software agents. This
paper builds upon these to develop ample software agents by
implementing the number of design issues regarding the
Autonomous agent.
The paper is organized as follows.
Next sections focus on definitions. Section III describes
proposed classification. Section IV explains the various norms
for Modeling Autonomous Agents. Proposed technology a new
approach for Autonomous Software Agent explains in the
section V. Section VI focus on Design Issues of Autonomous
Agent, finally section VII concludes and gives some future
works.
II.DEFINATIONS
To understand the Autonomous software agent clearly, this
section focused on definitions of software agent. Some of them
are as follows
Software agents are software entities that can act on behalf of
user in an autonomous fashion, and perform their action in
some level of proactivity and reactivity [2].
It is a persistent software entity dedicated to a specific purpose
and also it is computer program that simulate a human
relationship by doing something that another person could do
for you [3].
It is software entities, which functions continuously and
automously in a particular environment, often inhabit by other
agents and process [4].
With this, it is better to fix up the definition as given below by
considering the fact that ”Autonomous agents are not directed
by commands from a user. Instead, they are directed by a set of
tendencies [5].”
“An Autonomic Software Agent means that the agent has its
own behavior control mechanism and it can base on its
internal status and working environment information to carry
out its own tasks. Alone with the task changed, it can
automatically upgrade its own configuration to complete the
new tasks.”
 III.PROPOSED CLASSIFICATION offers a sound mathematical foundation and (4) an architecture
that supports run time activities.
AGENT OWL-P may contribute to autonomous software agent Web
service interactions in the cross cutting area of interactions and
protocols. OWL-P can offer both a language and a framework
for defining and executing the agent-Web service interactions
[9].
Agent Design XML and RDF
Application Technology Behavior structure XML technologies are being used in the presented agent
architecture. The use of agent and XML technologies in
Air Traffic Control OWL-P Mobile Agent Three Level Arch. autonomous architecture can help secure the interoperability
Network Management XML and RDF Reactive Agent Component base and reusability factors.
E-Commerce JINI and UPnP Interface Agent Open Design Arch
Agent base Firewall etc. Hybrid Agent Bus base Arch. XML plays an important role in the success of a management
Entertainment, etc. Information Agent etc. model based on grids of autonomous agents in what it refers to
the representation and passing of management information
Figure.1 Classification of Autonomous Agent Architecture among the several components of the grid [10].
A. Agent Application Resource Description Framework (RDF), it provides
Network Management interoperability between applications that exchange machine
understandable information on the Web. RDF emphasizes
Security enhancement software system using software agents
facilities to enable automated processing of Web resources.
are developed for preventing unauthorised access to corporate
RDF metadata can be used in a variety of application areas,
information systems is essential for many organisations. In
digital library by intelligent software agents to facilitate
which a core software agent resides on a server and user end
knowledge sharing and exchange etc [11].
software agents reside at each use workstation. By
downloading a pattern of typical user behavior and rules JINI
governing invalid behavior from a core agent to each user end
agent, all decisions and actions about a typical or invalid user Jini technology is a service oriented architecture that defines a
behavior can be taken by a user agent. This permits security programming model which both exploits and extends Java
detection to continue even when the core agent fails to operate technology to enable the construction of secure, distributed
[6]. systems consisting of federations of well-behaved network
services and clients. Jini technology can be used to build
Air traffic-flow management adaptive network systems that are scalable, evolvable and
flexible as typically required in dynamic computing
Air traffic controllers and airline personnel are limited
environments.
resources in air traffic management. To compensate for their
scarcity, Autonomous agents can be employed both in UPnP
simulations (i.e. human operators) and in actual operations
(assisting their human counterparts)[7]. Universal Plug and Play (UPnP) technology is a
distributed, open networking architecture that employs
Agent Base Firewall TCP/IP and other Internet technologies to enable seamless
Distributed firewall uses the autonomous agents to proximity networking, in addition to control and data
coordinately control the traffic on the network. Agents to transfer among networked devices in the home, office, and
implement a distributed firewall to improve the performance of public spaces.
classic firewalls [8].
C. Agent Behavior
B. Agent Technology a) Mobile agent: It has the ability to move around some
WOL-P network.
b) Reactive agent: It shows a reaction or response to the user,
Web Ontology Language for Protocols (WOL-P) has been which should not wait to be told what to do next.
recommended as a technology that will contribute to software c) Interface agent :It is a personal assistant who is collaborating
agents’ interaction with Web services and standard Web with the user in the same work environment.
service technologies. It provides a framework in which to (1) d) Hybrid agent: It refers to those whose foundation is a
develop, (2) specify and combine and (3) enact Commitment-
grouping of two or more agent philosophies within a singular
based protocols.
agent.
OWL-P components include (1) an OWL ontology to support e) Information agent: It performs the role of managing,
the specification of protocols elements, (2) a rule-based manipulating or collating information from many distributed
representation of protocols using for rule expression, (3) an sources; essentially, they help manage the vast amount of
algorithm for combining protocols that is based on algebra and information in wide area networks like the internet [12].
D. Design Structure
For constructing the advance architecture for Autonomous
agent, first figureout some past developed Autonomous Agent
architecture. Regarding that the paper review such type of
architecture from 1997 to 2008. In this review first study the
an “Architecture for Autonomous Agents Integrating
Symbolic and Behavioral Processing”, then “The Behavior-
Based Architectures of Autonomous Agency”, thirdly find out
some key point of “The goal autonomous agent architecture”,
then we study the “Service-Oriented Architecture”, after that
try to understand the” Model-driven Architecture” with
“Component Based Architecture” and “Open Design
Architecture”, finally study the recently develop “Software
Bus based Autonomic Agent Architecture”.
a) An Architecture for Autonomous Agents Integrating
Symbolic and Behavioral Processing
Chella, Gaglio, Sajeva, Torterola [13] investigate the
autonomous agent which links together in a principled way of
the two different research traditions i.e. functional and
behavioral approaches of robot architectures.
In this they make use of the three levels of representation of
new architecture for autonomous. The architecture integrates
the symbolic and the behavioral processing of data coming
from the robot sensors. The integration is based on the
introduction of a conceptual space representation that links the
subconceptual level, which is a repository of behavioral
modules, with the symbolic level, in which rich symbolic
descriptions of the Autonomous Agent environment take
place.
b) The Behavior-Based Architectures of Autonomous Agency
E Architecture, that is based on OMG (Object Management
B B, G
G, B
Reactive Reactive – Deliberative
Agent Deliberative Agent
Agent
Emergent Functionality Hybrid Functionality Useful Functionality
(may be undeliverable ) (may be useful) but closed world
Operation
Figure 2. Three dominate types of architectures of autonomous
agency. [Source paper 14]
Where B= Behavior, G=Goals and E=Environments
Amol Mali [14] presented the three dominant classes of
architectures for autonomous agency and describes key
features of several architectures from these classes by
reactive/behavior-based component in their architectures;
interact with their environments continuously. The author
classified the autonomous agent architectures in to three
categories, first is Reactive agents that follow the “sense-act”
cycle and exhibit functionality that arises out of sequential and
concurrent execution of behaviors. They design the behavior
base robotic system that provide the robot to navigate in an
uncertain and unpredictable world by encoding thee robot with
behaviors and coordinating them independently.
Second is the Reactive agents augmented with deliberative
control to prohibit undesired functionality. Deliberative
capability is generally provided in the form of a planner that
monitors and controls the execution of reactive behaviors.
Finally, third one is the Deliberative agent that use internal
world models and make the closed-world assumption.
c) The goal autonomous agent architecture
Z.SHEN et al [15] modeled the goal autonomous agent
architecture that act as an “anytime” autonomous agent.
The goal autonomous agent model and architecture have been
presented to describe the goal autonomous agent from the
logical view and the structural view respectively. The model
support both behavior autonomy and goal autonomy that was
base on Goal Net.
d) Service-Oriented Architecture
Tsai et al [16] provides the service-oriented architectures, in
which the applications are regarded as services accessible on
network.
Architecture of SOA application is dynamic, what means
(except other consequences), that application can by
composed in time of program execution using existing
services. SOA thus provided new way of creating software
architectures, where architecture is determined in time of
application execution and can be dynamically changed for
fulfillment of new software requirements.
e) Model-Driven Architecture
Sheng and Benatallah [17] design the Model-Driven
Group) standards and it isolates business and application logic
from technological platform. Core idea of MDA is using of
UML language for specification of static interfaces and also
dynamic behavior of components in platform-independent
models. One of the principal goals of MDA is support for
platform independent application development.
CM (Computation
independent module)
transformation
PM (platform
transformation independent module)
PSM (Platform PSM (Platform
Specification Model) Specification Model)
transformation transformation
Code Code

Figure 3. The Principle of Model-driven Architecture
[Source paper 17]
f) Component-Based Architecture
Sriraman and Radhakrishnan [18] suggested the Component-
Based Architecture that is base on distributed component
architectures and includes mechanisms and techniques for
development of coarse reusable business/technical
implementation units by considering environment/container
and decomposing service to more SBBs (Service Building
Block), which are connectable distributed parts associated
with: presentation logic, business logic, resource management
logic, integration logic, logic of net events, security logic and
other logics.
g) Open Design Architecture
Miroslav Beličák [19] design the Open Design Architecture
for simplifying IS design and maintenance. In this architecture
the author separating application functionality from
presentation logic such as graphics user interface and it
coupling IS design together with IS.
Open design architecture should provide better facility for
modification of Information System, because it prescribes
tight coupling of IS design with executable binary elements of
IS. Thus, user can easily change the structure of IS without
necessity of IS execution suspension or disconnection of
users. To achieve these goals the author, they utilize software
autonomous agent s and artificial intelligence agents.
h) Software Bus Based Autonomic Agent Architecture
System Environment
Other Agent (Public Blackboard, etc. )
Message Business Data
Process Task Process
Module Module Module
Software
Bus
Controlle
r avoided.
Figure 4. Software Bus based Autonomic Agent Architecture
[Source paper 20]
Zhou Chuan-Sheng [20] proposed the new and practical
software technology for software development and
integration.
It was based on software component and software reuse
technology, through certain standards and specifications, by
creating generic bus interface, communication protocol and
corresponding software components, to implement the
software components plug and play in the software bus.
This Autonomic Agent Architecture is decompose an into
some related Module like Message Process Module, Business
Task Module and Data Process Module, etc.
In which the Message Process Module combine software
component used to help Automatic Software Agent dealing
with message communication. As in different applications or
at different period, the message number and message types
will be different or changeable, then in order to improve the
Automatic Software Agent mobility and flexibility, during the
design of this module, it used software bus technology.
The second module is Data Process Module that also a
combination of software component to help agent working in
Public Blackboard Mode. In agent application system, different
business task using different data model, even in the same
application, with different period, the data model of an agent
uses will be different also. Therefore to improve the agent
adapts to business changes, software bus technology to design
this component.
Last is the Business Task Module for a combination
component. It is the kernel module in the Autonomic Software
Agent. The main responsibility is to complete the business task
controlling and processing of Autonomic Software Agent.
IV. NORMS FOR MODELING AUTONOMOUS
AGENTS
Norms represent the means to achieve coordination among
agents (for MAS), which are assumed to be able to comply
with norms, to adopt new norms, and to obey the authorities of
the system as an end. Norms prescribe what is permitted and
what is forbidden in an agent society. Thus, norms might be
created by the agent designer as built-in norms, they can be the
result of agreements between agents, or can be elaborated by a
complex legal system. A classification of norms into four main
categories: obligations, prohibitions, social commitments and
social codes [21].
Obligations and prohibitions are norms whose purpose is to
ensure the coordination of individuals in a society, and which
agents adopt once they become members of the society.
To represent it, imposing a constraint on punishments as
follows uses the schema of a norm.
Whereas obligations represent goals that addressees must
bring about, prohibitions represent goals that should be
Social Commitments
These are norms derived from agreements or negotiations
between two or more agents . They are part of a deal between
two sets of agents and, consequently, addressees participate
actively in their creation.
Social Codes
Third category of norms is social codes. These are norms,
which are accepted as general principles by the members of a
society or a particular agent group. Rather than being forced
through punishments or rewards, social codes are complied
with as ends in themselves.
V. PROPOSED TECHNOLOGY NEW APPROACH FOR
AUTONOMOUS SOFTWARE AGENT
The autonomous software agent should be compatible with
various recent technologies in the new era of information
technology. Everyday new technologies are launched and to
survive to this neck-cutting environment, autonomous agent
should be getting compatible with these all emerging
technology.
Some of these technologies are as follows
a) Semantic Web
b) SOA
c) Ontology
d) Localization
a) Semantic Web
Semantic Web technologies appear to be the right means to
provide the semantic integration between data and processes
across systems that can be owned by different enterprises.
Autonomous Software agents will play a fundamental role in
building the Semantic Web. Software agents will be
responsible for coordinating searches, comparing and
negotiating on the Web, and greatly reducing user effort .An
important distinction is that agents on the Semantic Web will
not act in a completely autonomous way, but rather they will
take care of the “heavier load” for their users. They will be
responsible for conducting the investigation, with the
obligation to present the results to the user, so that he or she
can make his or her own decisions.
b) SOA
Service Oriented Architecture (SOA) is a paradigm for
organizing and utilizing distributed capabilities that may be
under the control of different ownership domains and
implemented using various technology stacks.
Software agent technology is more and more considered one
of the most interesting technologies to successfully support
SOA. In fact, besides being an ideal mechanism for
implementing complex systems, agent technology is well-
suited to applications that are communication-centric, based
on distributed computational and information systems, and
requiring autonomous components readily adaptable to
changes. The central role that agents should play in a SOA
scenario is to efficiently support distributed computing and to
allow the dynamically composition of Web services.
c) Ontology
Ontology is description of problem domain, where entities of
the domain, its properties and its relations are described.
Ontology defines the meaning of the terms in used content
language and the relation among these terms. The model of
agent communication is based on the assumption that two
agents, who wish to converse, share a common ontology for
the domain of discourse. It ensures that the agents ascribe the
same meaning to the symbols used in the message. Using
ontology not only allows communication between agents but
also gives the possibility for agents to reason about the
concept.
d) Localization
Localization means ensuring that your software "speaks" the
language of your customers. In a world of globalization,
localization can make the web pages will be language
independent.
Autonomous Software agent will be compatible with most of
the languages and on the basis of that it can be communicate
with most of web pages and will keep the track on it.
VI. DESIGN ISSUES OF AUTONOMOUS AGENT
Following are the some design issues regarding the
functionality of autonomous agent [22].
 Autonomous agents are decision maker: the agent would
review the context and make a decision about what to do,
when to do them and what kind of information to
communicate to whom.
 Autonomous agents are reactive to contextual element of
environment:
Agent may need to monitor contextual element (like location,
delivery time, user’s role etc) to provide the information and
service to users. For this agent need the mechanism to perceive,
recognize disseminate different type of context information.
 Autonomous agents should communicated with each other
or directly to user and service:
For this agent should need the platform of communication that
enable them to convey information to other agent, user and
information resource by using some protocol of
communications.
 Autonomous agents act as proxies to devices: Agent
represent the device such as public display can be aware of
presence of other agent and user available in the
environment. These agents enable allowed users and other
agent to use the devices.
 Autonomous agents need a mechanism for authentication:
It needs a mechanism for authenticating users and agent that
want to access them.
 Autonomous agents need to communicate different types of
message:
Agent needs the communication language that enables them
to convey message for requesting the information from devices
or services and responding to such request notifying the
information to user and devices and requesting from other
agent the execution of an action.
 Autonomous agents may have a reasoning algorithm as
complex as the logic of its functionality:
Autonomous agents needs to aware of information regarding
the environment in order to generate its actions.
VII. CONCLUSION
This paper describes a proposed classification for Autonomous
agent architecture. It draws upon several existing ontologies,
but provides an all-inclusive classification that takes into
account the various aspects of agent research. Due to the wide
variety of approaches toward software agents, a simple
classification method that assigns each agent to a single
grouping is insufficient.
Determining which types of agents work well with others will
be of increasing importance as large-scale multi- agent
systems become more widespread across the Internet. This
will also become an issue when designing standardized agent
architectures. As a consequence, analyzing issues relating to
security and certification of agents are very important. In
addition, paper also focused on different emerging technology
and compability of autonomous agent with this new
technology.
[14] Amol Dattatraya Mali, Member, IEEE” On the Behavior-Based
Architectures of Autonomous Agency ”,IEEE TRANSACTIONS ON
SYSTEMS, MAN, AND CYBERNETICS—PART C: APPLICATIONS AND
REVIEWS, VOL. 32, NO. 3, AUGUST 2002
[15] Z.SHEN,R.GAY,Y.MIAO AND C. MIAO”Goal Autonomous Agent
Architecturer” , Proceeding of the 28th Annual Int. Conf. COMPSAC’
04,IEEE
[16] Tsai W. T., Fan Ch., Chen Y., Paul R., Chung J.-Y.: Architecture
Classification for SOA-based Applications, in Proc. of ISORC'06, April 2006,
pp. 295-302
[17] Sheng Q. Z., Benatallah B.: ContextUML: A UML-Based Modeling
Language for Model-Driven Development of Context-Aware Web Services, In
Proc. of the (ICMB’05), July 2005, pp. 206-212

Future work needs to be done on refining the categories, and [18] Sriraman B., Radhakrishnan R.: Component-based Architecture
reaching consensus on ontology for agent-based systems, the Supplementing SOA, Sun Microsystems, March 2005
next approach of research to develop such type of architecher [19] Miroslav Beličák” Open Design Architecture and Artificial Intelligence
that should be generic for all. Localization is one of the key Agent Application in Information Systems Practice” 5th Slovakian-Hungarian
topic of today’s agent era and it need to be researched into Joint Symposium on Applied Machine Intelligence and Informatics Jan 25-26,
further. 2007pp-349-359
REFERENCES [20] Zhou Chuan-Sheng” Research and Design of Autonomic Software Agent
[1] Mr. Gopal Sakarkar, Dr.V.M. Thakare,” A Survey of Various Software Architecture” International Conference on Intelligent Information Hiding and
Agent Architectures and Agent Communication Languages”, first International Multimedia Signal Processing,2008 IEEE
Conference MNGSA’08,Dec.2008 CSIR and IACSIT, India
[21] Fabiola L´opez y L´opez, Michael Luck”Modelling Norms for
[2].Q. H. Mahmoud and L. Yu” An Architecture and Business Model for Autonomous Agents” Proceedings of the Fourth Mexican International
Making Software Agents Commercially Viable ’,38th Hawaii Int. Conf. on Conference on Computer Science (ENC’03), 2003 IEEE
System Sciences-2005,IEEE.
[22]Marcela Roadriguez,Jesus Favela” Autonomous Agents in Collaborative
[3].N. Jennings and M. Woodldridge “Software Agent”, IEE Review ,Jan. Ubiquitous Computing Environments ”,Computation system Vol 10. No
1996.pp17-20 2,2006pp189-204 ISSN1405-5546.
[4] K. Decker,K. Sycara and M. Williamson” Intelligent Adaptive Information
Agents”.
[5] Hemme-Unger K., Flor T., Vögler G.: MDA Development Approach for
Pervasive Computing, OOPSLA’03, Anaheim, CA, USA, 18th ACM
SIGPLAN conf. on OOP, systems, languages, and app., Oct. 26-30, 2003
[6] Pikoulas J, Mannion M, Buchanan W” Software Agents and Computer
Network Security”, 2000 IEEE
[7] Shawn R. Wolfe” Supporting Air Traffic Flow Management with Agents”,
2006, American Association for Artificial Intelligence
[8] Dalila boughaci1, Habiba drias, Brahim oubeka1, Abdelkader aissioui1,
Belaid benhamou” A Distributed Firewall using Autonomous Agents”,
Proceedings of the International Conference on Dependability of Computer
Systems (DEPCOS-RELCOMEX’06), 2006 IEEE
[9] Dominic Greenwood, Margaret Lyell” The IEEE FIPA Approach to
Integrating Software Agents and Web Services ” AAMAS’07 May 14–18 2007,
Honolulu, Hawai’i, USA. 2007 IFAAMAS .
[10] Marcos Dias de Assunção, Carlos Becker Westphall, Fernando Luiz
Koch” XML in Multi-agent based Applications for Network Management“
[11] Ora Lassila ” Introduction to RDF Metadata”
[12] Erlin, Yusof Norazah, Abdul Rahman Azizah” overview on agent
application to support collaborative learning interaction” Jan. 2008, Volume 5,
No.1 (Serial No.38) US-China Education Review, USA
[13]A.Chella,S.Gaglio, G. Sajeva, F.Torterola “An Architecture for
Autonomous Agents Integrating Symbolic and Behavioral Processing
”,,1997,IEEE pp.-45-50