Agent-Based Approach to Software Engineering: Assignment 2

Desklib provides past papers and solved assignments for students. This report explores agent-oriented software engineering methodologies.

Agent-oriented Software Engineering and
Methodologies
(Assignment 2)

TABLE OF CONTENTS
INTRODUCTION...........................................................................................................................1
AGENT-ORIENTED APPROCAH TO SOFTWARE ENGINEERING.......................................2
CONCLUSION................................................................................................................................5
REFERNCES...................................................................................................................................6

INTRODUCTION
Invention of multi-agent technology and of autonomous agent is considered as one most
remarkable events in the sector of computer science. The Agents are adopted as new theoretical
computation model for effective engineering in distributed and complex systems. The increasing
software systems are also viewed as autonomous agents. It is quite a difficult task to developed
high quality software for its application in real-world due to complexities in flexibility and
number of component and interrelation among them. This above statement is always true
irrespective of techniques or model applied[1]. Software engineering is effective in offering
effective techniques and models for minimizing complexities and effective handling the
complexity for developing software. The software engineer can use various and wide range of
software engineering paradigms like object-orientation, software architectures, design patterns,
and component-ware. The main purposes of designing of the paradigm are to make the overall
process of engineering easier and to minimize complexities while building software.
The contemporary methods are effective in developing the software, however while developing a
distributed and complex system the contemporary model fails to establish proper interactions
between different computational entities as they are very inflexibly defined and representing the
system’s inherent is improper as there is insufficient mechanisms for that too. On the other hand,
agent paradigm responses effective considering these issues as agent-oriented approaches
improve the ability of the software designer’s ability to design, model and build very complex or
distributed systems[3]. Agent-based approaches can be considered as both logical and natural
evolution different type of contemporary approaches for effective software engineering. The
agent-oriented approach is likely to succeed in the near future as one of the important paradigm
for mainstream software engineering.
Despite of different developed languages, architectures, agent theories and achievement of agent-
based applications, not much of work has been devoted for applying and specifying techniques
for developing application using agent-based. The main role of the agent methodology is
providing assistance to the developer during very phase of agent-based application’s of life cycle
along with its management [6].This report focuses on focusing on some argument of agent
1

computing community for supporting the statement regarding “agent approaches will succeed to
b a mainstream of software engineering”.
AGENT-ORIENTED APPROCAH TO SOFTWARE ENGINEERING
The agreement related to agent-based approaches for being the most suitable approach for
distributed systems, engineer complex, for developing software real-world problems is
increasing. Some of the mandatory characteristics requited by a software engineering paradigm
for engineering for designing and developing complex systems are decomposition, abstraction
and organization[2].
Decomposition by Agent-oriented approach: Agent-oriented approach is effective in
decomposing problem considering autonomous agents engaging high-level and flexible
interactions. The agent has Autonomy property which means that system can be engineered with
various problem solvers having their own unrelenting control thread. Decentralisation is effective
in reducing complexities of system’s control. The agents are capable of making decisions
regarding scope and nature of the interactions makes engineering and management easier.
Abstraction of Agent-oriented approach: While designing a software, powerful abstractions is
related to minimising semantic distance among various units of analysis which are used for
conceptualising problem and for constructing solution paradigm[5]. Interplay between various
subsystems along with constituent components is considered as high-level social interactions.
The Agents systems are effective in cooperating for achieving objectives, establishing
coordination among actions. Agent-based approach is effective in providing rich structural set
for managing and representing organisational relationships
Organisation: Agent-oriented approach also offers explicit representation of structure and
relationships of organisation. Notion of primordial component varies as per requirement of
observer. The whole subsystem at one point can be viewed as primitive part. The recursive
structure of agent is stated as to have low complexity in terms of relationships.
Diverse organisations, relationships and interactions between agents are defined to be
cooperative, self-interested, and competitive and relationships among them can be changed
2

dynamically. Hence, the agent-oriented approaches effectively provide models that is explicit to
represent relationships even in dynamic or complex environment[9]
Agent-based approach a New Mainstream of Software Engineering
The agent-based approach can be viewed as natural evolution in software engineering
considering language and paradigms have key abstractions with Object Oriented Approach.
Agents -oriented view is the most natural way of characterising problems [6]. Basic building
blocks of programming models have rising degrees encapsulation and of localisation offering
each agent with their thread of control. Agents effectively reuse subsystem components and offer
flexible s interactions pattern wrapping the software with legacy code.
The agent-based techniques can also be considered as ideal computational model for software
development in open or networked systems. Agents are autonomous and active, effective in
solving problem for achieving effective objectives. Agent are proactive and reactive, this support
in inherent uncertainty while open environment can be solved effectively. The agents are overtly
effective in representing organisational relationships effective in becoming influential paradigm
for software development.
Agent-oriented Methodologies
For evading building any methodology from scrape, researchers emphasising on agent-oriented
methodologies usually follows approach of extending methodologies which comprises relevant
characteristic related to agent paradigm [7].
There are some similarities between agent orient paradigm and object-oriented paradigm is
effective in justifying that approach of expanding the methodology of object-orientation.
Recognition of the object-oriented methodologies can also be stated as potential advantage for
extending the object-oriented methodologies to the agent-based systems. Some existent
approaches related to extending OO methodologies to that of multi-agents systems is discussed
below:
Technique of Agent Modelling for BDI agents system: In this method two levels for modelling
Belief, Desire, Intention agents are defined. External view consists decomposition of agents and
3

their interactions. The overall system is carried out using either agent model that is effective in
describing relationship between agents and concrete agents hierarchically or through interaction
model that is effective in relating services, responsibilities, and interactions among agent along
with external systems. Internal viewpoint does modelling of every BDI agent class using belief
model, goal model, and plan model[8].
Multi-Agent Scenario-Based Method: This method is intended for MAS for CSCW or
cooperative work. Analysis phase includes activities such as, Scenario description for
identification through using natural language, Role functional description to define role of agent
roles, Data and world conceptual modelling, for modelling data through agent, System-user
interaction modelling for defining and simulating interfaces. In design phase MAS architecture is
used along with description of scenario. The next phase is Object modelling for defining
attributes, procedures and hierarchies. And Agent modelling phase used for specification of
elements.
Agent oriented methodology for modelling. This methodology is combination of both object-
oriented methodologies, enterprise modelling methodologies with Computer Integrated
Manufacturing Open System Architecture[9]. The models are:
Function Model that is effective in describing various functions with the help of IDEF diagrams.
In Use Case Model is yet another model theta is effective in describing actors that are involved
in function through OOSE using case notation. Dynamic Model is proposed for analysing the
interaction between object. Agent Oriented System comprises of Agent Identification,
Coordination protocols, Plan invocation and Beliefs, Sensors and Effectors. This component
functions together for modelling the software.
Extensions of Knowledge Engineering Methodologies: This methodologies provide basic for
MAS modelling, this so as they deal with system that is based on development of knowledge. As
agents encompass cognitive characteristics, with the help of these methodologies techniques to
modelling of agent knowledge can be attained[7]. Extension of the current knowledge
engineering methodologies is advantageous for acquiring experience for methodologies.
Additionally, existing tools along with problem solving method libraries and developed ontology
4

libraries can be used repeatedly for developing intelligent models for modular integration and
network management.
CONCLUSION
In the current report discussion have been made regarding some current research and directions
in the sector of software engineering and methodologies to built agent-based applications and
system. The report has analysed various abilities of agent paradigm for modelling significant
assets of a complex or distrusted systems including abstraction, organisation and decomposition,
and to understand the probable prospect of the particular approach. From the report above, it can
be concluded that agent paradigm is effective in establishing proper interactions between
different computational entities of system which are real-world and complex, providing effective
mechanisms to represent inherent of the system and its organisational structure[5]. Agent
paradigm can be stated as natural evolution of various contemporary approaches for engineer
software systems. In the current report, discussion has been made regarding various ranges of
agent-oriented methodologies.
The fundamental focus of the report is to understand agent-based model along with computing
having high potential for improving modelling ability, along with building and designing
complex or distributed software systems. The agent-oriented approaches are expected to succeed
eventually to become prominent mainstream paradigm for software engineering. Thus, against
the claim, the study covers limitation of the approach including unpredictability of agent
interactions along with possibility for emergent behaviour [3]. The concurrent and distributed
system are considered as the future trends of computer engineering and science therefore agent-
based models will be important in the near future due its open character and interaction
flexibility.
5

REFERNCES
Books and Journals
[1]Abar, S., Theodoropoulos, G. K., Lemarinier, P., & O’Hare, G. M. (2017). Agent Based
Modelling and Simulation tools: A review of the state-of-art software. Computer Science
Review, 24, 13-33.
[2]Caillou, P., Gaudou, B., Grignard, A., Truong, C. Q., & Taillandier, P. (2017). A Simple-to-
use BDI architecture for Agent-based Modeling and Simulation. In Advances in Social
Simulation 2015 (pp. 15-28). Springer, Cham.
[3]Gomez-Sanz, J. J., & Fuentes-Fernández, R. (2015). Understanding agent-oriented software
engineering methodologies. The Knowledge Engineering Review, 30(4), 375-393.
[4]Heckel, R., Kurz, A., & Chattoe-Brown, E. (2017). Features of Agent-based Models. arXiv
preprint arXiv:1712.09496.
[5]Lee, B. H., & Ahn, H. S. (2016). Distributed formation control via global orientation
estimation. Automatica, 73, 125-129.
[6]Lee, J. S., Filatova, T., Ligmann-Zielinska, A., Hassani-Mahmooei, B., Stonedahl, F.,
Lorscheid, I., ... & Parker, D. C. (2015). The complexities of agent-based modeling output
analysis. Journal of Artificial Societies and Social Simulation, 18(4), 4.
[7]Mao, X., Wang, Q., & Yang, S. (2017, January). A survey of agent-oriented programming
from software engineering perspective. In Web Intelligence (Vol. 15, No. 2, pp. 143-163). IOS
Press.
[8]Padmanaban, R., Thirumaran, M., Suganya, K., & Priya, R. V. (2016, August). AOSE
Methodologies and Comparison of Object Oriented and Agent Oriented Software Testing.
In Proceedings of the International Conference on Informatics and Analytics (p. 119). ACM.
6

[9]Passos, L. S., Rossetti, R. J., & Gabriel, J. (2015). An agent methodology for processes, the
environment, and services. In Advances in Artificial Transportation Systems and Simulation(pp.
37-53). Academic Press.
Railsback, S. F., & Grimm, V. (2019). Agent-based and individual-based modeling: a practical
introduction. Princeton university press.
Shehory, O., & Sturm, A. (2016). AGENT-ORIENTED SOFTWARE ENGINEERING.
SPRINGER-VERLAG BERLIN AN.
7