Computer Science

Analytical Agent Technology for Complex and Dynamically Evolving Environments


The importance of simulation environments stretches throughout the academic and corporate hemisphere as they have dealings in not only aeronautics, electronics, and chemistry, but also biology, and physicist. All these different industries rely heavily on simulators in order to better understand and validate the numerous theories that help to explain the complex and dynamic systems. The purpose of this project is investigating of simulation agent-based paradigm for designing and developing frameworks that potentially focuses on monitoring of the systems. Within this research, the use of multi-agent simulators and multi modeling agents will offer a new perspective that can be used to address the evolving complex and dynamic systems that will differ from the explanation through mathematical equations. The techniques created will be devised in developing multi-agent simulators that have the ability to track the dynamic evolving system simulated.

In developing the multi-agent systems, the project will rely on integration of computational intelligence aligned machine knowledge that will allow for analytical and reasoning with developing the multi-agent simulation systems.  The importance of developing the multi-agent systems is to self-manage and self-adapt to evolving environments that are able to deal with different real world scenarios and phenomena that are created from the behavior of the dynamic systems. That will yield tremendous societal and academic impact dealing within the study of human centered and industrial systems, which will provide input in natural resource management.

Problem Statement

Previous research has been conducted on the importance of implementing multi-agent systems in complex and dynamic evolving environments (Zhang, Xu, Shresth, 2008) (Bosquet, Barreteau, Le Page, Mullan, Weber, 1999). In complex industrial and real world situations where the simulated environment has dynamically complex phenomena is not easily described with the use of analytics or mathematical equations. The use of multi-agent modeling systems (MAS) are a necessary tool in order to develop, manage, and explain theories used in simulated systems. “Multi-Agent System (MAS) is a suitable programming paradigm for distributed information systems and applications, where resources, data, control and services are widely distributed.” (Zhang, Xu, Shresth, 2008) Within the multi-agent modeling, the system utilizes the collection of autonomous decision making agents based on the development of human systems, which are entirely complex to manage and sustain due to the changing nature of the dynamic environments. The use of these systems is use within different industries, but more importantly has potential to be used in economic, environmental, and financial systems.

Research has defined such phenomena that surpass the use of traditional mathematical equations to be large-scale, complex and highly dynamic. Earlier research based on the economic impact weighed the essential benefits of using the artificial adaptive agents over its linguistic and mathematical counterparts, categorizes (redefines) the modeled reproduction of this complex phenomenon as complex adaptive systems, which possess certain attributes (Bonabeau, 2002).  Bonabeau defined the systems as one that consists of a network of interacting entities (agents), where aggregate behavior and properties emerge as a result of the individual behavior of the agents.  Within human developed systems, the Bonabeau took a different approach by the  question posed of when Agent-Based Modeling (ABM) should be used, and concluded its necessity in the case of a potential emergent phenomenon where the individual behavior of the entities is nonlinear (if-then rules), deeming it difficult to model using mathematical models like differential equations. Memory, learning curve, and adaptation effects the individual behavior of the entities.  The average functions and homogeneous mixing exhibited in traditional mathematical models are not the truest adaptations of the actual modeled phenomenon. According to Bonabeau, stochasticity and randomness are an integral part of the entities behavior within the systems. Apart from capturing emergent phenomena, dependent on human created and controlled systems attributed ABM’s superiority to its natural framework of capturing human behavior using “behavioral” entities. Its features of flexibility, as it is easy for example to add or remove agents to the system and tune the level of complexity of individual agents. Current research has come far from the previous decades where the capabilities of agent-based modelling wasn’t greatly known, however, more research is needed in understanding the nature of the dynamic systems and the distinct approach to the simulators system development.

The aim of this project is to deliver a generic architecture that is non-specific in its design and capable of features used in dynamic and complex evolving environments where simulators are being applied. The use of the simulators have potential to be utilized in environments of road traffic systems, economic systems, biological change, climate change, Peer to Peer and ad Hoc complex network, forest fire simulation systems, and other industries dependent on adaptable technology in dynamic environments. In order to deal with different scenarios, the implementation will incorporate diverse computation methods and hybrid approaches. These approaches will have roots in diverse disciplines including artificial intelligence, machine learning, neural networks, evolutionary algorithms, genetic algorithms, data mining, and fuzzy logic among others.

Importance of the problem

Within this study, the agents aim to design and test an agent-based simulation prototype application, where the development of the agent-based systems are self-adaptive and self-managed in order to survive and use when different real-world and complex industrial scenarios. The agents would be autonomous in developing the analytical, and well coordination of others where the capabilities of reasonable thinking are necessary in the application of one’s own expertise and perception of the environment. The need for development of multi-model systems into agents that have the characteristics of intelligence, autonomy, and adaptability, in order to be integrated into systems that are flexible to changes is essential in the furthering of academic and social impacts. The research of the multi-model agents within simulations can be utilized as a primary tool in understanding the human societies within the various social processes of human activities. Previous research has been conducted on the topic has not fully realized the capabilities of using in the agented-based paradigm for the benefits of monitoring and managing the complex and dynamic evolving environment. The use of data intensive applications in order to be applied in the surpassing phenomena  for the purpose of decision making process that rely on sensors from the changing environment. This research will help to analytical develop a framework that is explained without the dependence of previous mathematical equations to sastisfy the goals of the project.

Research Questions

The questions that will be asked are based on the goals outlined within the project that will include, however, as the project progresses more questions will need to be asked in order to provide an adequate system capable of the objectives outlined:

What are the required human intelligence required to develop a multi agent system?

What are the different approaches needed in defining and developing autonomous agents and multi-agent systems?

What type of framework is needed to develop agent systems that can scale to real world applications?

What will be the different information technology applications that will be utilized in compiling non-monotonic logic theories for the simulators?

Is the developed system capable of self-adaptability and self-management?

What have been previous multi-agent systems used in simulation of complex and dynamic environments?

What types of methodologies are successful in data and computer integration for application requirements?


           In conducting the project the use of agents that will work autonomously in devising techniques in developing multi-agent simulators. With the use of previous published research based on the topic and potential fields of use, along with established qualitative and quantitative data developed of multi-agent simulators. The use of agent based technology in addressing the systems in combination with integral computational intelligence, and knowledge of machine learning techniques.


          The timescale will be dependent on numerous factors including the use of readily available research and resources from data streams of the dynamic environments that are used in simulation development prototypes. The technology will play a definite factor in the timescale of the project, along with collaboration and integration of information from other agents working on the goals of the project.


Alkhateeb, Faisal, Eslam Al Maghayreh and Iyad Abu Doush. Multi-Agent Systems – Modeling, Interactions, Simulations and Case Studies. Intech. 2011. Web. 28 May 2013.

Bosquet, Francosi, Oliver Barreeau, Chistophe Le Page, Christian Mullon, Jacques Weber. Environmental modelling approach. The Use of Multi-Agent Simulations. Advances in Environmental and Ecological Modelling. 1999. Web. 28 May 2013.

Bonabeau, Eric. Agent-based modeling: Methods and techniques for simulating human systems. PNAS. 2002. Web. 28. May 2013.

Brazier, Frances M.T, Barbara Dunin-Keplicz, Nick Jennings, Jan Treur. DESIRE: MODELLING MULTI-AGENT SYSTEMS IN A COMPOSITIONAL FORMAL FRAMEWORK. N.p. n.d. Web. 28 May 2013.

Hanna, Lindsey, Jonathan Cagan. Evolutionary Multi-Agent Systems: An Adaptive and  Dynamic Approach to Optimization. Carnegie Mellon University. 2008. Web. 28 May 2013.

Hongzhong, Deng. Multiagent Simulation of Complex Dynamic Evolving Network’s Survivability. Intelligent Systems (GCIS), 2012 Third Global Congress. 8 Nov. 2012. Web. 28 May 2013.

Lazar, Alina, Robert G. Reynolds. Computational framework for modeling the dynamic evolution of large-scale multi-agent organizations. Wayne State University. 2000. Web. 28 May 2013.

Li, Qingshan, Lili Guo, Xiangqian Zhai and Baoye Xue. Intelligent Collaboration Environment in Multi-Agent System Enabling Software Dynamic Integration and Adaptive Evolving. Xiadan University. 01 April. 2011. Web. 28 May 2013.

Luke, Sean, Claudio Cioffi-Revilla, Liviu Panait, Keith Sullivan, and Gabriel Balan. MASON: A Multi-Agent Simulation Environment. George Mason University. N.d. Web. 28 May 2013.

Uhrmacher, Adelinde M, Danny Weyns. Multi-Agent Systems: Simulation and Applications. Taylor & Francis Group. 2010. Book. 28 May 2013.

Zhang, Xiaoqin, Haiping Xu & Bhavesh Shresth. An Integrated Role-Based Approach for Modeling, Designing and Implementing Multi-Agent Systems. University of Massachusetts at Dartmouth. 2010. Web. 28 May 2013.