Essay Sample on Multi-Agent Systems: Agent-Based Modelling and Simulation

Introduction

Multi-agent systems refer to the computerized systems with intelligent agents that depict a multi-interactive ability. On the other hand, agent-based models are computational models used in the simulation of the interactions and actions performed by autonomous agents (Bonabeau, 2002). They are aimed at establishing an insight which explains the collective behavior exhibited by agents following simple rules. Technological advancements have led to the use of software to simulate the desired systems and their operation tested before they are made into actual systems. Also, simulation techniques have been used in agent-based modeling to come up with prototypes and robust system models. The resulting simulation tools have been adopted in the supply chain to create efficient systems with low failure possibilities. This literature review evaluates existing literature to explain the application of agent-based modelling simulation and the current research gaps being witnessed.

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Save 25%

Testing for the Robustness of a System

Agent-based models have been applied in investigations involving the robustness of policies by creating a virtual environment which simulates the real one. Santa-Eulalia, Ait-Kadi, D'Amours, Frayret, and Lemieuxcd (2011) used an agent-based modelling to create an environment simulating the advanced planning and scheduling systems. The researchers used industrial data set from companies located in Quebec in efforts of establishing the robustness of control and planning policies. It was found out that simulations give insights on planning methods and the control levels possessed by a system. Also, they provide directive followed to get the desirable robust configuration of parameters. The provision of the optimum configurations minimizes the likelihood of uncertainties occurring due to shifts in demand, manufacturing and supply. It should be noted that the production processes are guided by the present and future demand. As such, the application of agent-based simulation predicts whether a particular system can adjust to changes by simulating an environment bearing the investigated parameters.

Simulation in Planning

Adequate planning is a vital element since most of the systems operate under uncertainties (Ayub, 2015). Simulation methods aid in the prediction of the most likely events which may introduce friction in the activities such as the production system. Through virtual environments, simulations are in a position to create near perfect environemnts which expose the behavior of the system and appropriate measures taken in the event such scenario occurs in real-life. Ivanov, Sokolov, and Kaeschel (2010) back these statements after embarking on a survey to create a framework which can be used to introduce a planning tool for supply chain operations. After analyzing previous studies, they found out that simulations in the agent-based modelling lead to the creation of a stable supply chain which is resistance to a crisis and adapts well to changes when they occur. The use confirms the ancient literature by Wooldridge and Jennings (1995) who stated that planning would shift from laboratory to software use of simulations in planning.

Dai, Lin, and Long (2014) echo the same by explaining the uncertain nature of complex systems in the supply chain. In the global sphere, the supply chain is regarded as a complex system which renders its analysis a difficult task. They concur with the idea of the introduction of simulations as an ideal method addresses the stochastic, uncertain and dynamic nature of supply chain systems. Since a complex system has numerous systems, simulation in agent-based modelling is used to produce an all-round system with fewer uncertainties. With the use of simulation, the most uncertain aspects in the supply chain such as transportation, inventory and manufacturing can be determined.

Arbor and Parunak (1997) object the idea that simulation in agent-based modelling alone can aid in the streamlining of complex systems. In their arguments, the system designers must first consider the environment under which the system operates. They pay attention to the life of insects such as ants and wasps and associates their success with the ability to work in ways which do not induce a conflict with their immediate environment. An environment is considered as active since it contains process which can alter its original state and is independent of any actions preceding from its agents. The various agents in the environment comprise input and output coupled to the elements found in a particular state, but the environment fails to identify the coupled components. Barbati, Bruno, and Genovese (2012) validate this argument by explaining that simulation should be used to eliminate the optimization problems found in the heterozygous and distributed environment. In this regard, a system should be structured in a way that it interacts with its environment to eliminate system failures.

System Agility

Simulations in the agent-based systems are used to introduce agility in operations involving the processing of big data. With the average supply chain systems being regarded as complex, the introduction of the process of large data introduces a higher system complexity. Managing such a system requires the use of simulation and agent-based systems to initiate a faster responsive nature of the system. Giannakis and Louis (2016) carried out a research to investigate the working of a multi-agent- based system in introducing agility in a supply chain with significant data processes. It was found out that simulations increase the agility of a system by increasing its speed, flexibility and responsiveness. The same ideas are echoed by Wang, Wan, Zhang, Li, and Zhang (2016) who explain that simulations create a self-organized system with efficient coordination and feedback for significant data processes. Simulations lead to the creation of system designs which integrates risk management and production capabilities leading to the flexibility that minimizes delays during processing of big data.

System Restructuring

With the changes being witnessed in services, simulation is imperative in reengineering systems to meet the increases in demand. An example is the reengineering of the supply chain to accommodate the rise in consumer expectations. Although system reengineering is ideal, their analysis should be performed before the new systems are made to gauge their success. Simulation comes to the rescue by creating software which represents main constituents of a system. In the efforts of developing an ideal supply chain, Swaminathan, Smith, and Sadeh (1998) developed software which simulated transporters, manufacturers and retailers and possessed their respective control elements. After the simulation, they were in a position to create an ideal supply chain and evaluated the effects of new entries such as changes in supplier and addition of additional distribution centers on production.

Santa-Eulalia, Halladjian, D'Amours, and Frayret (2011) back the possibility of system restructuring by analyzing the existing literature on simulation. While still using the case study for the supply chain, they highlight the major areas which have been addressed by the present literature. For instance, they explain the current research have covered areas such as methodological approaches, application and the different types of modules which have been developed. With the presence of vast knowledge in the simulation sector, their argument implies that system restructuring is possible with the application of the right module. Also, the present breakthrough witnessed in the simulation is enough to create advanced planning systems for supply chains.

Santa-Eulalia, D'Amours, and Frayret (2012) Present conflicting ideas on system scheduling using simulation. After the analysis of the existing literature, they found that most of the survey lack clear identification of the specific experimental factors to consider when modeling. Also, Long (2014) explains that present simulation models lack a universal standard to be used when structuring new systems. Such issues emanate from the complexity of systems, for instance, the supply chain system, which leads to an endless list of the factors to consider when modelling. With this setback, researchers only pick a few experimental elements and leave the rest. The resulting systems are not as effective as the factors left out still affect the output. Although they present a framework for an agent-based simulation, lack of an existing one shows that system restructuring to achieve a near perfect one is currently impossible to achieve.

Research Gaps

Although the current Agent-based modelling and simulation techniques have been widely adopted, they lack a way of validating and verifying results (Loo, Tang & Ahmad, 2018). When a new model is formulated, its results are compared with the results produced by the previous models. However, there is no stipulated method which validates the new results obtained. Such a scenario has led to the production of models which record improvements over the previous models but introduce further complications. Lack of proper validation has also led to system failures as a result of errors which were not factored in during the development of software used to create new simulation models. For efficient simulation, a robust validation and verification scheme should be designed for researchers to establish the viability of their simulation tools.

Currently, there is no developed methodology for planning using simulation in the supply chain (Santa-Eulalia, Frayret, & D'Amours, 2008). The complex nature of supply chain renders it difficult for a simulation tool to predict the factors which contribute to the system's environment. With such limitations, planning for the perfect simulation environment cannot be achieved leading to the production of tools which do not fully replicate the actual environment. Research should be developed to come up with a simulation methodology which captures almost all aspects of the surrounding environment in which the systems being investigated operate. Such advancements will create nearly error-free agent-based modelling and simulation methods that exhibit a high level of efficiency.

Conclusion

In summary, agent-based modelling and simulation have been used in many functions in the field of supply chain. For instance, simulation has been used to develop a robust system, and this is achieved through the simulation of the real environment which enables researchers to predict system performance. It has also been used in the planning process whereby simulations are used to establish the behavior of a system and measures taken to prevent the occurrence of unwanted outcomes. However, some research has doubted the success of the simulation and explained its inability to achieve substantial success as expected. The setbacks explained include lack of universal standards and failure to simulate the real environment, contributing to inefficient simulation models. The current research gaps witnessed include lack of validation and verification methods to determine the viability of simulations and a methodology to predict the factors in the surrounding environment which affect system performance.

References

Arbor, A., & Parunak, H. V. D. (1997). "Go to the Ant": Engineering Principles from Natural Multi-Agent Systems. Annals of Operations Research, 4049(Artificial Intelligence and Management Science), 1-27.

Ayub, B. (2015). Risk and Uncertainty in Engineering Systems: Editor'a7s Introduction to The Aims and Scope of the ASCE-ASME. Journal of Risk and Uncertainty in Engineering Systems, 1(1): 1-5. Doi. 10.1115/1.4026400.

Barbati,...