Measuring the BDARX architecture by agent oriented system a case study

• Main Authors: Khan, Sundas Naqeeb, Mohd Nawi, Nazri, Witjaksono, R Wahjoe, Abdul Hamid, Noorhamreeza, Shahzad, Asim, Ullah, Arif, Rahman, Atta-ur-
• Format: Article
• Language: English
• Published: Penerbit UTHM 2018
• Subjects: QA299.6-433 Analysis; TA168 Systems engineering
• Online Access: http://eprints.uthm.edu.my/4562/1/AJ%202018%20%28793%29%20Measuring%20the%20BDARX%20architecture%20by%20agent%20oriented%20system%20a%20case%20study.pdf; http://eprints.uthm.edu.my/4562/

Distributed systems are progressively designed as multi-agent systems that are helpful in designing high strength complex industrial software. Recently, distributed systems cooperative applications are openly access, dynamic and large scales. Nowadays, it hardly seems necessary to emphasis on the potential of decentralized software solutions. This is because the main benefit lies in the distributed nature of information, resources and action. On the other hand, the progression in multi agent systems creates new challenges to the traditional methodologies of fault-tolerance that typically relies on centralized and offline solution. Research on multi-agent systems had gained attention for designing software that operates in distributed and open environments, such as the Internet. DARX (Dynamic Agent Replication eXtension) is one of the architecture which aimed at building reliable software that would prove to be both flexible and scalable and also aimed to provide adaptive fault tolerance by using dynamic replication methodologies. Therefore, the enhancement of DARX known as BDARX can provide dynamic solution of byzantine faults for the agent based systems that embedded DARX. The BDARX architecture improves the fault tolerance ability of multi-agent systems in long run and strengthens the software to be more robust against such arbitrary faults. The BDARX provide the solution for the Byzantine fault tolerance in DARX by making replicas on the both sides of communication agents by using BFT protocol for agent systems instead of making replicas only on server end and assuming client as failure free. This paper shows that the dynamic behaviour of agents avoid us from making discrimination between server and client replicas.