ABSTRACT:With significantly improving the overall manufacturing operations commonly known in smart factory practices, vertical and horizontal integrations of various components are introduced throughout the entire value chain. The proliferation of smart factory practices introduces the fourth industrial revolution called Industry 4.0. Industry 4.0 creates new challenges and the application of networked manufacturing systems is one of the important features required to handle the systems in an efficient way by communicating and processing information to each other. Distributed systems provide coordination to allow the global information to be available for a better decision making and consequently achieving high efficiency. This connectivity helps for a better decision making and therefore achieve high efficiency. However, the evaluation on the accuracy of global information between different architectures of distributed systems (centralized systems and decentralized systems) has little work mentioning in the state of current manufacturing systems under the conception of Industry 4.0. As a result, this research will fill the gap by providing an accuracy of the decentralized system over the centralized system together with their sensitivity analysis. In this study, the decentralized system is built based on the concept of Epidemic protocols or Gossip-based protocols, while the centralized system is a simple client-server. The epidemic protocol is a bio-inspired communication paradigm that imitates the behavior of virus when the outbreak occurs in a community. The accuracy in both systems has been monitored by means of simulations. The effects of message loss to the accuracy of centralized and decentralized systems are studied. By comparing the system accuracy between both systems, it was found that the accuracy of the decentralized system is generally more accurate when the system is used for a long period. The accuracy tends to be lower down especially when the information is not completely distributed, while the accuracy of the centralized system receives an excessive suffer from message loss.