A decentralised conflict and error detection and prediction model

Abstract

The objective of this research is to study the performance of a newly developed decentralised conflict and error detection and prediction model (CEDPM) over different networks. In the CEDPM, conflict and error (CE) detection and prediction agents are deployed at each collaborative unit to detect and predict CEs, and exchange information with the support of a CE detection and prediction protocol. Two metrics, detection and prediction time and conflict severity, are defined to evaluate CEDPM and CE propagation, respectively, for linear, divergence, convergence, and parallel coordination networks in which different task dependences exist among collaborative units. Experiment results show that the CEDPM performs significantly better for networks with parallel activities. The conflict severity reflects network complexity and increases as the dependence between collaborative units increases. The findings are useful for the design of emerging prognostics systems-of-systems.

Keywords:

• collaborative control theory (CCT)
• conflict prediction
• error detection
• information systems
• modelling
• multi-agent systems
• prognostics

Acknowledgements

This research has been developed with the support of PRISM Center. This article is partly based on Xin Chen's Master Thesis completed in 2005.