Research Abstract |
Our study considered a decentralized autonomous network system that consists of multi-layred decision making systems and evaluated the reliability of the network system. First, we proposed a method to evaluate reliability of I out of M ; G system, which is a fault tolerant system, based on a few data. Second, from the hierarchical point of view, we analyzed self-improvement process performed by meta-layr in each decision making organization. This autonomous process dynamically emerges new information and decreases uncertainty so that each multi-layred decision making system can deal with any situation which is not predicted in advance. We also proposed the conditions to manage the system anticipatively. On the other hand, from the network point of view, we found out that parallel genetic algorithms and immune network system models provided adequate models for analyzing decentralized autonomous network system. We showed that a cooperative control was performed by utilizing genetic algorithm. We also paid an attention to information system and database system. We indicated that each component system would be required to be intelligent for individually processing data. For database, we proposed the reflection in data model which makes it easy to revise the database. Lastly, we concluded from the above results that, to prevent global failures and to assure the fault tolerant, individual decision making system must monitor the relation the between itself and other decision systems on real-time and improve its internal evaluation system preventively. This conclusion reveals that the function of meta-layr which performs these activities is the most important.
|