WSEAS Transactions on Computers
Print ISSN: 1109-2750, E-ISSN: 2224-2872
Volume 13, 2014
Reliability Improvement and Reliability Assessment for Distributed Hardware-Software Multi-Agent Systems
Authors: ,
Abstract: The problem of reliability of new object of distributed hardware software (DHS) multi-agent system (MAS) is considered. DHS MAS is determined as a system that is based on agent technologies and consists of both agents and hardware components required for execution of agents and for interaction of agents with an environment. Reliability improvement, fault-recovery and several reliability assessment approaches for DHS MAS are presented. The reliability improvement methodology is built upon replication of unique functional components and redundancy of universal components. The fault-recovery methodology defines a set of fault-recovery procedures required for restoration of consistent system configuration after failures of its components. Methodology for operability function formation was developed to enable utilization of logical-and-probabilistic methods for reliability assessment. Another approach for reliability assessment is based on Markovian model and system state graph and was developed to overcome limitations of logical-and-probabilistic methods that are suitable only for systems with hot standby. The state graph based approach allows reliability assessment for DHS MAS with cold standby and different operating modes of system components. The state graph based approach is also applicable for a case when probability of failure of one of components depends on states of other components. New failure model for determination of failure rates of system components in accordance with system state is introduced. Computing experiments described in the article have validated developed methodologies.
Search Articles
Keywords: Multi-agent system, reliability, fault-recovery, operability function, redundancy, state graph, logical-and-probabilistic method
Pages: 670-690
WSEAS Transactions on Computers, ISSN / E-ISSN: 1109-2750 / 2224-2872, Volume 13, 2014, Art. #60