Self-adaptive Fault-Tolerance of HLA-Based Simulations in the Grid Environment

Jijie Huang,Xudong Chai,Lin Zhang,Bo Hu Li
DOI: https://doi.org/10.1007/978-3-540-77600-0_7
2008-01-01
Abstract:The objects of a HLA-based simulation can access model services to update their attributes. However, the grid server may be overloaded and refuse the model service to handle objects accesses. Because these objects have been accessed this model service during last simulation loop and their medium state are stored in this server, this may terminate the simulation. A fault-tolerance mechanism must be introduced into simulations. But the traditional fault-tolerance methods cannot meet the above needs because the transmission latency between a federate and the RTI in grid environment varies from several hundred milliseconds to several seconds. By adding model service URLs to the OMT and expanding the HLA services and model services with some interfaces, this paper proposes a self-adaptive fault-tolerance mechanism of simulations according to the characteristics of federates accessing model services. Benchmark experiments indicate that the expanded HLA/RTI can make simulations self-adaptively run in the grid environment.
What problem does this paper attempt to address?