Research On Checkpointing Simulation Tool In Mobile Computing Environment

As an important fault tolerance strategy which has been used widely in mobile computing systems, a checkpointing fault tolerance scheme has significant meaning for improving the overall performance of mobile computing systems. For the sake of algorithm design and performance improving of fault tolerance schemes, the application of simulation technology is an effective way to reduce the cost of actual network deployment, improve the quality of experiments, as well as shorten the development period. In this thesis, we studied the simulator for checkpointing fault tolerance schemes in mobile computing environments, and provided a test platform for the performance study of checkpointing fault tolerance schemes in mobile computing environments.Firstly, in this thesis, the checkpointing fault tolerance schemes in the mobile computing environments are been studied and categorized, and for each category, their common characteristics are been uncovered been represented as different discrete events. Meanwhile, we take multiple discrete events to denote one single process in a distributed checkpointing scheme and construct the simulation model of discrete events, and then the main frame of the simulator for checkpointing fault tolerance schemes in mobile environments is followed. On the basis of the main frame of the simulator, we analyze and design its primary functions, modules, interface, and parameters, as well the simulation model and the structure of each module.Secondly, based on the overall design, a mobile computing environment is constructed in this thesis,the handoff and moving mechanism of mobile nodes is simulated, the topological structure is set up, and the checkpointing fault tolerance scheme is designed. Subsequently, the generated model of discrete event generates discrete events according to the principles of discrete event-driven, and reads discrete events to simulate by adding new checkpointing protocols to the application layer to GloMoSim. The simulator developed in this thesis can guarantee all the checkpoints algorithms subjecting to same discrete events and then fairly comparison, and obtain checkpointing setting overhead, communication overhead, storage overhead, running with error-free overhead and so on. Furthermore, we described the contents and activities of each function, as well as the solutions for some important problems. This simulator can also generate a series of message sequence and system failures randomly in terms of the requirements of users. In order to enhance the scalability of this simulator, we leave interfaces in each function for the future development of various types of checkpointing fault tolerance schemes. This makes the simulator more practicability.Finally, we take the pessimistic and optimistic message logging algorithms to test the performance of this simulator. Experimental results indicate that the simulator developed in this thesis works in a good way, and it can simulate checkpointing fault tolerance schemes in mobile computing environments effectively, as well as meeting the needs of study.