Research And Application Of Mobile Agent Based Network Parallel Computing

With the increasing demand for computing speed, the traditional single computing can no longer meet our needs. Thanks to the development of the cheap high performance computers, workstations and high performance network technology, parallel computing based on network has become the focus of current parallel computing research. Parallel computing platforms such as PVM, MPI still need to be improved in terms of stability and expansibility. Mobile Agent features autonomy, reactivity and mobility. It provides a new computing model for parallel computing, and basically makes up for some shortcomings of parallel computing platforms. It is characterized with transplantation, expansibility and flexibility. In order to improve the performance of parallel computing, the paper presents a parallel computing module based on Mobile Agent by combining Mobile Agent with parallel computing. It first analyzes the current situation and development trend of parallel computing technology as well as technical features of Agent. On the basis of comprehensive analysis of Mobile Agent and parallel computing technology, the thesis proposes a parallel computing module based on Agent--ABPC. This thesis also gives a thorough introduction of the relevant module, such as communication module, transform module and data access module. It takes into consideration of potential problems in designing the module and explored possible solutions. Finally, by drawing upon the construction of FOPMAX, the thesis elaborates, analyzes and assesses the application and effect of ABPC and makes some suggestions for further improvement.