![]() | Only 14 pages are availabe for public view |
Abstract This dissertation discuses mainly the problem of reconfigurable parallel processing that solve a complex problem using the ant algorithm. The ant algorithm was modified in order to decrease the total algorithm execution time. To solve the main problem of this dissertation a number of procedures were performed. First the ant algorithm is to be modified to improve its performance. This modification results in speeding up the ant algorithm compared to the classical one. A simple implementation of this approach was tested on the traveling salesman problem (TSP). the results show that our modified ant algorithms have good performance compared to the original ant algorithm. Second, the generation of task graphs that are related to real problems from a source code is to be obtained automatically. These task graplıs are used as the input to the reconfigurable parallel processing system. Third, a number of algorithms were realized using the modified ant colony algorithm to assign each task in a given task graph to a processor in a reconfigurable multi processor architecture. Each group of ants is working collectively to find the interconnection architecture that best suited for the task graph and the best scheduling of that task graph on the chosen interconnection architecture. The total execution time is computed according to the knowledge of the individual run times of tasks and the communication requirements among tasks. The inter task communication time depends on the interconnection processor architecture. The results of using the modified ant colony algorithms in reconfigurable parallel processing system are compared with those obtained using genetic algorithm. |