Task scheduling using ACO-BP neural network in computational grids

Task Scheduling in computational grid is a complex optimization problem which may require consideration of different criteria such as waiting time, makespan time, throughput, communication time, and dispatching time. For optimal scheduling, the scheduler must know about the above factors and status of the resources in the grid and include these dynamic changes in the availability of resources while scheduling the tasks. For all situations, classical algorithms cannot adapt themselves with situations. The heuristic algorithms are proved to be more efficient than classical scheduling algorithms. This paper propose a method that tunes the Back Propagation Neural networks (BPN) using the capability of ACO algorithm to produce an optimal solution. Proposed method reduces the computation time by removing the unnecessary links in the neural network structure. The algorithm increases the efficiency the scheduling process and allocates the tasks to best available resources in the computation grid.