Parallel algorithms for optimal data allocation in a dynamic CIM network

An efficient, massively parallel optimization technique is developed for solving the dynamic data allocation problem in medium to large scale applications such as computer integrated manufacturing (CIM) systems. This method is based on a significantly reduced feasible state search space. A statistical evaluation framework compares the performance of the proposed technique with other dynamic data allocation strategies. Algorithms are actually implemented for a variety of I/O task activation scenarios, with the number of task activation nodes ranging from 50 to 250. The overall performance of the proposed method has a significant improvement over other optimization strategies, especially as the number of task activation nodes increases.<>