A Grid Implementation of a Cellular Neural Network Simulator

In this paper we present a cellular neural network simulator and its implementation on a grid infrastructure. We propose a master slave architecture with a divide and conquer approach by which, exploiting the data level parallelism, the image is divided in parts, according to the number of computational elements used in the grid infrastructure, each part is independently processed, and finally a conquer steps merges all the results. The simulator was applied to digital image processing. Some considerations on the boundary conditions are also investigated.