A Parallel Domain Decomposition FDTD Algorithm Based on Cloud Computing

摘要

This paper presents a parallel domain decomposition finite difference time domain (DD-FDTD) algorithm based on MapReduce architectural pattern in a Hadoop cloud computing cluster. The algorithm is implemented on a 6-nodes Hadoop laboratory test cloud computing cluster to compute the electromagnetic fields of lightning in the downtown area in Shanghai city, PR China. The speedup ratio under different numbers of computational sub domains is evaluated. It shows that the maximum speedup ratio of the algorithm implemented on our Hadoop cluster is about 2.4, which will increase with the scale of the mesh model and the nodes of the Hadoop cluster.