Parallel Hypercube Implementation of Haar Transform

Wavelets transform, especially discrete wavelets transform (DWT) is an important tool for many applications. This paper describes one parallel implementation of simplest wavelet transform, namely Haar transform. A theoretical and experimental analysis of this parallel implementation in a cluster environment is also described. Implementation is based on hypercube architecture and for experimental results, a simple cluster computer (with Red Hat Enterprise 4 and OSCAR 4.2 in computer network with 8 PCs) is used. The programming language used here is C++ with LAM MPI 2 paradigm. The results show the impact of computation and communication on the parallel implementation and its comparison with one process implementation which time is O(n).