A Parallel Algorithm for Observability Analysis on Networks of Workstations

VAZQUEZ, GUSTAVO E.; PONZONI, IGNACIO; BRIGNOLE, NÉLIDA B.

Dallas, Texas, Estados Unidos

Conferencia; AIChE Annual Meeting 1999; 1999

AIChE (American Institute for Chemical Engineering)

In this work we present a parallel depth-first search technique for the detection of all the paths of a given length existing in a graph. The implementation was carried out using the Parallel Virtual Machine (PVM) library. The algorithm constitutes the indispensable core for the parallelization of novel structural techniques for the matrix-partitioning problems that arise in process plant instrumentation design. The code was tested under a heterogeneous distributed environment. As to performance analysis, we propose a new metrics to obtain suitable speed-up measurements for environments where the machines have different computing powers.