Parallel optimal hierarchical control using a MIMD architecture

Hierarchical control problems for large scale systems solved via parallel processing were considered. The authors initially made use of the natural parallelism of the prediction of interaction algorithm. This implementation presented a reasonable computational performance in terms of global processing time when compared to the sequential implementation of the algorithm. Then, algorithm partition was implemented by subdivision of the coordination task into N subtasks, resulting in a superior computation performance. A parallel architecture with shared memory was utilized for both implementations