Comparison of four parallel algorithms for systolic array processing

Four parallel algorithms for solving least square estimation problems are studied. They are the recursive modified Gram-Schmidt (RMGS) algorithm, the a priori error recursive modified Gram-Schmidt (PERGS) algorithm, and the Given rotation with and without using square root operator (GRWSR and GRWOSR) algorithms. These algorithms based on orthogonal transformation provide a complete solution to the least squares estimation problem. They are highly reliable and have a highly modular structure suitable for systolic array implementation. The designed VLSI systolic array chip can provide a high throughput.<>