Comparison of methods for solving sets of linear inequalities in the bounded-error context

Effective recursive updating of the solution set of linear inequalities has recently gained importance in the area of parameter bounding for system identification, prediction and control. When it is not empty, this solution set is a convex polyhedron, usually a convex polytope in the context of parameter bounding. Several algorithms have been proposed in the literature to update this polyhedron when a new inequality is introduced. This paper describes three of them in a unified framework and compares them on the number of operations involved and the memory space required.