Evaluation of a sort-based matching algorithm for DDM

Abstract

The High Level Architecture (HLA) provides several Data Distribution Management (DDM) services to reduce the transmission and reception of irrelevant data. These services rely on the computation of the intersection between "update" and "subscription" regions. As the brute force approach does not scale well, the current implementations of DDM usually use a grid partitioning of the routing space to efficiently compute the intersection. Since this approach still has some shortcomings, we have focused our recent research on the design and the evaluation of different intersection algorithms for the DDM.In this paper, we introduce a new intersection algorithm in which regions are sorted before computing the intersections. The strong points of this algorithm are (1) the sort suppresses many useless checks, (2) the algorithm allows many optimizations at the implementation level. An experimental comparison of this algorithm with brute force and grid-based approach provides additional information about the good points and weak points of each of the three approaches.