A dynamic sort-based DDM matching algorithm for HLA applications

Abstract

Simulation is a low-cost and safe alternative to solve complex problems in various areas. To promote reuse and interoperability of simulation applications and link geographically dispersed simulation components, distributed simulation was introduced. The High-Level Architecture (HLA) is the IEEE standard for distributed simulation. To optimize communication efficiency between simulation components, HLA defines a Data Distribution Management (DDM) service group for filtering out unnecessary data exchange. It relies on the computation of overlap between update and subscription regions, which is called “matching”. There are many existing matching algorithms, among which a sort-based approach improves efficiency by sorting region bounds before the actual matching process, and is found to outperform other existing matching algorithms in many situations. However, the existing algorithm performs matching for all regions in one round and cannot dynamically deal with a selective region modification without processing all the regions once again. Realizing that in many spatial applications, only a small subset of all regions are actually modified in each time step, this article proposes a dynamic sort-based matching algorithm to deal with this efficiently. Theoretical analysis has been carried out for the proposed algorithm and experimental results show that the proposed algorithm has significantly better performance than major existing matching algorithms at dynamic matching.

Parallel Data Distribution Management on Shared-memory Multiprocessors

“…The ITM algorithm, while slower than SBM, can be easily extendable to cope with dynamic regions since the interval tree allows efficient insertion and deletion of regions. In fact, a version of SBM that can efficiently handle region updates has already been proposed [49], but it can not be readily adapted to the parallel version of SBM discussed in this paper. Developing a parallel and dynamic version of SBM is the subject of ongoing research.…”

Optimizing Pairwise Box Intersection Checking on GPUs for Large-Scale Simulations

“…-sweep-line-based algorithms: Bentley and Ottmann [1979], Bentley and Wood [1980], Six and Wood [1980]; -tree-based algorithms: Vaishnavi and Wood [1982], Edelsbrunner [1983], Petty and Mukherjee [1997], Zomorodian and Edelsbrunner [2002]; -uniform partition-based algorithms: Van Hook et al [1996], Tan et al [2000], Boukerche et al [2005]; -sort-based algorithms: Raczy et al [2005], Gupta and Guha [2007], Pan et al [2011].…”

A parallel data distribution management algorithm

“…SBM first sorts the endpoints, and then scans the sorted set (details will be given in Section III-D). SBM is extended in [8] to work efficiently on a dynamic scenario where extents can be moved or resized dynamically.…”

“…In most applications, extents can move and grow/shrink dynamically; if an update extent, say U j , changes its position or size, then it is necessary to recompute column j of matrix M. The brute force approach applied to U j alone gives an O(n) algorithm, since it is only necessary to identify overlaps between U j and all n subscription segments. An extension of SBM capable of updating intersection information efficiently has been proposed [8], with an asymptotic cost that depends on various factors (e.g., upper bound of the dimension, maximum bound shift in a region modification).…”