Title :
VSO: Self-Organizing Spatial Publish Subscribe
Author :
Hu, Shun-Yun ; Chen, Kuan-Ta
Author_Institution :
Inst. of Inf. Sci., Acad. Sinica, Taipei, Taiwan
Abstract :
Spatial publish subscribe (SPS) is a basic primitive underlying many real-time, interactive applications such as online games or discrete-time simulations. Supporting SPS on a large-scale, however, requires sufficient resources and proper load distribution among the simulation units. For load distribution, existing mechanisms either use a static partitioning, such that over-provisioning or overloading are bound to occur, or require manual adjustments unsuitable for massive workloads. We describe Voronoi Self-organizing Overlay (VSO), which extends a Voronoi-based Overlay network (VON) to automatically partition and manage a logical space to support SPS. Efficient resource usage thus is possible as only the units necessary to maintain the system are used. Load is also balanced among the resource units so that overloading or over-provisioning can be avoided. We use simulations to verify our design and describe some preliminary results.
Keywords :
computational geometry; message passing; middleware; SPS; VON; VSO; Voronoi self-organizing overlay; Voronoi-based overlay network; load distribution; self-organizing spatial publish subscribe; static partitioning; Games; Integrated circuit modeling; Load management; Load modeling; Logic gates; Radar; Subscriptions; publish subscribe; scalability; spatial computing;
Conference_Titel :
Self-Adaptive and Self-Organizing Systems (SASO), 2011 Fifth IEEE International Conference on
Conference_Location :
Ann Arbor, MI
Print_ISBN :
978-1-4577-1614-0
Electronic_ISBN :
1949-3673
DOI :
10.1109/SASO.2011.13