Evaluation of Grid-Based Relevance Filtering for Multicast Group Assignment

This paper examines the performance of a grid-based relevance filtering algorithm. The goal is the reduction of network traffic between simulation entities to that which is relevant to their collective interaction. This implementation of relevance filtering utilizes the formation of multicast groups to allow entities to communicate with only those directly affected by their actions. A grid-based filtering technique is discussed and evaluated for its potential to reduce network loading and its consumption of multicast group addresses, using militarily relevant scenarios. An idealized relevance filtering algorithm is illustrated to determine the benchmark which defines the minimum network traffic necessary to support the simulation. The shape of an entity’s area of interest and grid alignment relative to battlefield activity are both evaluated for their effect on filtering performance. The current analysis leads to a recommendation for a grid cell size which minimizes network traffic flow and optimizes use of scarce multicast resources, while realizing these results may be somewhat scenario and simulation dependent.