Human Entities' Effect On Server Performance, In Distributed Virtual World Training

Keywords

Distributed Simulation; Performance Analysis; Virtual World Simulation

Abstract

The use of virtual worlds for training continues to expand in the military as advancements in, simulation technology have enabled more efficient and effective simulation-based training. One of the potential, major advantages of virtual world training is the ability to support collective training without the need for, individuals to be physically co-located with each other. In order for distributed military collective training to, become a reality however, the virtual world simulation architecture must be able to support distributed, synchronous and non-deterministic training. This paper is a continuation of our research in which we attempt to, optimize the Military OpenSimulator Enterprise Strategy (MOSES) server architecture in order to support, collective, distributed training. In this paper, we examine the effect that the number of human users have on the, server's processing memory so as to support our development of a predictive model that determines how many, resources are required to support a target number of concurrent users in the virtual world. We discovered a, statistically significant difference in the amount of processing memory utilized based upon server hardware, configuration. Furthermore, we found a positive, linear association between the number of human avatars in the, virtual world and the amount of processing memory required by the server. These observations allow virtual, world designers and administrators to know the resource demands associated with each human-user. The, results of this paper confirm our hypotheses and provide further insight into optimizing the server architecture, to support virtual world training.

Publication Date

1-1-2015

Publication Title

2015 Fall Simulation Interoperability Workshop, SIW 2015

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

Socpus ID

84964005610 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84964005610

This document is currently not available here.

Share

COinS