Crowd Simulation as a Service: A scalable, real-time architecture.

Abstract

In this thesis on real-time crowd simulation as a service, we explore how crowd simulation applications can benefi?t from cloud computing. The aim of this thesis is to provide a candidate architecture for real-time cloud-based simulation software that is scalable and highly performant. We also aim to document our design process to show how and why decisions were made. First, we introduce the subject by motivating the need for cloud computing in simulation software. We then provide an overview of related work about both crowd simulation and computer and cloud architecture. Next, we describe the problems we are trying to solve and we list a collection of requirements that will help guide our design process. We then combine our knowledge into a candidate architecture for real-time Modeling and Simulation as a Service (MSaaS). After presenting our candidate architecture we perform research into the performance sensitive parts of MSaaS and discuss how we can apply our results.