Flexible and Scalable Results Collecting in Distributed Spatial Simulations

Among many distributed spatial simulation systems each has its own approach to the problem of results collecting and analysis.The volume of results can be huge, while not all results are finally needed.The presented solution is to provide a unified form of defining the range of data to be collected and the methods for efficiently collecting them during the simulation runtime.Simulation results can be represented as a stream of records, where every record has the same structure.This observation means, that simulation can specify one or more data schema, equivalent of the CREAT ET ABLE command in an SQL database.Then data selection and analysis comes down to writing proper SELECT statements.The paper describes three main parts of the proposed, SQL-inspired results collecting method: parsing and query analysis, distributed computing and integrating all parts together.The method has been integrated with the HiP U T S, a distributed urban traffic simulator.