harryprince / GeoSparkSim

Scalable Microscopic Traffic Simulation

GeoSparkSim is a scalable microscopic traffic simulator, which extends Apache Spark to generate large-scale road network traffic data and help data scientists to simulate, analyze and visualize large-scale traffic data. GeoSparkSim converts road networks into Spark graphs, simulates vehicles to Vehicle Resilient Distributed Datasets (VehicleRDDs) and provides a simulation-aware vehicle partitioning method to parallelize simulation steps, balance workload and handle the dynamic spatial distribution.

It is mainly developed by the contributors from Data Systems Lab

Installation

• Download GeoSparkSim repository or run command

  git clone https://github.com/zishanfu/GeoSparkSim.git

• Run cd GeoSparkSim

• Run mvn clean install under GeoSparkSim folder

Get Start

• Run command java -cp target/GeoSparkSim-1.0-SNAPSHOT-jar-with-dependencies.jar com.zishanfu.geosparksim.GeoSparkSim -h

• Run GeoSparkSim with -o to show the user interface

• Run GeoSparkSim in standalone mode

  ./spark-folder/bin/spark-submit
  --class com.zishanfu.geosparksim.GeoSparkSim
  target/GeoSparkSim-1.0-SNAPSHOT-jar-with-dependencies.jar

• Run GeoSparkSim in distributed mode

  ./spark-folder/bin/spark-submit
  --master <master-url>
  --class com.zishanfu.geosparksim.GeoSparkSim
  target/GeoSparkSim-1.0-SNAPSHOT-jar-with-dependencies.jar

• More details from GeoSparkSim Wiki and Demonstration Video.

References

• Building a Large-Scale Microscopic Road Network Traffic Simulator in Apache Spark Zishan Fu, Jia Yu, Mohamed Sarwat. To appear in proceedings of the IEEE International Conference on Mobile Data Management, in Hong Kong, China June 2019.
• Demonstrating GeoSparkSim: A Scalable Microscopic Road Network Traffic Simulator Based on Apache Spark Zishan Fu, Jia Yu, and Mohamed Sarwat. To appear in proceedings of the International Symposium on Spatial and SpatioTemporal Databases, in Vienna, Austria August 2019 (Demo Track). Best Demonstration Paper Award Runner-Up.
• GeoSpark: A Cluster Computing Framework for Processing Large-Scale Spatial Data[Project Website] Jia Yu, Jinxuan Wu and Mohamed Sarwat. In proceedings of the ACM International Conference on Advances in Geographic Information Systems, Seattle, WA, USA November 2015 (Short Paper)

Questions

Please contact [e-mail]