TrajEdge: An Efficient and Lightweight Trajectory Data Analysis Framework in Edge Environment

Introduction

This repository holds source code for the paper "TrajEdge: An Efficient and Lightweight Trajectory Data Analysis Framework in Edge Environment".

Environment Preparation

• Java 11
• Docker
• Docker Compose
• CentOS 7.0

To set a edge simulated environment, you need to install Docker first, then configure the docker cluster:

cd TrajEdge
docker-compose up -d 

Then you will get a cluster of storm in docker container.

Next you can use tcconfig to set the network latency and bandwidth between docker container for better simulated the edge-to-edge and edge-to-cloud communication.

tcset eth0 --delay 100ms --rate 100Kbps --network 192.168.0.10

BTW, we also provide detailed scripts for configuring the network by running:

chmod +x ./scripts/configureNetwork.sh & ./scripts/configureNetwork.sh

Last, you need to run python ./scripts/allocateNode.py for mapping the external node into edge device.

Datasets Description

We use 2 publicly available real-world trajectory and road map data, which can be obtained from Geolife, T-Drive. And the synthetic dataset of Oldenburg can be generated in LINK.

Don't forget to upload your data to all docker container in cluster by using:

chmod +x ./scripts/copyData.sh & ./scripts/copyData.sh

Usage

1. Prepare your trajectory data like below:

   [Lat] [Lng] [TimeStamp]
   

   Each column is separated by a blank.

2. Modify your data loader class in directory Spout

3. Pack the project into fat jar TrajEdge-storage.jar and TrajEdge-stream.jar using Maven

4. Set up the storage cluster of TrajEdge

   chmod +x ./scripts/deployStorage.sh & ./scripts/deployStorage.sh

5. Copy the jar to docker and run the topology TrajectoryUploadTopology to store trajectory data

   docker cp ./TrajEdge-stream/target/TrajEdge-stream-1.0-SNAPSHOT.jar nimbus:/opt/storm/lib/
   
   docker exec -it nimbus storm jar /opt/storm/lib/TrajEdge-stream-1.0-SNAPSHOT.jar org.example.TrajectoryUploadTopology test-1 true 1000

6. Run the query topology, including TrajectoryIdQueryTopology , TrajectorySpacialRangeQueryTopology, TrajectoryTimeRangeQueryTopology and TrajectorykNNQueryTopology.

   # ID temporal
   docker exec -it nimbus storm jar /opt/storm/lib/TrajEdge-stream-1.0-SNAPSHOT.jar org.example.TrajectoryIdQueryTopology 1 0 8 geolife
   # space range
   docker exec -it nimbus storm jar /opt/storm/lib/TrajEdge-stream-1.0-SNAPSHOT.jar org.example.TrajectorySpacialRangeQueryTopology 1 true 1 geolife
   # space time range
   docker exec -it nimbus storm jar /opt/storm/lib/TrajEdge-stream-1.0-SNAPSHOT.jar org.example.TrajectorySpacialTemporalRangeQueryTopology 1 8 geolife
   # k-nn
   docker exec -it nimbus storm jar /opt/storm/lib/TrajEdge-stream-1.0-SNAPSHOT.jar org.example.TrajectorykNNQueryTopology 1 true 8 geolife 1

Citation

If you use our code for research work, please cite our paper as below:

@article{,
  title={TrajEdge: An Efficient and Lightweight Trajectory Data Analysis Framework in Edge Environment},
  author={Changhao He, Ziquan Fang, Linsen Li, Yunjun Gao},
  year={2025},
}