LionTao / misty

Code repository for paper "Misty: Microservice-based Streaming Trajectory Similarity Search"

Misty: Microservice-based Streaming Trajectory Similarity Search

UPDATE: spark-based solutio was uploaded at spark folder

Overview

Implementation code for our under-review paper "Misty: Microservice-based Streaming Trajectory Similarity Search".

Misty is a microservice-based real-time processing framework for streaming trajectory similarity search query. The entire framework is developed based on Dapr. As a future direction, we are working on Misty's extension to spatial join and query plan optimization for higher throughput and more elastic index scaling.

Quick start

conda env create -f env.yml
conda activate misty
python tests/test.py

Dataset

We use a selected subset of T-Drive dataset presented in data/filtered folder. The preprocessing code can be found in preprocess.ipynb

Module list

Module name in framework	Folder name
Point Stream	ingress
Assembler	assembler
Index	index
Coordinator	coordinator
Executor	executor
Qyery Agent	agent