App-Driven Intelligence with IoT data

Solution Overview

This solution uses MongoDB's multi-cloud database platform to analyze IoT data from a rocket launch. It uses an open-source data set from a Blue Origin rocket launch, along with fictional data.

A typical rocket launch spans an eight-hour period from when the initial countdown begins to when the rocket payload is in orbit. During this window, approximately one million datapoints are generated per second by sensors that capture the rocket's performance.

While the metrics from the rocket make up the bulk of the data in this use case, there are two other sources of data:

• Notes are created by rocket engineers and an automated system. Rocket engineers create notes when they want to mark anything to remember after launch. Additionally, an automated system watches metrics as they stream in and creates notes whenever values are outside of their expected threshold.

• Weather data is retrieved from a third party, stored in an Amazon S3 bucket, and analyzed in combination with the launch data post-launch.

Reference Architectures

The following diagram displays this solution's architecture. MongoDB's flexible document model and Atlas tools streamline how you can store and interact with your data. Real-time sensor data, engineer observations, and real-time weather data can all be combined in MongoDB Atlas.

Figure 1. Reference architecture with MongoDB

Data Model Approach

This solution stores launch data in two Atlas collections: launchData and notes.

Initially, the rockets produce metrics as 4-element tuples with the following structure: (device, timestamp, metric, value). These values are then aggregated and written to the launchData collection as a single document that represents metrics from a specific device at a specific time, like in the following example document:

{
   _id: ObjectId("62f2f8b5800b621ee724bb94"),
   time: ISODate("2020-10-13T13:33:30.219Z"),
   meta: { device: 'truth' },
   TIME_NANOSECONDS_TAI: Long("1602596010219040000"),
   truth_pos_CON_ECEF_ECEF_M2: -5268929.31643981,
   truth_pos_CON_ECEF_ECEF_M1: -1387897.36558835,
   truth_pos_CON_ECEF_ECEF_M3: 3306577.65409484,
   truth_vel_CON_ECEF_ECEF_MpS2: -0.00810950305119273,
   truth_vel_CON_ECEF_ECEF_MpS3: 0.00414972080992211,
   truth_quat_CON2ECEF1: -0.458400879273711,
   truth_quat_CON2ECEF2: -0.176758395646534,
   truth_quat_CON2ECEF3: 0.511475024782863,
   truth_vel_CON_ECEF_ECEF_MpS1: 0.00220006484335229,
   truth_quat_CON2ECEF4: 0.7049953208872
}

This document model also works with MongoDB time series collections.

This data can be found in the aerospace.archive.gz file in the solution GitHub repository.

Build the Solution

The solution's GitHub repository provides the data and code necessary for you to get started. Additionally, the repository includes links to a three-part livestream recording that goes over this solution in detail.

To get started with this solution, perform the following steps.

Key Learnings

MongoDB's flexible database platform enables application-driven IoT analytics by providing the following capabilities:

• Modeling data across multiple data types: MongoDB's flexible document model allows you to store multiple types of data in one document, collection, or database, streamlining how you interact with your data and reducing the need for different tables.

• Integrating functions to react to real-time data: MongoDB's fast query capability allows you to analyze data in real-time.

• Minimizing ETL processes with in-place aggregation: MongoDB's aggregation pipelines enable you to transform your data into computed results and store them in a new collection.

Authors

• Jay Runkel, MongoDB

Learn More

• Build an IoT Data Hub for Smart Manufacturing

• Power Smart Meter Analysis with MongoDB

• Real-Time Audio-Based AI Diagnostics