GPS Data client and Server implemented in node.js with Raspberry Pi GPS Data Logger Use Case

• node.js
• npm
• sqlite

The node.js package manager, npm, is recommended to install module dependencies. The server database is managed with sqlite3

For installation, refer to the instructions for your OS on the installation page of each of the dependencies.

$ git clone https://github.com/p5a0u9l/gps_db # clones the repository
$ cd gps_db
$ npm install # installs the local module dependencies

gps_db comes with a config.json file with modifiable entries for

• SERVER_IP
• PORT_NUMBER
• CLIENT_KEY_FILENAME
• DATABASE_NAME

Entering

$ node server.js

creates a new sqlite3 database using the path found in config.json and binds the server application to the configured port.

Before pushing any gps records, enter

$ node gps_client.js register

This creates a new client entry in the server's database and returns the key to the client application, which in turn, writes it to the file path found in config.json.

Different clients can be registered with the server, however, currently, only one client key file is read when running the client application. To manage two separate clients, clone the repo to two seperate directories, or two seperate machines.

Assuming you have a local file called my_gps_file.json, issuing

$ node gps_client.js consume ./my_gps_file.json

will stream the file contents to the server and store them in a gps_records table indexed by your unique client id.

The server expects messages formatted by gpsd, a common Linux/FreeBSD daemon that receives data from GPS receivers. The messages should follow the JSON format below

{
    "class":"TPV",
    "mode":3,
    "time":"2016-12-07T01:58:33.000Z",
    "lat":47.421816667,
    "lon":-122.251965000,
    "alt":11.900,
    "speed":0.005,
}

A sample history of records is found under record_stream.json

The server/client registration can be tested using

$ node gps_client.js consume ./sample_data/record_stream.json

Issuing the command

$ node gps_client.js fetch

will fetch the entire history indexed by your key and write them to the local JSON file called out in config.json.

The following walks through a typical gps_db use case, assuming the necessary dependencies are already installed.

The server application for this example is running an EC2 instance on AWS. The config.json file currently points to the server.

$ git clone https://github.com/p5a0u9l/gps_db # clones the repository
$ cd gps_db

$ node gps_client.js register # registers your key with server
$ node gps_client.js consume sample_data/record_stream01.json # push a sample set of GPS records
$ node gps_client.js consume sample_data/record_stream02.json # push a sample set of GPS records
$ node gps_client.js fetch # retrieve the entire history

$ node gps_client.js register # registers your key with server
$ node gps_client.js consume sample_data/record_stream01.json # push a sample set of GPS records
$ node gps_client.js fetch # retrieve the entire history

This mini-project implements a simple GPS data-logging system to record position, time, and velocity during my morning and evening commutes from Maple Valley, WA, to Kent, WA, and back.

The data-logging system is implemented on a Raspberry Pi 3(RPi) single-board computer, connected via a USB-TTL adapter to an Adafruit Ultimate GPS Breakout module. The module is built around an MTK3339 GPS chipset which features a high-sensitivity receiver and an onboard patch antenna. Finally, gpsd serves as a monitor and parser of incoming messages and makes them available on a local TCP/IP port for client applications.

The system is powered-up at the beginning of a trip by connecting to a battery and powered down at the end of the commute by disconnecting the power source. The 10000 mAh battery is intended for mobile recharge of smart phones and will repeatedly power the combined RPi/GPS receiver for several weeks at 1 hour per day usage. The logging software is wrapped in a systemd service that enables it to go into action on bootup and run continuously until losing power.

At the end of each day, the system is connected to a home network via a USB Wi-Fi adapter. A second, very simple systemd service is enabled, which listens for a network connection event. On connection, it synchronizes any on-board GPS logs with a remote server running on AWS.

The following image shows a trip from Maple Valley to Kent colorized by speed.

This section contains a review of the homework assignment for UW EE590 - Advanced Embedded Systems - that initiated this project.

1. The node API to sqlite3 seems slower than I anticipated. It may be the implementation, although I tried a few different strategies.

2. The asynchronous nature of node causes the server to respond before the disk write is complete.

1. After pushing the file under sample_data or files of similar size, give the server about 30 sec. to a minute to completely write the records to disk.

2. For the second issue, issuing fetch a second time will flush the server responses and write the gps records to the file.

This project could be extended first by correctly handling events on the server side such that responses are delayed until are records are written to disk.

Another extension would be developing the visualization of gps data.