This application was designed to run on a throwable, cuddly, computer (the WebSphere sphere):
This sample shows two different ways of making sensor readings from a device available in a web page:
- Run an application server on the embedded device itself
- Run an application server in the cloud, and publish sensor readings to it by messages
The second is the more traditional IoT model, but both are interesting for different reasons. The same web application can be run in the cloud and on a low-cost single board computer. The application uses a range of Java EE programming models:
- Websockets
- JPA
- EJB startup singleton and timer beans
- Servlets
- JSPs
In the case where the application server is running in the cloud, a small Java application regularly publishes messages to a configured MQTT broker.
This application is designed to run on a pcDuino (and optionally Bluemix), and runs on WebSphere Liberty.
This project has not been optimised for efficiency. It is designed to demonstrate usage of a range of Java EE programming models, and show how a modular runtime can make the Java EE programming models a great option, even on resource-constrained environments like a pcDuino or Raspberry Pi. However, that doesn't mean that using every Java EE programming model in the same resource-constrained application is a great idea.
To set up Eclipse projects, run
gradle clean
gradle eclipse
Navigate to the sphere-war folder and run
gradle runServer
The application should be available on http://localhost:9080. By default it will run in emulation mode, so it will simulate pcDuino pin readings. You can control this by unsetting the EMULATE_PINS system property in the jvm.options file](liberty-usr/servers/pollServer/jvm.options).
### Deploying to a single board computer
To create a zip with the application and all dependencies (including the server), run
gradle packageServer
To create a war, run gradle from the sphere-war folder.
This can then be pushed to Bluemix with
cf push -p build/libs/sphere-war.war
If the web application runs in the cloud, you will need to deploy an MQTT agent to the device. To build the MQTT agent, run gradle in the pcduino-sensors folder.
To run the MQTT client on a non-pcDuino machine (with emulated pin readings), run gradle emulate. This will use the quickstart MQTT broker, so you'll be able to see how the client works, but your Bluemix application will not have access to readings, or be able to send messages.
To use a configured IoT Platform service, navigate from your Bluemix dashboard to the IoT Platform dashboard (by clicking on the service) and either:
-
create a device with type pcduino and id 'emulated', make a note of the org and access token, and run
gradle emulate -DiotConfig=" "
-
create a config file from the device configuration provided, using the format in example.conf, and run
gradle emulate -DiotConfigFile=""
You will then be able to see simulated values for this device in the IoT dashboard for your new device.
See the Internet of Things Platform device samples for more information on using MQTT with Bluemix.
This sample uses WebSphere Liberty, the Internet of Things Platform Java client helper libraries, Eclipse Paho MQTT client, Java EE interfaces, and the [webjars] bundles of the Bootstrap UI framework and C3 graphing library.
