Jetson TX2 and a data collection application swarm

Product support
1

I want to have remote devices with my ML application instences running, and collecting, data, which is stored in JSON and presisted on Local MangoDB.

I have some engineering tasks, that I need to accomplish before this setup makes sense.

Step1: being able to remotly deploy and maintain the application (the application is already docarized), hence the use of Balena.IO makes sence.

Step2: Remote device (Jetson TX2) connected to internet via 3G modam (This is the challange I am facing now, as the 3G modem is not direclty supported within the Jetson Device, so bought a extender which will interface with the 3G modem and provide a wifi hotspot to the Jetson device,) Does anyone have a better solution?

Step3: Presisting local MangoDB( already running) and a central location to store all devices inferance data, with which I need some guidence on how to approch it.

Looking forward to any insights that any one can provide.

Best regards

2

Hi, welcome to the forums!

  1. The balenaCloud platform is designed to do this, though you will probably need to adjust your Dockerfile to use one of our base images (https://www.balena.io/docs/reference/base-images/base-images/)
  2. @majorz Do you know of any good working solution for connecting a jetson tx2 via a 3g modem?
  3. are you planning to store the collected data separately on each device or in a central database? Its not clear from your description
4

Thank you for taking the time to reply.

  1. Ill look into it, thanks for the direction.

  2. I was thinking to have one central Database, where I can identify data collected from every device (Jetson TX2), and use it with an analytics application, to gnerate realtime information for the project dashboard. I am open to any architechtural suggestions, as I havent started this part at all.

5

I would recommend using AWS to host your database, they support mongoDB ( https://docs.aws.amazon.com/quickstart/latest/mongodb/overview.html ), though if you are willing to make some changes to your project you can use their excellent NoSQL offering DynamoDB ( https://aws.amazon.com/dynamodb/ ).