8

I have a requirement to design M2M communication devices for a factory which has 1000 nodes (machines) within 100m of each other. The devices will be wired to get power and will transmit a payload of 8Kbps.

What is the most suitable wireless technology for this? I have considered LoRa (433 MHz), BLE, Zigbee, Wi-Fi HaLow for this upto now with each having pros and cons.

Edited: Each node will send 1 Kilo-Byte data packet every second. Decision will not be constrained by the investment.

  • 1
    Do you have to think about the server, database, etc, or just the nodes? – Mawg Aug 30 '18 at 8:32
  • 1
    @Mawg: No significant constraints on that investment. Technology is the only consideration. – SamHW17 Aug 30 '18 at 10:25
  • 3
    Just double checking, are we in a 100m radius/sides of a square? If so does the solution have to be wireless? Given you are already stringing power could you not send data over the same wire e.g. PoE – hardillb Aug 30 '18 at 13:20
  • 1
    I've moved some comments to chat so the requests for clarification don't get buried and the discussions can be continued more easily there. Just a reminder to all that comments here are intended only to suggest improvements or request more information; feel free to post your solutions below as answers. You can also edit in any relevant updates from the comments to the question. – Aurora0001 Aug 30 '18 at 19:55
8

Your challenge is handling the throughput demands with channel contention. Your devices are all close enough that you need to assume they are all potential interferers to each other so you will need to use a protocol that is robust to this - probably using some sort of coordination between devices to ensure they all take a fair slice of the channel.

You have a payload of 1 Mbps, but also need to factor in the link overhead. A protocol using spreading codes or multiple channels will permit more concurrent connections. As a starting point, I think you want to be looking for a technology which peaks at closer to 100 MBps at a minimum.

You also need to consider the impact of lost data. Do you want to abandon or retry (abandon is much more efficient), particularly if there are other sources of interference.

This is a bit incomplete as an answer, but I wanted to flag up that I think this is a relatively hard problem to solve right. I suggest you don't accept this answer unless I dredge up a lot more detail.

  • 1
    you need to assume they are all potential interferers to each other so you will need to use a protocol that is robust to this - maybe a mesh network? Something with store & forward, like DTN? Btw, @SamHW17 is it critical that the data arrive in total realtime, or could you tolerate a small delay? – Mawg Aug 30 '18 at 12:30
  • 2
    No, something that is more intelligent than 'everyone transmit at a random time offset', or 'there is 10% time window in the frame for clients to request an access slot'. – Sean Houlihane Aug 30 '18 at 12:33
  • 2
    @Mawg tangential reference patents.google.com/patent/CA2373270C/en – Sean Houlihane Aug 30 '18 at 12:36
  • 2
    @Mawg No, the payload will contain a timestamp so latency would not be a big problem. I was also looking at mesh network. @ Sean: Isn't it 1Kbps? Not 1Mbps? Will be abandoning the lost data; data loss will be minimised because the payload will contain 3 rolling data sets. – SamHW17 Aug 30 '18 at 12:40
  • 1
    @SamHW17 1k times 1000 endpoints. So you also need a protocol which can be configured to be lossy/fire-and-forget. Never designed with a mesh, but I'm dubious unless you use very short range devices. – Sean Houlihane Aug 30 '18 at 12:56
3

Your best bet is DASH7.

It's pretty unknown even among nerds, but if you want a solid industrial stack, DASH7 it is. In case you haven't heard of DASH7 before, it's the official IoT standard of all NATO member countries and the Department of Defense:

DASH7 has been mandated by the U.S. Department of Defense (DoD) and NATO alliance.[1]

The beauty of DASH7 is that it specifies all layers of the OSI model, which is really a relief in the current chaos surrounding IoT protocols. There's no fuss about whether some layer stitches well with the protocol you've chosen for the layer below, etc. It is also not in the heavily populated 2.4 GHz ISM band, but can work on 433 MHz, for example.

And since it is backed by such powerful organizations, you can expect it to be here for a long time. Furthermore, the stack is derived from ISO/IEC 18000-7 (hence the name "DASH7") and the full specification is free for download.

DASH7 was, by the way, used in the Gulf War for logistics. There are some development boards available and it is also targeted towards Texas Instrument's MSP430.

  • 2
    Will it support the traffic density at 433 MHz? – Sean Houlihane Sep 2 '18 at 17:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.