Ratio of mesh-only (Xenon) to internet-enabled boards (Argon, Boron)

The first use case I thought of would be smart badges at a conference where there could be 100-1000 people in a room at a time. I was trying to determine costs when every Nth badge needs to have Wifi or LTE to make the thing work, then saw “making networks of 50+ devices possible” and got bummed out. If the range of the mesh radio is “tens of meters” you could have several hundred people within that radius at a conference.

Hopefully this idea is still viable once everything is out of the lab!

Usually, there is too much interference at large conferences/press releases when you have that many devices radiating that close together.

Example:

1 Like

I personally think that memory could become an issue a some point. The way I imagine it, the network topology needs to be stored somewhere. Devices should know to which device they should talk to get the message to the gateway, rather than double-backing twenty times. At some point you’re going to run out of space to store that ‘network map’ on, I think.
That said, that’s just my uninformed speculation :speak_no_evil:

1 Like

The answer you seek will likely be here … https://openthread.io/

IIRC there is a hard limit of 250 devices per Thread PAN, but there is memory impact in maintaining state between neighbors in the mesh, etc.

2 Likes

Great discussion! We’re still working on documentation, but here’s some details to get us started:

Gateways, Repeaters & Endpoints
Particle Mesh devices can play different roles in a mesh network: gateways, repeaters and endpoints. Repeaters make up the backbone of the mesh network. They are always on and can be trusted to relay messages across the network. Endpoints are typically sensors or actuators. As oppose to repeaters, they are not required to always be on and therefore can be battery-operated and sleepy. They pair with a nearby “parent” repeater, who queue messages on their behalf when a child endpoint is asleep. Gateways are special types of repeaters that have a backhaul back to Particle’s Device Cloud. The devices we announced today can play different roles, depending on how their software is configured:

  • Argon: Gateway, Repeater or Endpoint
  • Boron LTE/2g3g: Gateway, Repeater or Endpoint
  • Xenon: Gateway (with Ethernet FeatherWing,) Repeater or Endpoint

Theoretical and Practical network sizes
At launch, we’ve developed a lot of the software needed for Particle Mesh. However, we have a lot more development to go before we ship. Some of this development will impact the total network size possible.

A repeater can support a finite number of children endpoints and therefore the total network size is dependant on the number of repeaters/gateways in a network. The total network size increases with the number of repeaters/gateways in a network. A network can support a finite number of repeaters so therefore there is a finite maximum network size. A network can support multiple gateways which allows for redundant connections to the Particle Device Cloud.

When planning your network, we recommend you first start thinking how many sensors and actuators you’re deployment might need. That will govern how many repeaters are needed for the network. Then decide how many gateways you need to guarantee connectivity.

Note: while endpoints are ideal for sensors and actuators, you can attach sensors and actuators to gateways and repeaters because they share the same capable MCU as the Xenon, which might save some cost on endpoints.

Particle Mesh & Thread
Particle Mesh is based on the Thread networking protocol. As someone noted, Particle Mesh uses OpenThread and openthread.io is the best place to learn more about the technical details of OpenThread & Thread. While you’re reading their site, note that Particle Mesh uses more consumer-friendly terms for Thread roles. For reference:

  • Particle Mesh Endpoint = Thread End Device
  • Particle Mesh Repeater = Thread Router
  • Particle Mesh Gateway = Thread Border Router

Thread networks support a theoretical limit of 32 routers, each with 512 children. Therefore, a single Thread network could reach 16,384 devices. However, practical limits like memory on device will impact real-world numbers. As a baseline, the Thread Group demonstrated a network of 250 devices in a lab environment. We’ll improve our software over time and in the months before we ship but we’re haven’t finalized our evals. We intend to support (at minimum) a network of 1-2 gateways & 3-5 repeaters - with 10 endpoints each. That brings the total to 55ish devices in a network for this first release.

That’s a lot for now, phew! Hope this helps! Keep the questions coming :slight_smile:

11 Likes

@jberi

Basic question: What do you think is going to be a range of x in meters (whatever unit, just label it!) between Xenons.

I have used the Wifi Photon outside of my house and know that it’s Wifi connection maximum range can some days be only about 20 meters (~65 feet), others days I can walk a fair distance lets say 60 meters (~200 feet) from my house and still have a solid Wifi connection without the duck Antennae.

Other users have posted between 75 feet and 400 feet with Antennae Particle Photon Max WiFi Range which seems comparable, remembering that Antennae use and Antennae settings come into play.

I am interested in X (blue arrow) in the diagram below, for the obvious reason of it would influence how many devices would be able to monitor a farmers field for various sensor readings and actuator settings.

I am guessing 10 meters (~30 feet) would be a working reliable value for X, and the multiple rows should increase the reliability. Any opinions? This would be a good experiment when the new devices come out. If the actually number is only 3 meters (10 feet) for reliable mesh operation that would be very valuable for developers to know.

In the image below the green boxes are Xenons (Mesh and Bluetooth only) and the Red boxes are Argons (Wifi, Mesh and Bluetooth).

The Bluetooth classes define the range for this, I believe. The datasheet shows Bluetooth classes 1, 2, and 3 for this chip. Class 1 is 100 meters, although with other products I’ve consistently gotten about 50-60 reliably.

Hard telling at this point but the Bluetooth 5 and its long-range lower bandwidth is said to be capibable of up to 4 times longer range.

See this from the Nordic website about BLE 5 https://www.nordicsemi.com/eng/Products/Bluetooth-5

Being able to switch the data rate and increase the X distance would be very useful. Looks like people (@Mjones ) are thinking the X value in my diagram (a few posts above) might be able to be around 50 - 60 meters (~160 - 200 feet) . That would be incredible and really allow industries to implement IOT without any major infrastructure changes.

Unfortunately, I am a bit of a pessimist about future tech, so I really need to wait and see the new devices in action. Anyone else got opinions about a useable X value for Xenons in the above diagram.

1 Like

While we don’t have validated test results to share yet, I’m fairly confident you should be able to achieve 10 meters or more for a point-to-point connection over Thread. For a point of reference, Nest’s latest product Nest Secure is a security system that uses Thread for communication between the motion sensors (Detect) and control pad (Guard.) According to their FAQs Detects can be up to 50 feet away from a Guard. And that’s in indoor conditions, with wires and metal in the walls and your use case appears to be more ideal.

Based on your diagram it looks like the green boxes are all acting as endpoints but it is worth calling out that if some of those were repeaters (solar?,) the effective range from one corner of the network to the other would increase by taking advantage of the mesh.

2 Likes

Looking at the specs for Bluetooth 5, class 1 range could be 800-1200ft. Apparently they’ve changed quite a bit since v4. However, we all know real world applications are different. Even if we get half of that, which is pretty low in my opinion, 4-600ft would be awesome.

For clarity, the Mesh uses the 802.15.4 radio on the nRF52840, not the Bluetooth 5 part. Range for 802.15.4 is as I quoted above.

2 Likes

What if in my diagram all the Xenon’s were repeaters? 24/7 powering them is not a problem. Repeaters can still have some sensors, correct? By all of them being repeaters does that strengthen the mesh or overload it?

Yup! Repeaters can be identical to an endpoint, except it has to be always on, so they can have sensors or actuators, and even run the same application code. You could modify your diagram where some green Xenons were Repeaters and some Endpoints but all have the sensors you’re interested in. Than, the total distance would be ~50 ft between each device + hoping between repeaters, plus the total number of daisy-chained repeaters to get to your effective range.

Does that help? Note I explained Repeater and Endpoint limits above but based on your diagram, your network size should be do-able.

3 Likes

Ahh, so your not using Bluetooth for the mesh network.

How does the power consumption levels with 802.15.4 compare to BLE5? Any ideas?

I know Wifi is much more power hungry compared to BLE so figure 802.15.4 may be power hungry also but not sure.

Found this on Nordics forum about power consumption:

https://devzone.nordicsemi.com/f/nordic-q-a/19589/nrf52840-max-current-consumption

1 Like

So I understand the mesh is designed to use the 802.15.4 radio, would it be possible to talk with two BLE units to achieve larger range?

@jberi Thanks for all the clarifications, especially about the Mesh using the 802.15.4 radio, I also was confused and thought the Mesh was powered by Bluetooth. Really excited about the new devices.

Here is an updated farm image with Xenon repeaters and Xenon endpoints (purple blocks). I would test really slow sensor readings say every 15 minutes (staggered so they read at different times) just soil moisture and temperature. Then motion sensors on trigger.

I also added a few Borons for cellular redundancy, fun to see if that would work as a Wifi backup.

So if the Xenon has a connection radius of about 15 meters (50 feet) the next question is, how long could the field be? My diagram shows a field with a length y of about 60 meters (200 feet) using about 40 devices, but only 9 Xenon repeaters. If the connection radius x is greater than 15 meters then y obviously could be relatively longer.

Really curious about how far away from the original Wifi a situation like this could achieve and still be reasonably stable sending data to the cloud.

2 Likes

Where “always” is to be taken with a grain of salt - as long there are enough “spare” repeaters to bridge the gaps any redundant device can “disappear” without breaking the mesh but causing some “rerouting effort” .

1 Like

Thanks for helping to clarify, @ScruffR! I should also clarify that when I said “always on” I meant, “line powered (or really big battery) and could always be on.” In practice, repeaters won’t be transmitting/receiving constantly even in fairly chatty network but will draw more current than endpoints because they’re relaying messages across the network, must keep their transceiver on and can’t go into deep sleep.

2 Likes

With the estimates we have so far, the diagram should be achievable.

1 Like