Bluetooth Range and number of data points

[vkadal]

I need to collect data from approximately 20000 points. I want to form a BLE mesh network and collect the data. The sensors may be buried below the ground level about 30 to 50 CM deep, and the sensor will be kept in side a concrete / brickwork / rigid PVC / steel chamber. ( The selection of suitable chamber is in our choice) The next data point will be within 5 meters. 

Can your Bluetooth devices work under this condition?

[Daniel Chisholm]

Interesting question.  We need to figure out if the nodes can talk to each other, and if so, can a BLE mesh network can handle the load?

To the first part: Is it fair to say that the bluetooth chip (more importantly, its antenna) has to be with the sensor, in which case this problem statement begins with "can the bluetooth signal of a device buried up to 50cm deep travel up to 5m through the soil?"  Or can the problem be solved by having the antenna closer to the surface, ideally at or above grade level?

[vkadal]

Antenna can be placed at the top of the chamber , in other words, closer to the earth surface. Then it may need travel through 5-10 cm of soil or chamber top wall. Next Bluetooth device will be within 5-7 meters 

[Edvin]

Hello,

I believe you need to test these conditions with your equipment. There are two concerns. Range and throughput.

1 - Range:

My intuition says that Rigid PVC is your best option. At least if this encapsulates the entire nRF (with antenna). 

5 meters in itself is not an issue, but the 50CM below the soil worries me. It depends on the soil. Remember that water absorbs RF, so if the soil is moist, this may be an issue.

2 - throughput:

You say you want to collect data from 20 000 points. Does that mean 20 000 unique devices, which will report in the same network? If that is the case, how often will they report? Bluetooth Mesh is not a great protocol for high throughput. And the overall throughput goes down when the number of nodes in the network increases.

Bluetooth mesh works well with a lot of "passive" devices, such as light bulbs that listen for a message that is intended for many devices. The reason for this is that all nodes will repeat all messages several times, so the message is flooded with that message, to make sure that the message covers the entire network. This is not great if you have many (!) devices that report their own custom message. It means that the message from all devices will be relayed (repeated) at least once by all other nodes. Comparing these scenarios, a light switch message will be repeated typically 3 times by all nodes. If you have 10 000 devices, that means 30 000 messages going through the network. If all 10 000 nodes send out their own message, each of these will be repeated 30 000 times. 10 000 * 30 000 messages = 300 000 000 messages. 

Of course, you can limit the number of retransmissions, and turn off the relay functionality in some nodes, but that also impacts the coverage of the network.

Perhaps you can elaborate the throughput that you need. Will 20 000 devices send their own messages? How often?

BR,

Edvin

[vkadal]

Thanks for the detailed reply.  As you suggest, I can use a rigid plastic enclosure and avoid moisture. So the range should be as required. The data rate is very slow. May be, one or two data per hour per device. And I can reduce the number of data points to reduce the congestion. That will not be an  issue. 

Can you suggest me the right part number to do a POC? I have nRF 9160 DK with me 

[Edvin]

For Bluetooth Mesh I would recommend the nRF52833 or the nRF52840, depending on your application size. It is easy to port the application between the two, so you should start with either of those DKs. 

When you have an application that is working, you can port it to the nRF52840 Dongle if you like (less expensive and smaller, which can be used for a larger scale of testing).

Best regards,

Edvin