Mesh low power sensor

[Artur]

Hi all,

I am trying make some temperature sensor with Segger and nRF52 based on Mesh BLE v4.0.

Currently I am researching documentation and examples to pick best solution. 

In simplest case my network would consist of two nodes:

A) Main powered client communicating through some UART with my smart home

B) Temperature sensor which sends data in intervals (battery power would be great)

So my questions for now:

1) Which example is best for understanding friend software?

2) Would I pick mesh LP Node or non mesh beacon for sensor solution? (or something else)

Best regards, 

Artur

[Hung Bui]

Hi Artur, 

We do have the friend node example which is integrated in the light switch server. You can use the light switch server to test with the Low Power Node. 

Could you described your application a little bit more ? What is the scale of the application, how much data you want to send ? Is there any real time requirement for the data ? 

[Artur]

We are now making some simple prototype and we are going to scale app in future. We want to send at least 2 bytes (just temperature value), is there specific generic type for temperature (like binary is Generic OnOff)?

We do not need strict real time, just send new readout circa every 5 minutes. 

I probably also need some timer to wake-up LPNode for sensor readout before sending frame?

[Hung Bui]

Could you describe your application, I want to know about the number of nodes, what would be the backbone of the network. How many sensors vs the number of collectors ? 

There is a model for temperature but it's mainly for light temperature. Please refer to the Mesh Model Specification: https://www.bluetooth.com/specifications/mesh-specifications/

You can define your own vendor specific model if needed. 

You can use timer to trigger a sensor read out and send data. 

[Artur]

I would say 4 low power nodes (temperature sensors) with one friend node collecting data. Friend node would be some kind of UART proxy and would send data (got wireless from low power nodes) through UART to network master (eg. Raspberry Pi).

[Hung Bui]

My question is do you really need mesh in this case ? If the friend node connect directly to UART then I don't see any need of making a mesh network here. You can do much easier with lower power consumption with normal Bluetooth Low Energy instead of mesh. Your collector can be a BLE central and the sensor are BLE peripheral. The peripheral wake up occasionally, establish a connection, transmit data then sleep. We support up to 20 peripheral per central. 

[Artur]

We think that mesh is good solution in the case of scaling application in future. It is distributed system so we do not need strict master in this topology. Could you shortly describe what in your opinion are best advantages of MESH instead of BLE?

[Hung Bui]

Mesh actually running on top of BLE MAC and PHY layer. The biggest advantage of mesh to BLE as you mentioned is the scale. If you are making a network with more than a dozens of backbone nodes (collector, relayer) that covers a large area a normal BLE network can't cover (more than 50m for example), then mesh would be better option. But if you want to focus on ease of use and low power consumption, then BLE is better.

Bluetooth Mesh also a better solution if the network is dynamic, the nodes move around in the network and don't have a dedicated node that it has to connect to.