Maximum practical nodes
-
I've searched around for this question, but haven't really found a good answer.
I see what the maximum theoretical node count for a full network would be: 255 Nodes per Channel, 125 Channels.
But, not having much experience with RF applications, I'm curious what the practical maximum node count might be, as I'm assuming at some point the environment would just be too noisy? I dunno.
What is the largest implementation anybody has seen so far? I couldn't find any good info on this either.
I'm evaluating the NRFs for use in a network that could potentially have up to 5000 nodes in a 5000 sq ft area. Is that just crazy talk?
-
I've searched around for this question, but haven't really found a good answer.
I see what the maximum theoretical node count for a full network would be: 255 Nodes per Channel, 125 Channels.
But, not having much experience with RF applications, I'm curious what the practical maximum node count might be, as I'm assuming at some point the environment would just be too noisy? I dunno.
What is the largest implementation anybody has seen so far? I couldn't find any good info on this either.
I'm evaluating the NRFs for use in a network that could potentially have up to 5000 nodes in a 5000 sq ft area. Is that just crazy talk?
@Nicholas-Underwood it depends on how much data is sent and how often. At 250 kbit (the default speed for nrf24) sending the maximum message size (32 bytes) takes just under 1 millisecond. Rf networks usually handle about 20% utilization before degrading. So we can send for 200ms out of every second.
That means about 200 messages per second, so in a 254 node network each node should be able to report almost as often as once per second before the network starts to degrade.
If the messages are shorter than the maximum, even more messages can be handled.
According to the nrf24l01+ datasheet, no channels overlap when using 250kbit. So you can use all channels at the same time. That means 254*126=32,004 nodes. (There are 126 channels, not 125. The gateway will use id 0 so 1-254 can be used for "regular" nodes.)
If you need more, increasing the data rate to 1mbit will give you 4x capacity (at the expense of range). The channels still don't overlap. Going to 2mbit won't help you though, since 2mbit uses 2MHz wide channel so the channels will overlap.
This is just a back of the envelope calculation though, based on the information in the datasheet. There might be other bottlenecks. I am not sure gateways and controllers will keep up.
-
Thanks @mfalkvidd
The good news for me is that speed is not that important, so with these numbers, even 250kbit should be sufficient. Only a small percentage of the nodes will need to report data at any time back to the gateways, with a very small payload, only a few bytes. The bulk of the data would be sent to the nodes. Also typically a small percentage of the entire network at one time, but potentially occasionally to a large portion of the network, with roughly 12 byte payloads.
Thanks again!
-
Thanks @mfalkvidd
The good news for me is that speed is not that important, so with these numbers, even 250kbit should be sufficient. Only a small percentage of the nodes will need to report data at any time back to the gateways, with a very small payload, only a few bytes. The bulk of the data would be sent to the nodes. Also typically a small percentage of the entire network at one time, but potentially occasionally to a large portion of the network, with roughly 12 byte payloads.
Thanks again!
@Nicholas-Underwood you're welcome. Just a note: sending data to the nodes will use just as much capacity as sending from the nodes - the radio channel is a shared medium.
If you have a lot of data that should reach all/many nodes you might benefit from sending to the broadcast address (id 255) but that would require some custom code for handling acknowledgements and selecting which nodes should act on the message.
-
@Nicholas-Underwood it depends on how much data is sent and how often. At 250 kbit (the default speed for nrf24) sending the maximum message size (32 bytes) takes just under 1 millisecond. Rf networks usually handle about 20% utilization before degrading. So we can send for 200ms out of every second.
That means about 200 messages per second, so in a 254 node network each node should be able to report almost as often as once per second before the network starts to degrade.
If the messages are shorter than the maximum, even more messages can be handled.
According to the nrf24l01+ datasheet, no channels overlap when using 250kbit. So you can use all channels at the same time. That means 254*126=32,004 nodes. (There are 126 channels, not 125. The gateway will use id 0 so 1-254 can be used for "regular" nodes.)
If you need more, increasing the data rate to 1mbit will give you 4x capacity (at the expense of range). The channels still don't overlap. Going to 2mbit won't help you though, since 2mbit uses 2MHz wide channel so the channels will overlap.
This is just a back of the envelope calculation though, based on the information in the datasheet. There might be other bottlenecks. I am not sure gateways and controllers will keep up.
@mfalkvidd said:
According to the nrf24l01+ datasheet, no channels overlap when using 250kbit. So you can use all channels at the same time. That means 253*126=31,878 nodes. (There are 126 channels, not 125)
I guess you need one Gateway for each channel?
-
@mfalkvidd said:
Sounds like a cool project by the way. Can you share any information about it?
Unfortunately I don't think I should share anything just yet, but once the system architecture gets finalized and the product starts to come together, I'll definitely share then. :)
-
@mfalkvidd said:
According to the nrf24l01+ datasheet, no channels overlap when using 250kbit. So you can use all channels at the same time. That means 253*126=31,878 nodes. (There are 126 channels, not 125)
I guess you need one Gateway for each channel?
