Hello, I’m trying to evaluate UWB for peer to peer communication. My primary need is only communication and not RTLS. In this use case two devices should paired without user intervention or without the need of any other technology like NFC or QR code etc.
Can you please let me know if this can be achieved using UWB?
Also, how much power is needed for UWB circuity, probably a ballpark no?
What will be the complexity of integrating the UWB to linux host?
Please share your thoughts?
Yes it can be done but it’s not what the parts are intended for.
The maximum packet size is fairly small and there is no automatic way to handle missing packets. And there will be some missed packets.
So unless you are OK with some data loss you will need to implement your own acknowledgment and retransmission protocol on top of the UWB link.
The parts natively use SPI, generally to integrate with linux you’d be looking at a small microcontroller to handle the UWB and act as a UWB to uart interface. A DWMx001 is in effect this and so would be the logical off the shelf hardware but in terms of firmware you’d be looking at writing a fair amount yourself.
edit- with regard to power, it depends on things like the duty cycle. Receive is the highest power mode so the more time you spend all the time listening for packets the higher the power usage will be. And since you will probably need a small processor in there that will also impact power draw. Basically it’s too complex to give a meaningful answer without far more details about the implementation and usage. The data sheet gives power for each mode that the chip can be in, you can use that to estimate what the power will be.
1 Like
Thank Andy for your response!
I understand that packet loss depends upon multipath propagation, interference and distance etc. But in comparison to BLE, will packet loss be on the lower side or higher? In my use case, two devices will be in close proximity around 15-20 cm, so I’m anticipating that multipath propagation will be lesser.
Thanks for letting me know that I’ll be needing a micro to talk to UWB chip over SPI.
I’ll look into the datasheet of an eval board to get an idea of power consumed, just to get a ball park no.
BLE automatically handles the packet loss and retransmission. Or at least does to a certain extent. At short ranges like that the packet loss should be fairly low but you should never assume it’ll be zero.
To give you an idea we will normally see >95 of range attempts succeed (which implies 4 packets without an issue), over 99% is common, implying less than 1 in 400 packets getting lost. But in some poor environments this can drop to 75%. Also these are short packets with minimal data, longer ones containing more data will have more chances for errors.
If the loss is low enough you could possibly get away with simply sending everything twice. However if there is an interference source then they tend to be bursty, you’ll lose a couple of packets in a row and then it all comes back again. So if you went with that approach you’d probably want to leave a short gap between the repeats.This approach is far simpler but higher power and less robust than the a full two way transmit / acknowledge type approach.
Thank you Andy, this is so helpful.
If I have fully understood the math, you’re saying it’s 95% success in good environment and probably 75% in case of interference. Is this correct
In a poor environment 75% of ranges work. Each range requires 4 messages without error.
Assuming all things are symmetrical that makes the probability of a single message getting through 0.75^0.25 = 0.93.
So around 93% of individual messages make it.
Where as 95% of ranges would imply 98.7% of individual messages.
Thank you, Andy! Appreciate it!