Synchronisation of transmitter and receiver for direction finding.

[cheezballs]

Hi there,

I have a rudimentary understanding of programming with the nRF chipsets and Segger.

In our project, we are trying to locate the position of a transmitter (nRF52833 DK with external antenna) relative to a receiver (nRF52811 based receiver with antennae array). 

We are finding that our azimuth angle outputs are quite inconsistent, they seem to be accurate for a few samples, and then very inaccurate. We believe this could be due to multipath/signal reflection in the testing environment. One way we hope to mitigate the effects of multipath/signal reflection is to synchronise the transmitter and receiver. 

We have had the following two ideas for synchronisation:

• Transmit based on a shared clock 
• Have a calibration phase on the receiver, whereby the transmitted signal is analysed for the transmission (pulse) interval. This is under the assumption that the initial transmitted signal has the highest RSSI, and the subsequent multipath/reflection signals have relatively low RSSI.

These resulted in the following questions:

• Is there a clock library available which would allow for intermittent, synchronised transmissions between the transmitter and receiver?
• Or is there a way to emulate a spectrum analyser on the receiver in order to interpret the interval between transmission pulses from the transmitter?

If neither of these is possible, are there other alternatives for wirelessly synchronising transmitter and receiver devices?

Also, if you have any ideas as to why the angle outputs could be inconsistent, please let us know.

Thanks!

These

[Dmitry]

Hi,

are you making AoA or AoD system? Is it possible to test it outdoors, without any reflecting surfaces nearby? Multipath is a most likely cause, but the result shouldn't change in static environment (e.g. no people walking near antennas). For AoA system, multipath reflections from walls can be compensated using a multi-element array with an optimal algorithm. For AoD wearable with a single antenna, the varying reflections from human body makes the problem much more complex.

Do you have any plots or CSV of I/Q data for good and bad cases?

The method implemented in BLE5 don't require any time sync. You cannot generate a short pulse with GFSK, as the signal is passed through  filter to keep an emission within an 1-MHz bandwidth. Also, max sampling rate of radio ADC is 8 Msps - more than 30 meters a path of radio wave, it's not sufficient to detect and compensate a reflected signal.

[cheezballs]

Thanks for this, yes it is an AoA system with an array of eight isotropic whip antennae (vertically polarised) for receiving. The transmitter has a single isotropic whip antenna which is not always vertically polarised (i.e. it is moving around). 

I will try to collect data for good/bad cases. I see, this makes sense. Thanks for your help!