RF time of flight

[wildnwackywill10]

Hi,

I've been trying to build some prototype indoor positioning circuit systems. I've read about some ultrasonic positioning circuits which measure the delta between an ultrasonic signal and an RF signal - these seem to be remarkably accurate however the inability of the ultrasonic signal to tolerate obstacles is an issue for the application I have in mind.

The biggest difficulty of rf time of flight is the synchronisation between transmitter and receiver. however I've had an idea for what I hope would be an answer to this issue.

if a moving robot were to transmit an rf signal and a stationary node were to reflect this signal at a different frequency then if the robot could detect this signal if would be able to tell the time difference between sending and receiving the signal. The time difference would be made up of two component 1. The time delta due to the stationery node (time between receiving and reflecting the signal) and 2. the time due to the signal having to travel the distance. The time delta due to the reflector node should in theory be constant and the time to travel through the air should be variable depending on the distance from the reflector node.

There should be no synchronisation errors due to both sending and receiving being down by the same device - provided that the 'processing time' by the stationary node stays constant this should be able to measure distance

Please let me know what you all think of this idea

Thanks,
Will

 

[Ylli]

Sounds like you are talking about something similar to a radar altimeter. Years ago when I worked in avionics...
A radar altimeter transmits a signal that is being swept in frequency by a sawtooth waveform. Let's say the frequency sweeps 0.5 Hz/ns. The signal is reflected from the ground (or other object) and received by the altimeter where it is mixed with the current transmitting frequency. The output of the mix will be a frequency proportional to the time it takes the RF to reach the ground and return - propagation speed of RF in free space is about 1 foot/ns.

So if the ground is 50 feet away, the signal will be delayed 100 ns. At 0.5 Hz/ns, the beat frequency will be 50 Hz. Count the frequency and display.

 

[Nanren888]

One ns a foot, there abouts. c = 3E8
That constant time delay to turn round rx to tx has to be accurate if you want accurate results. Two get twice the path length, of course.
Passive reflector? Corner reflector = no delay.
Sometimes as Ylli says a chirp, sweep can be workable and the differenct frequency is proportional to the range. Some times easier to measure frequency than time, accrately. Takes some bandwidth though, so the RF compnonents get a little more difficult than CW. If your object has any speed, Doppler will figure.
You also might want to think about that the waveform you send has duration.You detect, some threshold of the waveform for arrival and the second arrival. In any practical channel, this is affected by noise, so you get another variable delay.
Multiple path, scattering can also get in your way.
UWB is popular these days for indoor localisation: chips available, but the RF circuitry for ultra wide band can be tricky, not to mention antennas.
BLE beacons also have been used with various degrees of accuracy. Take these carefuilly, as there are a few that are not very clever and will never be accurate.
Not sure what you want to go round, but lidar or simple optical ranging is getting cheaper.

 

[WHONOES]

Do what they do in some radar systems. In the responding device, set a time before transmit period which remains constant regardless of what ever processing is taking place. Obviously the set period needs to be longer than any processing that takes place. Then when receiving the return that period is subtracted from the "turn round time" which is then divided by 2 leaving the "distance to object" period which should the distance simple to calculate.
You must bear in mind that the RF pulse length will determine the minimum distance measurable.