Tracking a radio signal

[Mike]

Hey all,

Some of you may remember me from a few years back, I was active on 237 (and these boards) from '04 to '08. Few years later I have a little bit of a project brewing and I needed some people to vet an idea to. What better than to give some off-season guys some thinking to do?

Scenario:
Multiple beacons transmitting on 433Mhz a simple data packet:

SYNC byte
ADDR byte (unique identifier for each beacon)
TX_ID byte (random identifier for each transmission)
CHKSUM byte (to verify integrity)

So lets say we have multiple beacons transmitting in a wide area. What I need to be able to do, flying from approximately 150 feet at relatively low velocity is find a beacon. So I need to take the digital data in, find a unique beacon to choose (for now, randomly pick one) and then get as close as possible to that particular beacon; "home in on it".

Here was the original idea presented to me. Four receivers on all corners of the plane (nose, tail, each wingtip). Small plane, about three meter wingspan. However this may be expanded to a larger (8 meter+) application. Using the transmission ID byte, track the time it takes between hitting each receiver and calculate the distance away, or at a minimum the direction to go. The problem here is that I calculated the speed of a 433Mhz wave at 299,792,000 m/s. That gives us 10 nanosecond difference between hitting two receivers 3m away. I think that idea is pretty much nixed.

Next idea: a receiver antenna on the plane is to be shielded by a rotating tube with a small slit in it. By using the position of the slit when we receive a transmission from the beacon we're looking for we can find out where the beacon is transmitting from.

I've never really dealt with radio waves this directly before, so what I'm looking for is some opinions on the feasibility of my second idea, as well as what kind of antennas and shielding material would work best and avoid detuning, etc.

The project is a pretty cool one with great potential. It involves Unmanned Aerial Vehicles with a unique, necessary and possibly profitable application. If you're in the Space Coast, FL area and would like more information shoot me a PM or email: msorrenti2008@fit.edu

Thanks,
Mike Sorrenti

[jee7s]

Mike-

Your slit idea is probably more trouble than it is worth. That slit would have to be pretty large to reliably pass a signal at 400MHz. The physics of EM get funny like that.

You were pretty on track to a potential solution with your multiple antenna idea.

I can understand your intuition about the 10 nanoseconds being too fast to detect reliably. However, you don't need to deal with time, you need to deal with phase. At the carrier's frequency, 433MHz, the wavelength is about 70 centimeters. So, using the phase difference beteen your antennas, you can get an angle to target. Using several antennas in an array (preferably a good distance apart), you can triangulate the location of the transmitters.

This isn't to say that the solution is a piece of cake, but it is possible. I'd suggest looking into ARRL literature (like QEX and QST) for more details about beacon finding. Many in the amateur radio community track beacons as a hobby.

-Jeff, KG2OV

[Jared Russell]

Originally Posted by jee7s

However, you don't need to deal with time, you need to deal with phase.

Just what I was coming here to say. However, with this approach you are going to want more than two receivers to help fully determine the location of the transmitter. (see: http://en.wikipedia.org/wiki/Multilateration)

[DonRotolo]

Mike,

What you want to do is a well-known science, no need to reinvent the wheel. Ham radio operators do this all the time, there is even an international sport of RDF (Radio direction finding)

What you want is a Doppler array for RDF. Ramsey Electronics makes a kit (DDF1) that works with a computer sound card and displays the direction to the transmitter in 22 degree increments. Google "Doppler Radio Direction Finder" for loads of other kits, articles, and reference papers. There is a LOT of info out there.

Your antenna array for 433 MHz would be 4 antennas each about 8" long arranged in a square pattern about a foot on each side.

As for picking the strongest transmitter, FM receivers have an unusual characteristic known as the "Capture Effect", where only the strongest signal is demodulated, all other signals (even just a little weaker) are ignored by the receiver. The receiver controls this.

It would also be possible to use the receiver's RSSI (Received Signal Strength Indicator) to roughly determine proximity, but radio signal strength at 433 MHz is not always linear with distance, research the terms Multipath Interference and Foliage Signal Attenuation.

Your part in all this would be some software that combines what you know about the aircraft's speed, direction and location, what you know about the transmitter's identity, location and signal strength, and the flight characteristics of the aircraft, to guide the aircraft to the beacon.

Let me know if you need more information. As I said, this wheel's been invented already. You're just painting it a different color.

(Cool stuff like this is why I became a Ham in the first place)

Don

.

[Jared Russell]

Originally Posted by Mike

Next idea: a receiver antenna on the plane is to be shielded by a rotating tube with a small slit in it. By using the position of the slit when we receive a transmission from the beacon we're looking for we can find out where the beacon is transmitting from.

Although the phase-based approach that Don posted is simpler to implement and somewhat more robust to multipath, this approach would work with an omnidirectional antenna and a well-designed slit (which is admittedly hard to make).

I just want to point out, though, that there is no need for a slit if you use a directional antenna. Simply rotate the antenna; the antenna's azimuth at the peak magnitude response in a revolution should correspond to roughly the right heading. Your accuracy would be limited by the "sharpness" of the antenna's response curve, however.

[Jeff Pahl]

Originally Posted by Jared341

Although the phase-based approach that Don posted is simpler to implement and somewhat more robust to multipath, this approach would work with an omnidirectional antenna and a well-designed slit (which is admittedly hard to make).

I just want to point out, though, that there is no need for a slit if you use a directional antenna. Simply rotate the antenna; the antenna's azimuth at the peak magnitude response in a revolution should correspond to roughly the right heading. Your accuracy would be limited by the "sharpness" of the antenna's response curve, however.

At 433 MHz, to attempt to rotate a directional antenna you would end up with something that looks like an AWACS. The antenna length would be a significant percentage of your wingspan to get any sort of a reasonably narrow beamwidth.

A slit is also going to have a very wide beamwidth.

As Don suggested, a doppler system it probably going to give you your best results. It also has no moving parts, so your reliability will be higher and you will not have any issues with the rotating mass affecting the flying.

Do multiple beacons transmit at the same time? That will be a problem, as due to the capture effect you will only hear the loudest beacon, which may not be the one you want. If they transmit one at a time, then you can listen for the id of the beacon you want.

[DonRotolo]

Originally Posted by Jared341

Simply rotate the antenna

A valid approach, but you then have the issue of coupling the RF signal path through a rotating joint, not trivial.

Originally Posted by Jeff Pahl

Do multiple beacons transmit at the same time? That will be a problem, as due to the capture effect you will only hear the loudest beacon, which may not be the one you want. If they transmit one at a time, then you can listen for the id of the beacon you want.

The same thought occurred to me last night. Unless they have some kind of synchronization, you're going to get data collisions all over the place (look up Hidden Transmitter Syndrome) and in a practical application you won't hear any of the transmitters well enough to decode their data.

I asusme they are simply low powered transmitters, not equipped with enough intelligence to synchronize, and without a receiver for a DAMA-type synch protocol.

If they were laid out in a known pattern, you can use whatever you receive to know 'where you are' so you can get to 'where you need to be'. But if they are scattered randomly _or_ mobile you have to just search until you hear what you are looking for and zero in on it before you lose the signal.

Also consider how often each one pings. If the ping is 5 bytes (TX_ID is 2 bytes) you have 40 bits, at 1200 baud that's (40/1200 = 1/15) of a second per ping, add a small randomization in there (two TXs that collide every time is a problem) you can see maybe 10 pings per second, with 200 TXs you might be able to ping once every 20 seconds (at best), which is far too slow even if you fly at 30 MPH (44 fps) since you'll move (44*20=880 ft) between pings.

I want to emphasize that this entire "problem" is absolutely solvable (with college-student-level resources) but not trivial. There is plenty of info published on these topics. Again, ask whatever questions you have.

[Alan Anderson]

Originally Posted by Jared341

Your accuracy would be limited by the "sharpness" of the antenna's response curve, however.

It's often possible to arrange for a very sharp "null" in the antenna pattern at the sides. Aiming for minimum signal can be a good strategy when trying to determine the direction to a transmitter.

This obviously does not work if you're simultaneously attempting to decode the signal, but an obvious workaround is to use one receiver with an omnidirectional antenna to read the data, and another receiver on the directional antenna to determine where the signal is coming from.



Is the signal format set in stone? A wideband "spread spectrum" transmission with a separate spreading code for each transmitter would be more immune to interference between transmitters.

[Mike]

Thanks for the help guys.

Just a short info blurb:

Multiple low powered beacons with no synchronization between them laid out in a random pattern. For this proof of concept, maybe three. For the real world application, maybe thirty thousand ()... lets focus on the proof of concept first.

I'll begin looking into some kind of doppler system. I saw the ham radio guy's competitions and stuff but I can't have anything backpack size with the huge antennas like I saw in those pictures. Unfortunately a google of "doppler radio direction robotics" doesn't have much, except for some guy on a forum named Chief Delphi

Not sure what you're referring to Alan. Nothing is really set in stone. The main requirements are an ability to mass produce, cheaply and efficiently, a large amount of beacons and then be able to find them when scattered in a wide area (proof of concept: square mile; production: entire town).

[DonRotolo]

Originally Posted by Mike

Unfortunately a google of "doppler radio direction robotics"

Drop the "robotics"

The Ramsey kit is the size of a textbook (plus antennas), you need a radio receiver for the frequency of interest (you pick the size) and a power source (ditto). I could fit it all into a Cheerios box (but the antennas would poke out) with room for breakfast.

3 or 10 over a square mile as PoC is doable, but scalability will frustrate you even at the hundreds level. The only way around that is a fixed (land-based) antenna network, which doesn't sound practical here. (A friend's company does exactly this, providing equipment location within a hospital. Handy if you need that <equipment name> NOW to save a life)

Research Spread Spectrum, Alan's comment is absolutely valid and when you get past the PoC stage allows for scalability.

(Here's a thought experiment for you: Imagine instead that each device chirped an audio tone, and your 'antenna' was a microphone. How many chirpers before you can't tell one from another? Repeat with chirpers on any of 500 random frequencies (=spread spectrum). Now how many?)

By the way, the Ramsey kit is just an "antenna", you can easily connect up a spread spectrum receiver to it as well.

[Mike]

Originally Posted by Don Rotolo

Drop the "robotics"

The Ramsey kit is the size of a textbook (plus antennas), you need a radio receiver for the frequency of interest (you pick the size) and a power source (ditto). I could fit it all into a Cheerios box (but the antennas would poke out) with room for breakfast.

3 or 10 over a square mile as PoC is doable, but scalability will frustrate you even at the hundreds level. The only way around that is a fixed (land-based) antenna network, which doesn't sound practical here. (A friend's company does exactly this, providing equipment location within a hospital. Handy if you need that <equipment name> NOW to save a life)

Research Spread Spectrum, Alan's comment is absolutely valid and when you get past the PoC stage allows for scalability.

(Here's a thought experiment for you: Imagine instead that each device chirped an audio tone, and your 'antenna' was a microphone. How many chirpers before you can't tell one from another? Repeat with chirpers on any of 500 random frequencies (=spread spectrum). Now how many?)

By the way, the Ramsey kit is just an "antenna", you can easily connect up a spread spectrum receiver to it as well.

Proof of concept is all I need for now.

So using this DDF kit I will be able to point, within 22 degrees, towards the beacon with the strongest signal strength? How would I go about decoding the signal then, another directional antenna pointed in the approximate heading? What happens if there happens to be two or three beacons there? I feel like the DDF guiding the directional antenna doesn't provide enough resolution in case there are two beacons in that heading. Can I use the DDF kit to both find the direction of the radio and decode anything incoming?

Thanks
Mike

[DonRotolo]

To receive the signal, use an omnidirectional "ground plane" antenna. Very inexpensive and small.

The DDF signal varies too much to decode data from it, generally speaking. There are ways around it, but not easy ways.

Yes, you may in fact receive signals from more than one transmitter, but your flying algorithm will not simply pick up 'one' signal and fly at it, instead you'll fly past it and note where it was, then turn and fly at it. Or, said another way, you need to find the right one before you can steer towards it.

Try this on a regular map. At one spot, you get the target at a bearing of 110 degrees. At another spot, you get the target at a bearing of 150 degrees. You turn 90 degrees and get the target at a bearing of 120 degrees (all measured relative to the aircraft) again. Draw lines to infinity from those locations, where they cross is the target. Now go there.

Again, this isn't a trivial problem to solve, but I still think it is within your budget and reach. Before committing to air, I'd first learn how to find the transmitter on the ground. By breaking this problem into manageable pieces, what you learn on foot will be easier to use int the air.

[dtengineering]

Originally Posted by Don Rotolo

A valid approach, but you then have the issue of coupling the RF signal path through a rotating joint, not trivial.

Would it be possible to have the receiver rotate with the antenna, so that only power and a data feed need to be transmitted through the rotating joint? The rotating antenna doesn't sound like the best solution for this project, but I was wondering if it was easier to pass power and data than to pass the RF signal... honestly, I don't know if this is a practical technique or not, but thought I'd ask.

I've enjoyed this thread... nice to see a slightly different question and the typically excellent CD answers from people who really know what they are talking about. (And in the RF field, at least, I am not one of them!)

Jason

[Al Skierkiewicz]

Theoretically you could have a shuttered antenna array such that multiple dipoles backed by reflectors would keep antennas from receiving in multiple directions at the same time. A simple RF switch would then allow one receiver to sense signals from different directions based on the switching. The tough part is scaling down an antenna array at this frequency. A dipole is still over one foot long. I suppose you could effectively couple from something smaller, saya few inches. So the device looks like pie segments from the top. A dipole is fit into each corner so that it effectively can only receive in 90 degree increments for a four section assembly. Four dipoles gives you four separate directions. If you sense data and position the platform so that your are flying/receiving from the correct transponder, while the others transponders could be used for stabilizing the path.
You might also take a cue from instrument flying. There are high gain antennas pointed straight up at specific distances from the target that can only be received when directly overhead. (you might remember the phrase "I'm over the outer marker now" in some flying movies) This system would allow you to know distance to target with another receiver.
Remember that RF in free space is essentially traveling at the speed of light. Phase over short distances at the speed of light are extremely difficult.

[Alan Anderson]

By the way, you can simplify the problem greatly if you make the beacons a wee bit more expensive. Add a cheap GPS receiver to each and have them broadcast their coordinates. Combine that with spread spectrum modulation and you've got a very scalable scheme that could potentially handle hundreds of beacons in range simultaneously, with no special antenna setup required.