Position Tracking Concept - Low Power Use

Hi everyone, pretty new to the forums so I hope I've come to the right spot.

I've recently started an endeavour, which brought me across the arduino boards. I'm no electrician or micro-controller expert, but I do have programming experience. I live in South Africa where livestock theft is quite large. I'm trying to create a livestock tracking system that will interface with an online application, to alert farmers of this. At the same time, I want to build it into a sort of management system where farmers can track which cattle are in which area.

I don't need precise position tracking but more of an area indicator. ie. if a cow breaches the fence line, send a push notification. This brings me to my problem... I need to create a low power emitter that will be attached to a cow's ear tag. I was thinking something with a motion-generated recharger as opposed to micro solar panels. The emitter needs to be as cheap as possible since farmers can have over 5000 head of cattle at any given time.

Does anyone have any suggestions or alternative solutions for this?

I'm no cattle expert, so take my remarks in that context. :)

I'm inclined to recommend some sort of low-power, low-range transmitter which you could attach to each animal. For example, the JeeNode:

http://jeelabs.com/products/jeenode

Judging by another page I found, you might expect a couple of hundred meters range, depending on terrain. This should be good enough for detecting if the cattle is in the paddock, or has wandered off or been stolen.

I would have each one sleep most of the time, and at some interval you determine wake up and transmit a short message to a base station. You could put a serial number (cattle ID) into the EEPROM of each chip, so each one sends a different code.

To save power, have a long time between messages, if you want to catch rustlers in the act, maybe a shorter time (eg. 15 minutes?).

The base station could be in the middle of the paddock (thus you get 200 m in each direction from the station, so you could have a 400 x 400 m paddock) and then base station could then repeat recent transmissions via (say) CB radio or a cell phone link.

Stuff about his measured power consumption here:

http://jeelabs.org/

My own pages about saving power:

http://www.gammon.com.au/power

The simplest power source might be a button cell (eg. CR2032) which should last a year or two if you don't transmit too often. For example, my car keys use a CR2032 and I lock and unlock the car multiple times each day, and it lasts a few years.

As for the rustlers, if they leave the tag on the cow, then you have some hope of finding it. If they don't (and personally I would remove something as obvious as that) then you may find the gadget in the paddock, but not attached to Bessie any more. To try to stop that, you could have the wire that holds the tag to the cow register when it is cut. So they cut the wire, and the gadget then wakes up and transmits a "theft" signal. If they destroy the tag before cutting the wire, well you still get the "no cow" signal at the next appointed time (because it is not sending anything).

If they are smart and strip the insulation from the wire, and bridge the ends before cutting it (with a longer wire) then that scheme might fail. But you might also have a temperature sensor built into the device (a thermistor) which normally would register "cow body heat" but some other temperature if it has been removed. Another approach might be to have a light sensor on it. If the device is held against the cow's body then not much light would get in, if it is removed more light would get in (during the day, anyway).

At the same time, I want to build it into a sort of management system where farmers can track which cattle are in which area.

If you have multiple base stations (say half a km apart in each direction) then the one the animal is nearest to should register a signal, so you would have a rough idea of which cattle are where.

What sort of area and/or boundary length were you hoping to cover? It seems to me that anywhere big enough to hold thousands of cattle is going to be a big place, which implies a massive scalability problem - you either need thousands of transmitters capable of transmitting over a long distance to a small number of receivers, or a vast number of receivers so that there is always one close enough to every transmitter to pick up a short range transmission. Whichever way you do it, you're going to need a lot of something.

PeterH: Whichever way you do it, you're going to need a lot of something.

It's beginning to sound like a great idea which is not presently available not because no-one thought of it before, but because simply does not happen to be practical.

They are now putting RFIDs on supermarket products as I understand it. Thy can now make them cheap enough for this. But they have a range of mere centimetres. As I have pointed out in another post, electromagnetic radiation (as with fluid diffusion) operates on an inverse square law; the power level or concentration drops away with the square of the distance.

So what can be done at a certain price 10 cm away, requires a hundred times the power level at a distance of one metre and will tend to be a hundred times as expensive. Even if it were only 30 times, it will be another 30 times that at ten metres and so on. 30 times 30 is more-or-less 1000.

PeterH: What sort of area and/or boundary length were you hoping to cover? ... Whichever way you do it, you're going to need a lot of something.

Perhaps. I don't know how much one cattle costs (is one cattle a cat?), but if it is a few hundred dollars you can probably afford maybe $20 to stop it being stolen.

The active transmitter idea shouldn't be too hard to implement, and if it has a range of a few hundred meters then you only need a receiver every six or seven hundred meters apart. Let's imagine for example there is a watering trough, that would be a logical place for the receiver as the cattle would go up to it for a drink.

I don't know enough about cattle behaviour to say for sure, but if they hang around in herds, you probably only need to tag one out of 10. After all, if they stick together if you know where one is, you know where the other nine are. And if someone steals one he will probably steal a dozen.

I did some testing with the JeeNodes I had to hand. With direct line-of-sight we seemed to get about 65 m reception. This isn't exactly 200 m so maybe that isn't acceptable.

However a different transmitter module, higher power, another frequency, may all influence things for longer range.

Plus, if you had a better antenna on the receiver it would probably help. I just had a short piece of wire.

Thanks for all of your replies. They are very informative.

So to answer a couple questions...

Each cow costs about R8000 ($750), if we take the average over a whole herd. The idea is to re-use the device when the cow is sold and just replace the battery. So $20-$30 per emitter, seems very plausible.

Each field varies in size (up to 2 acres - 8094 m2) and shape, and some have foliage inside. Each setup would have to be custom, using basic math to try to find the best placement for each reader.

With regards to battery life, I was thinking of using a motion charging battery, since cattle shake their ears around alot (similar to this -> http://www.saikatbiswas.com/web/Projects/Livecell.htm - not sure what options are available on the market though)

Placing readers 50m apart doesn't seem plausible, and stands a chance of them getting stolen, along with the cattle.

if a cow breaches the fence line, send a push notification.

Interesting problem. Instrumenting an already instrumented (RFID) animal to detect physical separation of the cattle from the home-turf.

Perhaps flip the problem around. Instead of the cattle transmitting a signal, perhaps have each instrumented cattle receive a signal only when they were near the boundary. Receivers can go a very long time on a battery where transmitters consume power. This concept is like the "Invisible Fence (R)" dog containment system.

When the animal was very near the invisible fence (transmitting antenna), the unit on the animal would then transmit a signal powerful enough to be picked up by a centrally located (elevated) receiving antenna. This way, the transmitting is limited to only those beasts that need to be monitored and only for the time they require active monitoring.

You can publish for free on sites like Webmonkey.

Ray

I've read about several asset-tracking products recently that use small and inexpensive-looking tags and are described as suitable to track luggage, lost keys etc. The previous generation used a lot of power and were basically using a BlueTooth link as a proximity detector which would be useless for what you're trying to do but perhaps these new ones will be more useful. The problem I see is that they seem to rely on having an infrastructure round them capable of communicating with them, and while this may be practical in a city it's hardly going to be practical out on the plains of Africa. The conventional solution for that would be a GPS and GPRS unit. While that may be practical to track heavy equipment worth hundreds of thousands of dollars, it's hardly going to be cost effective to buy and operate one of these for each cow. If you're out of cell coverage (which seems pretty likely) then the problem gets even worse and you would be looking at sat phones - obviously out of the question.

So it seems to me that the problem is not the technology - it's getting it to scale up to such a massive size without costing a fortune. At the moment I can't see any tracking solution being cost effective. Perhaps you should be looking at an electric fence to detect things crossing the boundary, and a bunch of quadrotor drones to go out at night and see what's happening.

mrburnette: Receivers can go a very long time on a battery where transmitters consume power.

I would dispute that.

Consider the wireless doorbell. The transmitter uses a CR2032 (well, the one I just installed does,) the receiver uses AA (not AAA) batteries. The receiver will require new batteries far more regularly than the transmitter.

The difference is that the transmitter pulses occasionally, the receiver has to be on all the time (and due to noise, is continually decoding the signal looking for information).

RodRitter: Each field varies in size (up to 2 acres - 8094 m2) and shape, and some have foliage inside. Each setup would have to be custom, using basic math to try to find the best placement for each reader.

OK, so if they are square that is ?8094 = 90 m per side. And if the receiver was in the middle it would be within 63 m of any cow (even on the hypotenuse). So the rough figures for the JeeNode transmitter (and I have no particular reason to push that, except I have one on hand) would seem to be in the ball park.


Let me show you some photos. This is the JeeNode board, which is the processor chip and the transmitter/receiver module on the right (circled):

Clearly the transmitter/receiver is quite small, so if you replaced the DIP-28 processor by a SMD one, the whole thing would be much smaller. And if you ran it from a button cell you don't need the voltage regulator, diode, filter capacitor, etc. And you could just have pads for the SPI programming and use a "pogo spring" arrangement if you needed to reprogram.


Slightly blurry close-up of the transmitter. You can see that is a separate board, I presume you can buy those off the shelf.

That board is 15 mm x 15 mm approximately.


One option for power, a single AA cell on a boost-converter board.


The whole thing together:


So, a custom board made along those lines could be quite small, run off a single button battery, and have the anti-theft wire attached.

Caveat: The CR2032 gives 3v, that might be too little for the transmitter, I believe that requires 3.3v, you might need to check that sort of detail. Still, a boost-converter could handle that part, for slightly extra cost.

And maybe there are other transmitters around with more range. And in this case the transmitter/receiver combination might be overkill.

RodRitter: Each field varies in size (up to 2 acres - 8094 m2) and shape, and some have foliage inside. Each setup would have to be custom, using basic math to try to find the best placement for each reader.

Really, so small? Are these just holding paddocks, or fields that the herd live in? If they're really that small, surely you would be dealing with hundreds of fields, then.

This idea is very good, so that those cattle who can save most of the time, but because the scope is too broad, it seems too costly

Thank you everyone for your input. I like the idea of using the JeeNode (Thanks Nick). Going to have to do a few tests to see the power usage between using the tags as readers or emitters. There seems to be a split idea regarding this. I would assume the reader would use more power as Paul__B stated.

Anyway, I’ll let you know what happens. Thanks again everyone!

First, let me say that I don't know much about cows except I like eating and drinking dairy products, I like eating their flesh and some of the offal, and they go "Moo!". ;)

But how much weight can/should go on one of those ear tags? Might a collar be a better option to allow a larger package (particularly for a larger battery) and a better choice of weather and dust sealed off-the-shelf enclosures?

Good luck with the JeeNode experiments.