APRS Repeaters Get The Signal Out Of Mammoth Cave

APRS Cave-Link uses the amateur radio’s Automatic Packet Reporting System (APRS) inside caves to get their position data (and other messages) out.

Imagine that you’re coordinating a large scale search-and-rescue mission in a cave. You need to know where all your groups are, and whether or not they’ve found anything. But how do they all communicate to the command center?

You’d guess radio, but you’d guess wrong. Radio doesn’t propagate well at all in a maze of twisty passages, all alike; rocks absorb radio waves, especially in the VHF/UHF range that’s best suited for most small radios. In the past, you’d run wire and transmit along it. This article runs through the options in detail. But adding miles of wire to your already heavy caving and climbing gear is a nuisance or worse.


Some experiments by groups of amateur radio operators, and cavers, with APRS repeaters aim to change that. Digipeaters, as they’re known in the APRS world, take an incoming message and forward it on again. On each successive hop, the station that received the signal appends its name to a list of paths that are sent along with the message, which assures that the message propagates but doesn’t get repeated around forever in a loop.

Digipeaters and battery packs are dropped, in Hänsel and Gretel fashion, as the cavers work their way through the cave. The trick is to make sure to place one repeater before you’ve entirely lost the radio signal from the previous one, of course. But the APRS Cave-Link project got one mile’s worth of transmission in Mammoth Cave without using wires at all. That’s not bad!

waveguide_tnNow, GPS still doesn’t work underground, so the cavers need to bring an accurate map along with them and keep track of their own location. But even getting important messages (“we found him!”) passed around inside a cave environment is enough of a challenge.

We’ve seen APRS used for tracking high-altitude balloon payloads, and we can’t help but wonder if the same attention to weight-saving that’s demonstrated in these DIY versions wouldn’t also be useful in a caving context.

Have you made any cool networks of APRS links? Under adverse conditions? Let us know in the comments.

And thanks [Travis Goodspeed] for the unintentional tip.

55 thoughts on “APRS Repeaters Get The Signal Out Of Mammoth Cave

  1. I always wondered about such a spelunking idea like this. I’ve seen a recent DARPA robot that drops a bread crumb trail of throw-away digipeater signal extenders for itself. I think they use LiPo batteries and maybe a solar cell to extend life. Solar cells don’t work in spelunking ops though.

    There is a defunct digital 2-way radio American toy designed by ex-Soviet scientists living in Chicago. It was a great 900 Mhz toy that would really fit in here as a digipeater relay app. It was called CYBIKO. It could be programmed to repeat broadcast digital text messages to another CYBIKO (or group mode) ad infinitum. However, it’s on-board battery sucked and was not a conventional one so you’d have to out-board another via its external power jack. There used to be a CYBIKO hacker;s forum that lasted for decades, Now I think the interest has died down to nothing. You can still get them on EBAY and Amazon I think. I think the forums are still up but no traffic lately.

    RE: Search & Rescue apps:
    OK I know this is really off-topic but I just needed to chime in here. Recently I've seen a ARDUINO based device that senses chemicals via a gas analyzer module that is dirt cheap from China. It can sense particular gasses in tiny amounts of PPM,

    AMMONIA SENSOR!!! During the Vietnam War General Electric had a top-secret gadget called Operation Snoopy. It was a PEOPLE SNIFFER. Evidently people (and apes) give off copious amounts of ammonia from behind their knees and elsewhere on their bodies. This was detectable with a portable unit on a helicopter that had a suction fan in front of the detector, It was meant to find the enemy hiding underground and in the bushes. It could even find recently DEAD people too,

    What an idea for a Hackaday project! A search and rescue people sniffer with an Arduino as it's CPU.I really hope someone builds one and uses it to save hard-to-find stranded and missing people,

      1. davidcapper – US Soldiers had that SAME exact problem when using Snoopy on the ground. Once they put it in a helicopter it stopped detecting the soldier carrying the Snoopy unit. Obviously the helo introduced a closed environment and the intake port was outside the craft limiting Ammonia Hits to just the caves and bush, In a cave environment some sort of Ammonia isolation would be needed. Maybe a ATV robot carrying HaD Snoopy? A rescue dog which does not give off a lot of ammonia (except urine) with a Snoopy strapped to him? Or possibly the Snoopy Technician could wear a HazMat suit? A tight fitting suit with one way valve air supply would be great for spelunking rescues. Since the underground temperature is probably close to 21° C (70° F) (s)he won’t overheat in the HazMat suit. Then the only Ammonia targets will be humans (dead or alive) or a monkey (either way win-win).

        How exactly do you get hypothermia in a underground cave? I would think only in a ice cave or if you get wet. Otherwise it is basically room temperature down there isn’t it?

        1. The earth is a huge heat magnet. Think about sitting on a rock, or a chunk of concrete. The heat gets sucked out of you. With no thermal input from the sun, or a fire to replenish the heat absorbed by the surrounding stone, your body uses calories to produce the heat to keep you warm.
          Caves are rarely dry pristine environments. They have streams and pools, and mud. Hypothermia is a definite possibility.

          Also, bat guano has huge amounts of ammonia..

          1. Mike – I totally forgot about the bats!

            SOLUTION: Since this would be a search and rescue mission:
            1) Any lost person(s) would not be near the cave entry where bats like to hang out. Likely would be several thousands of yards deeper somewhere.

            2) Activate clearing fans near the entry to clear away ambient ammonia and the ambient fan noise will clear away the bats too.

            3) Sprinkle copious amounts of barley straw over the guano and put up black lights to control bacteria risk to breathing toxic guano fumes. .This method is used to clean up green algae-plagued ponds infected by water fowl droppings.

            4) Setup ultrasonic anti-rodent transmitters powered by car battery and inverter. This to keep bats from coming back while your working on your SAR mission.

            I like AussieLauren’s idea. Put a autonomous UAV in the cavern system with wall collision avoidance and a portable FLIR detector. When it returns it downloads it’s FLIR capture based on it’s inertial navigation mapping. Wire tethered guided would help with battery life. If it is self-retracting tether you could use that to detect how far the UAV flew into the caverns. Tight spaces would have to be avoided as rotors and tether may foul.

            This is all only BRAINSTORMING not speaking in any authoritative or dogmatic fashion here…


    1. That is actually not a bad idea!
      I’m and EE and a volunteer firefighter at DEMA here in Denmark, and part of our for Heavy USAR team.

      We use dogs to sniff out humans buried in gravel and of course cameras, listening devices and such.
      Souch a sensor on a ROV of some sort, car or quadcopter might not be a bad idea.

      Our dogs are great but they require a whole lot of support to bring in a plane and keep alive and happy for a few week w/o using local resources…

      Could you guys help me in the direction of such a sensor? :)

      1. martinmuunk – The Ammonia sensor I found where from China and were only a few DKK’s. They also said they could interface with an Arduino. I think I found a web site that actually had a diagram for that. You would only need to add a PVC pipe for the air intake and a 12 volt DC muffin fan to pull in surrounding air. A quadcopter may not be able to handle the payload as the batteries would weigh it down, The fan would draw too much power I think. A ROV sounds better like something from USA’s Boston Dynamics or iRobot.

        A Great Dane, German Shepard, or a St Bernard would be ideal to carry this payload under their necks. All they need to do is roam around the caves with the Snoopy People Sniffer. Their noses would not be be needed.Just let them pee before going down so it doesn’t false from their urine. Train them to bark when Snoopy starts to beep. You hang back by several meters so Snoopy doesn’t false off of you. Other animals can work too like cats, ponies, goats. sheep but NO MONKEYS! :-)

        1. Dogs don’t sweat! Also the ammonia from humans and monkeys comes from behind the knee and your wrists. That’s according to EE’s from GE who invented this SNOOPY PEOPLE SNIFFER in the late 1960’s. It was featured in a Charles Bronson movie where he played a USSS agent protecting the US President. The Snoopy Unit picked up hostiles in a barn up ahead of Marine One.

        2. Would such a sensor be anywhere nearly sensitive enough to detect a human?

          Well. I dont see your point about the dogs. They have a perfect gassious-compound-detector build-in from the factory themselves (their noses) ;)
          The whole point was to eliminate the dogs from the USAR team and have more dedicated rescue personnel.

          1. martinmuunk – That’s why a suction fan and PVC air guide is needed to increase the PPM rate to the sensor. The sensor is already quite sensitive but experimentation will be needed to see how much. GE has never evolved this technology and its been sitting idle for 50+ years waiting for an application. Who would have thought that a military weapon could be re-tasked as a SAR (search and rescue) method?

            RE: DOGS your talking about TRAINED dogs trained in scent tracking. I’m just talking about a non-sweating MULE for the Snoopy Package or your typical mutt from home who doesn’t know the first thing about tracking humans or anything else. Just talking about good ol’ FIDO who probably would only be sniffing other dogs bums other times :-)

            FIDOS are much cheaper than ROV’s, UAV’s, and HazMat suits. Just let them roam around aimlessly looking for food treats you’ve hidden along the cavern system. The Snoopy will trigger off when it smells ammonia. You are back a few meters so to not false the Snoopy system and will hear the high-pitch beeping so FIDO doesn’t have to. However, if FIDO takes a whiz the Snoopy will go off. That’s why you let him go pee BEFORE he enters the cavern and don’t let him drink any water until he is done with the sweep.


      2. Call me crazy, but why not use one of the 4 fans on the quad copter to push air past the sensor vs a completely separate fan and tube? I mean it after all is flying in the same air it’s trying to sample right?

    1. Rud Merriam – to solve that problem get some LED flashers from let’s say the local Dollar store toy department. The Halloween flashers take AAA batteries. The others take lousy Hg cells, 3 in all. You could use some sort of long life battery replacement. A backpack full of these switched flashers and you could bread crumb your route into the caves every few hundred or more feet. Since you’d be able to see these things for a long way (even around wet corners) you wouldn’t need to be too generous with them. Super-Alkaline batteries should last for several days. A lot cheaper than laying expensive Calumen sticks which don’t last that long.

      If you wanted to leave a legacy cave navigation map for others then print off a ream of paper QRCODE sheets with an incrementing number scheme on them. Then you could put up a web app that could be downloaded BEFORE you go down into cave. The code sheets correspond to cabe waypoints that you have mapped out on an interactive map that even has your comments (text an/or verbal) that describe that waypoint in detail. If each code sheet is folded up and put into a ziplock baggie and nudge under a rock sticking out, a spelunker could take it out and use their smartphone to scan the QRCODE and the built-in app you made would interact with the splelunker. No outside internet or radio link required. It would all be standalone app.

      (this is my alter-ego: https://i.ytimg.com/vi/p6k9tpzF38I/hqdefault.jpg)

  2. I suggest you take a look at what this guy has been working on (I was attending his Phd thesis presentation 7years ago).
    Thomas Watteyne: https://twatteyne.wordpress.com/publications/
    If I remember correctly, he was able to extract some kind of virtual map only by the radio links and packet route on the network (he was working at the time on self organisation in sensor wireless networks).
    I will let you guys to it …

  3. APRS on hot air balloons is made lightweight and low power.
    Lightweight might be nice, but low power would be awful in a cave.

    Balloons are way up in the air, and don’t need to have power to punch through or bounce off rock and dirt.

    1. jon – The new Calcium batteries are more powerful than lithium ones. Caves are full of CaCO3 or calcium carbonate. Just figure out an invention to exploit that into a cheap but powerful power source in caves, However, with repeaters High RF power is not needed so much. Also UHF+ frequencies tend to bounce around caves until reaching around corners and out corridors.

      1. High RF power is not usually needed with a repeater because they are usually mounted above the surrounding terrain. In a cave, more RF power would be needed.
        As the hams that live in my neck of the woods know: granite makes a really good RF filter.

        1. My CYBIKO idea is 900 MHz and only operates at around 100 mw – that’s a QRP rig. I think the RF would bounce around the cave system making it’s way to the other repeater several thousand feet away (maybe 1/8 to 1/2 miles? Normally they operate about 1-3 miles LOS on the surface). I don’t think 900 MHz can penetrate limestone no matter how high the RF power (within reason of course). I don’t know the frequency band APRS operates at. Also no license is needed for Cybiko as it is FCC Part-95.

          1. APRS is ran on the 2 meters band typically. However, it does not matter what frequency it is ran at the beacons and trackers will still work. A few of us have been kickin the idea around of trying to make a repeater from an old Linksys router using the 2.4 GHz that is already on board. I would think your idea would work with APRS on 900 MHz

  4. 9-dof inertial navigation module
    9g digipeater with Li chemistry
    50g UAV with basic collision avoidance and low-signal seeking algorithm

    Multiply by 50

    Map your disaster site.

    P.S. Please include “BecomeSkynetCost = 999999”

  5. I had another flash on an idea. Using APRS or even my Cybiko repeater idea, in either case you introduce latency to the time to travel of the message the longer the media content is. Each repeater hop slows down the message a few milliseconds. That aggregate latency at the other end could be on the order of several minutes.

    How many different contingencies can there be in a cave environment? Couldn’t you reduce just about any event or contingency to a 3-digit code? Even your identity and location in said cave could be reduce to 3 digits.By doing that you’ve cut message latency down to almost nothing. Then by sending just the 3-digit code your device could decode it from a built-in database. You wouldn’t even need to remember the codes as that could be encoded from a drop down menu on the UI (user interface).

    And if you do happen to remember some of the codes by heart you could handwrite them on a cave wall with chalk to convey something important to other passing spelunkers. HOWEVER, if you just have to convey a non-categorized message content you could send 000 or 999 for standby for free-form text message.

    Typical message: 123 012 333 translates to:“This is Harry ID 123. I am at location 012 which is the large cathedral room at 1000′ level into the cave near the large pond of crystal clear water. Condition 333 = I found some evidence of previous human activity here.”


    1. Good thoughts… I agree that each hop of the information would add a few milliseconds latency, but the information transmitted could only be delayed by several minutes if there were literally 10s of thousands of repeaters in the system (12,000 repeaters assuming 5ms per repeater producing a 2 minute message latency). I’m no APRS expert, but with that many repeaters, you’re probably hitting a limit to the message length in the first place – there’d be more repeater ID’s in the message than the message itself. You’d need to be dropping them every few feet to get that many down.

      The message tokenization you’re suggesting would reduce the message size, but not the latency for the message to arrive. The message latency is fixed – based on the number of APRS repeaters.


      1. John C – Your talking about APRS and it’s messaging overhead. I am thinking of CYBIKO (check my earlier posts here) in where you can use it’s SDK to tweak the messaging overhead and reduce it to raw data and no special handshaking overhead. Of course my method has no error checking routines or automatic re-transmission for lost packets. It’s a WYSIWYG system. Maybe it could send 3 times to compensate.

        I would like to explore that new PSK over SSB mode HaD did an article on recently. I wonder how that would work in a cavern system. Oh SNAP! LIGHT BULB over my head again!!! How about an ultrasonic ELF system that transmits this PSK signal not as a repeater but as an actual walkie-talkie transceiver. Imagine the ELF signal making it’s way through the cave system. No long wire either just a good ELF transducer inside a small plastic parabolic reflector. And a emergency hand-crank recharging system to keep battery topped off.

  6. How about 2-way laser communications pods? Put one emitter/receiver pair on a 2-axis swivel so it can be aimed at the previous unit. They would also need an adjustable base, like those GoPro mounts with the legs that wrap around things.

    A bit of beam spread and a lens that can focus an off-axis beam onto the photoreceptor would be good for this. Also, a couple of 1 mw visible lasers for aiming.

    When not transmitting/repeating, they’d power down to just monitoring for the IR laser hitting the receptor, so each message would be preceded by a wake up packet. Could even have a remote command to turn on the aiming laser to aid placement of downstream nodes as the cavers proceed.

    Transmission speed would be fast and should be able to have the bandwidth for video. The downside of such a system is that if something knocks a node out of line, the link is broken. Would be easy to tell which one with an automatic alert to both ends of the chain with the ID of the node that’s out of contact.

    IDs would not need to be pre-assigned. Set the first placed one to #1 and each one placed after auto sets to the next number.

    Can a laser beam be widened to 10mm or more?

    1. Galane – I love LASERS so this is a GREAT idea! How about a method to mitigate aiming? You know those laser jigs carpenters use? It paints a laser line on the walls in a 360° pattern. Imagine a modified version that also has a vertical component so the entire room is painted in infrared laser light. You just prop it up on a ledge or on a rock in the middle of the cave room. The receiver is a Argos (many eyed Greek god) style photo-diode array. It could sit on top of laser transmitter so it could be desensitized from it and only see remote laser transmitters without needing to be aimed. The Argos is a Styrofoam ball with a bunch of photo-diodes or solar cells embedded in most of it.Would look like a funky disco ball with wires coming out the bottom.

      Yes you can diverge a laser beam with a common lens.


      1. Oooo, yeah. A diffraction grating to project laser grids. Decently omnidirectional, just aim in the general direction of another node. Could have the grating able to shift around a bit in case the receiver happens to be in a grid hole. It would only need to shift X/Y enough to cover the area of one hole to be able to have total coverage within the bounds of the grid.

        Heck, the laser grids may as well serve a dual purpose for 3D mapping the cave.

  7. It would have to use 900 MHz and above. VHF doesn’t work at all in tunnels. Using handie-talkies, I’ve seen the person in the next subway station and been unable to talk to them. UHF is better and will work with large subway tunnels. But at cellular frequencies, I’ve tested a 10W ERP CW signal at 885 MHz and still had -80 dBm a mile away. At 2GHz the results are similar. Still… this was tested in smooth subway tunnels. Anyone have experience with these rough cave walls, forks, and odd shapes?

    1. Tony Kay – I would think you were experiencing multi-phase cancellation of the received signal hitting your antenna from multiple reflections arriving at your antenna at the same time. A diversity antenna could fix that. I would think the irregular wall shape of caverns MIGHT mitigate that physical anomaly but I can’t be sure without actual experimentation.

      A trick that mining companies use is to erect two antennas at both ends of a mine’s entrances. Like one at south end and another at north end. Then run a high-loss coaxial cable between them down through the underground caverns. The cable is so lossy that signals can be injected to it from walkie-talkies or portable mobile units and RF signals from topside are also carried down below. Of course you can hear the guy on level 3 at level 1 as the cable runs through every level.I would think that all units need to be good power level like 5-10W or more. Nothing is actually attached to the cable except the two antennas.


  8. APRS is extremely basic and primitive
    the problem seems to be size and shape of tunnels, solution could be SDR repeaters, or at least tunable transceivers with wide available spectrum coverage
    thin passage blocking longer wavelengths? tune transceivers for best frequency at that particular hop

    are rocks causing reflections? or do they soak up rf? reflections can be dealt with phase array antennas and mimo transceivers

  9. I just wanted to put my two cents in, if I remember back in the 70’S some people were sending radio signals in caves with VLF very low frequencies. They were able to send all the way to the surface. And the Navy uses VLF to communicate to subs under water.

    1. @Thomas – True VLF is still used but ELF is being used too. The USN is becoming more dependent on the SLC System which used B/G Lasers from satellites. But Through-the-earth communications techniques has taken a new twist by Lockheed Martin (Magnelink MCS) and a Canadian TTE comm company (RockPhone). Apparently they came to the preconceived conclusion that magnetism has little trouble with water and rocks. You can use your magnetic compass almost anywhere. So if you have a moderately powerful electromagnet generator (magneto inductive) and a decent sized loop antenna, you can pulse-modulate the magnetic loop (PCM) and a receiver about 1500~2000 feet can pick up your voice, text, or Morse code message. I imagine that the frequency is lower than ELF. I think I read it may work from 0-300 Hz (0 being DC but pulsed in this case). Here is a vido for the Ultra Electronics RockPhone:

  10. There are 29 men lying inside the coal mine at Pike River (West Coast, New Zealand) who have been there for five years now. They have been unable to be returned to their families because of dangerous, unknown and potentially explosive gas levels deep inside the mine. I have always felt something like this would be the first step in getting the miners home.

    1. THat was a really sucky situation, even worse was the fact the robot they sent in was not waterproof and broke down during the search. It was a total fiasco. until recently I had an EOD that would have been perfect for the job and instead of electronics used specialised military sealed relays. beast was able to tow a car. Sometimes New Technology is not always the best.

      As for underground communications there is a really cool stuff called Leaky Coax, the coax has slots cut in the shielding and works quite similar to a waveguide, something like this would extend the communications a huge amount.

      I wonder why they don’t use ELF for the transmission.

      1. Part of the problem is that there is a rockfall at the end of the drift which blocks access the the mine proper. I am picturing a remotely operated digger or mini TBM. Technology will have the answer, we just need someone with the passion.

    2. Matt – The Americans have a special “suit” not unlike IRONMAN that can brave the methane fires and lift and carry the bodies out one at a time. However, it is quite expensive and slightly classified. Boston Dynamics has an android called ATLAS that is being considered for going into Fukushima to turn some knobs and stuff. But I’m not sure he is ready for that or Pike River yet. Not sure he 100% fire-proof like DATA from Star Trek. He is better suited than Honda’s Asimo too. I understand 4 robots are stalled permanently in Pike River? They are entombed with the 29 men too I guess. So tragic for those poor families…

  11. Wouldn’t some form of seismc communication work? The users would have to press a hammer pin to some solid rock then it could tap out a message. I think it would work best with digitally encoded text messages. Not sure how well the signals would propagate or how you would detect them but it may work.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.