Underground Comms: Let's talk about wireless mesh & future standards

[Piotr Pasieczny]

Hi everyone,

I’ve been browsing the forum for a while and I noticed there hasn’t been much recent discussion dedicated specifically to modern underground communication standards. That’s why I decided to start this thread.

Brief intro: I represent a company specialized in critical communication systems for isolated environments (mining, tunneling, maritime). We have a fully operational, battery-powered mesh system that is currently used to protect workers in these hazardous conditions. We know our system works, but we also know that speleology has different requirements than a mine shaft. We are not here to sell, but to learn and initiate a dialogue that helps us continuously improve our technology.

We want to hear from the experts:

Since we don't see many alternatives to wired phones or earth-current radios being discussed – is there a reason for that?

1. Is it just about the physics/geology, or have ergonomics and weight been the main barriers for wireless tech so far?
2. If you could design the perfect comms node for your expeditions, what features would you prioritize (apart from range)?

We want to bridge the gap between industrial safety and exploration needs, so your honest feedback is invaluable to us.

Best regards,
Piotr Pasieczny
Sybet International

 

[PeteA]

I am in touch with Piotr on this matter.
We would welcome other opinions please - especially those outwith the cave rescue world.
Pete

 

[wellyjen]

Caves are often too wriggly, compared to mines, needing mesh nodes every few tens of metres at best. Cost goes up rapidly.
The most common use of underground comms is rescue, where speed of setting up is important and must be able to survive high flood waters and rescuers manouvering themselves past them. There was a skill in speedily laying out 'phone wire to avoid getting broken by passing cavers. Rescue teams are not rich, so can't afford to buy many mesh nodes. How much do mesh nodes cost these days?

 

[andrewmcleod]

'Ordinary' caving doesn't really have any need of communication.
For rescue, you are going to a cave that may not be known _that_ well to the rescuers, that you can't have pre-placed gear in (there are far too many caves in the country to litter all of them with mesh nodes in advance, and far too few cavers or reasons to keep such a network updated), where you may not have that many people (and can't spare a significant number to organize communications), where carrying kit in/out is a significant amount of effort, where passages are often tight and awkward and winding, where passages may flood or not have any dry floor, and where the current systems (e.g. surface Cavelink and underground Cavelink, or wired phones) are relatively low-effort.

For our expedition, we have a surface Cavelink unit that uses around 200-300 m of antenna wires and is placed about 40 minutes walk away from our base camp. We leave this on the surface for the three weeks and communicate with it via text. Underground, we have a second Cavelink unit in our Camp 1 which plugs into fixed underground antenna of again around 200 m or so. This gives us reasonable comms (except when it is very very dry and the surface antennas get bad grounding) through about 700+ m of rock. It takes one trip to set up the surface unit and one trip to retrieve it, and the only thing that people need to bring underground is a single Cavelink unit (which is still not inconsiderable as there are 550 m of pitches and then about an hours caving to get to the first camp). I don't think a mesh network would suit us for this?
This does only give us comms in a single location which isn't actually best placed for two of our current leads; the passages are large but there are probably 1-2 km between each lead (2-4 hours of caving).

 

[Piotr Pasieczny]

Caves are often too wriggly, compared to mines, needing mesh nodes every few tens of metres at best. Cost goes up rapidly.
The most common use of underground comms is rescue, where speed of setting up is important and must be able to survive high flood waters and rescuers manouvering themselves past them. There was a skill in speedily laying out 'phone wire to avoid getting broken by passing cavers. Rescue teams are not rich, so can't afford to buy many mesh nodes. How much do mesh nodes cost these days?

Hi Wellyjen,

Thank you for the insight. You are correct—relying purely on RF in tight, meandering passages can be physically inefficient compared to wire.

1. Hybrid Deployment
Nodes support a hybrid mode. They can be interconnected via wire through difficult sections (sumps, tight meanders) and switch to wireless mesh only in larger chambers or for the mobile forward team. This significantly reduces the node count compared to a fully wireless setup.
2. Placement Optimization
The handsets display real-time Quality metrics, allowing the team to place repeaters at the actual limit of the signal range rather than estimating distance.
3. Cost
The current list price for a MiniNode is 1199 EUR. We are aware this is significantly higher than analog wired solutions used by volunteer groups, but it reflects the autonomous mesh capabilities.

Your feedback on deployment speed vs. flood resilience is incredibly valuable for us. It helps us understand where the real operational limits lie.

Best regards,

 

[first-ade]

A few of us recently had an experiment with some Meshtastic nodes in Valley Entrance and managed to get a fair way in with them, in the end we reckoned you would need around 15 nodes to get from the entrance to the pitch head (hopefully someone will be able to say how far that is). But this does require a bit of time to work out when a node can no longer see the previous one and move back until it reappears, drop a node and move on. There's also an art to finding a location to leave the radio where it gets good signal but isn't going to get knocked off it's ledge or destroyed by other passing cavers.

Other considerations would include the licensing of the frequencies used. Meshtastic uses ISM bands, whereas IIRC the Heyphone and Nicola radios use around 68khz for the signal to get through the rock (I guess the cavelink uses a similar frequency). Both choices mean that you don't need an amateur radio license.

The best use for a mesh radio at the moment would likely be getting a messages to through-the-ground radio like Nicola, Heyphone, cavelink etc from a site a bit further away from the cave radio. For example, when the cave radio is in a known good location for reliably communicating with the surface, but you would ideally want Comms from your moving casualty on a rescue, mesh radios could bridge that gap between the casualty and the cave radio.

 

[andrewmcleod]

I _think_ we got comms in Valley Entrance from the entrance to the pitch heads with hand-held radios using six radios on a rescue practice? Starting at the entrance, one radio and operator, then an unattended repeater radio, then an operator with two radios (on different channels?) to manually relay messages from the entrance onwards, then another unattended repeater radio, then the pitch head radio and operator. I could be misremembering (I wasn't involved in that). With the set-up we have, you can't let two repeaters 'see' each other or they get overexcited and start just looping between themselves.

 

[Piotr Pasieczny]

A few of us recently had an experiment with some Meshtastic nodes in Valley Entrance and managed to get a fair way in with them, in the end we reckoned you would need around 15 nodes to get from the entrance to the pitch head (hopefully someone will be able to say how far that is). But this does require a bit of time to work out when a node can no longer see the previous one and move back until it reappears, drop a node and move on. There's also an art to finding a location to leave the radio where it gets good signal but isn't going to get knocked off it's ledge or destroyed by other passing cavers.

Other considerations would include the licensing of the frequencies used. Meshtastic uses ISM bands, whereas IIRC the Heyphone and Nicola radios use around 68khz for the signal to get through the rock (I guess the cavelink uses a similar frequency). Both choices mean that you don't need an amateur radio license.

The best use for a mesh radio at the moment would likely be getting a messages to through-the-ground radio like Nicola, Heyphone, cavelink etc from a site a bit further away from the cave radio. For example, when the cave radio is in a known good location for reliably communicating with the surface, but you would ideally want Comms from your moving casualty on a rescue, mesh radios could bridge that gap between the casualty and the cave radio.

Hi first-ade,

That’s a brilliant experiment! Using Meshtastic proves the concept works. And you hit on the two biggest challenges of mesh in caves:

1. Node Count: Yes, 868MHz doesn't like corners. You can need a chain. But that is why hybrid (mesh and/or cable) nodes are quite useful
2. Placement Guesswork: The "walk forward until signal drops, then step back" dance is slow and frustrating.

This is exactly what we solved in our industrial system (Sybet):

Since we build this for rapid rescue deployment, we couldn't rely on guesswork.

Our solution:
The handset (mPhone) has a built-in "Link Quality" analyzer.
Instead of guessing, the device actively shows you signal strength.
This removes the "step back and forth" trial-and-error. You walk, you drop a node, you keep walking. It makes deployment much faster and ensures you use the minimum number of nodes required (saving battery and gear).

Regarding your "Bridge" idea:

"mesh radios could bridge that gap between the casualty and the cave radio"

100% Agreed. This is the "Hybrid" tactical doctrine.
Use TTE (Nicola/CaveLink) for the heavy lifting through the rock to the surface.
Use Mesh to extend that link from the base station to the moving casualty/stretcher party.

It’s great to see DIY experiments confirming exactly what we see in professional deployment. If you guys ever want to compare Meshtastic vs. a purpose-built industrial rig on a drill, let me know!

 

[Piotr Pasieczny]

I _think_ we got comms in Valley Entrance from the entrance to the pitch heads with hand-held radios using six radios on a rescue practice? Starting at the entrance, one radio and operator, then an unattended repeater radio, then an operator with two radios (on different channels?) to manually relay messages from the entrance onwards, then another unattended repeater radio, then the pitch head radio and operator. I could be misremembering (I wasn't involved in that). With the set-up we have, you can't let two repeaters 'see' each other or they get overexcited and start just looping between themselves.

Hi Andrew,

That "looping" nightmare you describe is the classic problem with standard analog/DMR repeaters ("dumb" repeaters). Without smart routing, Repeater A hears Repeater B, re-transmits it, B hears A, and you get an infinite feedback loop that jams the channel.

This is the fundamental difference with a true digital Mesh (like Sybet):
Our nodes are "smarter". They don't just blindly repeat everything they hear.

1. Packet ID: Every voice packet has a unique ID.
2. No Loops: If Node A hears its own message coming back from Node B, it ignores it. It knows "I already processed this."
3. Self-Healing: You don't need to worry about "repeaters seeing each other". In fact, in our system, we WANT them to see each other. The more connections, the more robust the mesh. If one path is blocked, the data flows through another automatically.

What you tried to do manually (alternating channels, manual relays) is exactly what our protocol handles automatically in milliseconds. You just turn them on, and the network sorts itself out without the feedback scream.

 

[PeteA]

And we are working on connecting the Meshtastic radios to the Nicola 4. Have the skeleton code working and now need to integrate into the Nicola code.

 

[Piotr Pasieczny]

Dear Pete,

That sounds like an ambitious project. It is great to hear the skeleton code is working.

How is the integration progressing overall?

I am asking out of genuine interest. We know exactly how complex underground communication projects can be. If you are interested, I would be happy to outline specifically where our system offers superior performance and what broader capabilities are available out of the box compared to custom integrations.

We wish you the best of luck with the development, regardless of which path you choose to continue.

 

[Long Drop]

For a lot of us going caving is a nice break, away from 'comms'. Just saying.

 

[Piotr Pasieczny]

For a lot of us going caving is a nice break, away from 'comms'. Just saying.

That’s a very fair point, and I completely understand the desire to disconnect and enjoy the cave as a place of solitude and escape. For many, caving is indeed a break from the constant connectivity of modern life. That’s actually one of the main reasons I practice sport myself.

At the same time, I believe there’s also a place for communication technology in underground environments—not to disrupt the experience, but to provide an extra layer of safety and peace of mind. The right system should be unobtrusive, only active when needed, and never intrusive to the adventure itself. For those who want it, reliable comms can mean the difference between a minor incident and a major emergency.

 

[mountainpenguin]

Some random thoughts on meshing in underground comms.

First a little background I have played with lora meshing a bit and at once point created the time the largest lora mesh network connecting nodes on Yr Wyddfa, Scarfel and a mountain in Northern Ireland as well as a few other nodes. We did this with some off the shelf lora boards and3d printed antennas for range. This was pre meshtastic I have also used the heyphone quite a number of times with varying success.

Meshtastic is quite good however does suffer in a lot of deployments and the network has never felt to stable for me.
Meshcore has in my experience been a much better experience and feels better thought out.

I have often considered how best to combine these technologies within a cave environment especially during a rescue.
I was always struck by the amount of co ordination and how much people power went into getting messages / needs communicated effectively and how some of the meshing tech would be a good answer to these problems provided that you knew when and how to drop a repeater and how to deal with the power budgets for the comms.
I have dreamed of building small "store and forward parrot style repeaters" with memory and dual radios that would self select local TX channels to avoid interference. A user would then be "locked" to a mesh repeater while they are need by with their TX and RX tied to their local Repeater.
The user experience was very simple and cheap (think PMR with a loraboard wired into it for the user) and the nodes used the current state of the mesh to selectively activate / de activate the parrot repeaters to forward the voice messages around. Controllers would have a good view of who was near where and users could send voice comms slowly to any other user or all users with the mesh control providing locking while the "backbone" forwards messages between repeaters.

The state of uk licensing laws does put a bit of a downer on any of this experiment though as the prospect of requiring everyone in a rescue to sign up as a licensed radio amateur makes it a non starter.

 

[aricooperdavis]

The state of uk licensing laws does put a bit of a downer on any of this experiment though as the prospect of requiring everyone in a rescue to sign up as a licensed radio amateur makes it a non starter.

I'm sure this is something that could be negotiated by UKSAR along with the existing VHF bandplan if it was adopted widely enough to be considered worth bothering with.

 

[mountainpenguin]

I'm sure this is something that could be negotiated by UKSAR along with the existing VHF bandplan if it was adopted widely enough to be considered worth bothering with.

interesting I hadn't even considered that they would have the option to do that! I know the surface teams don't really have much of a problem with coverage these days so it would only be of use by the underground teams but I shall give things more of a ponder.

Would it cover things like non type approved/experimental usage ? I can not image a supplier making the kit but of the shelf parts could certainly be used

 

[Piotr Pasieczny]

Some random thoughts on meshing in underground comms.

First a little background I have played with lora meshing a bit and at once point created the time the largest lora mesh network connecting nodes on Yr Wyddfa, Scarfel and a mountain in Northern Ireland as well as a few other nodes. We did this with some off the shelf lora boards and3d printed antennas for range. This was pre meshtastic I have also used the heyphone quite a number of times with varying success.

Meshtastic is quite good however does suffer in a lot of deployments and the network has never felt to stable for me.
Meshcore has in my experience been a much better experience and feels better thought out.

I have often considered how best to combine these technologies within a cave environment especially during a rescue.
I was always struck by the amount of co ordination and how much people power went into getting messages / needs communicated effectively and how some of the meshing tech would be a good answer to these problems provided that you knew when and how to drop a repeater and how to deal with the power budgets for the comms.
I have dreamed of building small "store and forward parrot style repeaters" with memory and dual radios that would self select local TX channels to avoid interference. A user would then be "locked" to a mesh repeater while they are need by with their TX and RX tied to their local Repeater.
The user experience was very simple and cheap (think PMR with a loraboard wired into it for the user) and the nodes used the current state of the mesh to selectively activate / de activate the parrot repeaters to forward the voice messages around. Controllers would have a good view of who was near where and users could send voice comms slowly to any other user or all users with the mesh control providing locking while the "backbone" forwards messages between repeaters.

The state of uk licensing laws does put a bit of a downer on any of this experiment though as the prospect of requiring everyone in a rescue to sign up as a licensed radio amateur makes it a non starter.

Thank you for your very thoughtful observations and for sharing your experiences with LoRa mesh, Meshtastic, and Meshcore. It’s clear that you’ve put a lot of effort into experimenting and analyzing what works and what doesn’t—this is truly valuable, especially in the context of underground rescue operations, where reliability and speed are critical.

I understand that licensing and technology stability are major challenges, particularly in situations where every detail matters. In our systems, we strive to address these issues—we operate in bands that do not require individual amateur licenses, while ensuring reliable underground communication. In practice, this allows our users to focus on rescue operations, rather than administrative barriers.

I appreciate your ideas and innovative approach—these kinds of insights are exactly what help the entire industry move forward.