Why a 10% Airtime Mesh Network Will Always Hit the Wall

Airtime is not a performance setting. In Europe, on 868 MHz, it is the wall.

Meshtastic and MeshCore did not fail because LoRa is bad.

They did not fail because mesh networking is impossible.

They failed, again and again, because people tried to build a real mesh backbone inside a legal framework that was never designed for that job.

The culprit is not mysterious.

It is the 10% airtime limit.

For a quiet personal tracker, 10% duty cycle sounds generous. Six minutes per hour is a lot if the device sends an occasional position, a short text, or a sensor reading.

But a mesh relay is not a quiet personal device. A mesh relay forwards other people’s traffic. It repeats packets, handles adverts, relays messages, participates in routing, and often sits on a high site where everyone can hear it and everyone wants to use it.

That is where the 868 MHz dream breaks.

Either the node respects the 10% airtime limit and the network begins to stall, delay, and drop traffic, or the operator disables the limit and the network may appear to work — while moving outside the normal European SRD framework.

In other words, the network collapses, or the network goes illegal.

The two airtime budgets people confuse

There are two different airtime budgets in a LoRa mesh.

The legal airtime budget

The first is the legal airtime budget.

In Europe, the 868/869 MHz SRD bands are not amateur-radio spectrum. They are short-range device allocations with specific power, bandwidth, access method and duty-cycle limits.

In the commonly used 869.4–869.65 MHz sub-band, many LoRa mesh users rely on the well-known 10% duty-cycle rule.

That means one device may transmit for only a limited fraction of time. It does not mean that a complete mesh network gets guaranteed usable capacity.

The RF channel airtime budget

The second is the RF channel airtime budget.

This one is not legal. This one is physics.

If enough nearby nodes transmit on the same channel, the channel becomes busy. Packets collide. Retries increase. Delays increase. The mesh becomes slower and less reliable.

The 10% legal limit does not coordinate a mesh. It only limits one transmitter.

Ten nearby nodes each transmitting legally at 10% can already occupy the channel for 100% of the time in the same RF neighbourhood.

At that point, the fact that each individual node is “legal” does not make the mesh usable.

A shared LoRa channel needs spare airtime, randomness, quiet periods and controlled retransmission. It cannot be planned by giving every box its own 10% slice and pretending the slices do not overlap.

That is the trap.

The older articles were already pointing at this

This is not a new problem. It is the same issue that keeps returning from three different angles.

The article Meshtastic, MeshCore and the Legal Framework made the basic point: every transmitted packet belongs to a legal framework. A LoRa packet is not automatically amateur radio because the operator has a callsign. A board is not automatically legal because it runs Meshtastic or MeshCore. A CE logo is not a full compliance file. And a 10% airtime setting is not a magic legal shield.

The article Meshtastic, MeshCore, 868 MHz and the Ham Radio Trap focused on the dangerous European assumption: “I am a radio amateur, so 868 MHz is ham radio for me.” It is not. The common 868/869 MHz LoRa world in Europe is SRD spectrum, not an amateur-radio band. If the device is operating as an SRD, then the SRD rules apply.

The article Meshtastic, MeshCore, CE Marking and the Hardware Trap added the other half of the problem: cheap LoRa boards, high-power modules, external antennas, firmware settings and webshop CE logos do not automatically produce a compliant station. The real question is whether the exact product, in the exact configuration, is compliant for the intended use.

Put the three together and the pattern becomes obvious.

A hobby mesh on 868 MHz tries to behave like amateur infrastructure, but it is sitting inside an SRD framework.

The moment it becomes useful as infrastructure, it starts to need more airtime, better antennas, higher sites, more relay behaviour, more forwarding, more coverage and more coordination.

Those are amateur-radio instincts.

But 868 MHz SRD is not the amateur service.

Why a real mesh eats airtime faster than people expect

A point-to-point LoRa link is simple. One node talks, one node listens.

A mesh is different.

In a mesh, one user message may become several transmissions. A relay hears it and forwards it. Another relay may forward it again. Other relays may hear the same packet and suppress duplicates, or delay retransmission to avoid collisions, but the network still spends more than one packet’s worth of airtime to move one message across multiple hops.

That is not a bug. That is what makes the mesh work.

So the problem is not simply “a packet takes airtime”.

The problem is that a public relay’s airtime is not only its own airtime.

It is everybody’s airtime.

A private node can stay quiet. A backbone node cannot. A high-site relay becomes the station everyone wants to use.

The better the site, the more traffic it hears. The more traffic it hears, the more it must decide whether to forward. The more it forwards, the faster it reaches the airtime wall.

On 868 MHz SRD, that wall is not just technical.

It is regulatory.

MeshCore makes the fork visible

MeshCore makes the issue especially visible because the duty-cycle behaviour is configurable.

That is technically honest. It allows the operator to configure the system for the correct band, country and legal framework.

But on European 868/869 MHz SRD, it also exposes the unavoidable fork.

In the normal hobby SRD situation, “it works better” is not that lawful basis.

This is why the problem keeps returning in every 868 MHz mesh discussion.

The community wants infrastructure.

The band allows short-range devices.

Those are not the same thing.

The correct answer is not “ignore airtime”

The answer is also not to pretend airtime no longer matters.

A ham-only mesh on an amateur allocation still needs discipline. It still needs low power, good sites, clean transmitters, filtering where needed, sensible hop limits and a willingness to reduce power or change configuration if interference appears.

The difference is that amateur radio can legally support amateur-radio behaviour: callsign identification, coordinated frequencies, non-encrypted technical traffic, responsible experimentation and properly authorised unattended relay stations.

That is why the RFC matters.

The RFC: move the mesh into the amateur service

The proposed RFC moves the idea away from 868 MHz SRD and into a coordinated 70 cm amateur-radio framework for IARU Region 1.

The RFC is available on GitHub at https://github.com/Guru-RF/meshcore-rfc-iaru-r1.

The purpose is simple: a ham-only MeshCore / LoRa digital mesh layer using a single coordinated 70 cm frequency, avoiding fragmented and uncoordinated use of 433 MHz and 868 MHz devices by licensed radio amateurs.

The proposed channel sits in the narrow window above the 433 MHz ISM/LPD433 area and below the 435.000 MHz amateur-satellite boundary.

The RFC also requires that no station emit above 435.000 MHz.

This is not a proposal to create a free-for-all.

It is the opposite.

The RFC makes the regulatory point that fixed relaying nodes are not just “personal radios on a hill”. A fixed node that retransmits automatically is an unmanned amateur station in the repeater sense and needs the appropriate authorisation and coordination.

That requirement is not a burden.

It is the point.

A legal, coordinated mesh should look legal and coordinated.

Density calculation: Belgium has enough operators

Now the practical question: is there enough amateur-radio density to support such a network?

Assume roughly 5,000 Belgian radio amateurs and a Belgian surface area of approximately 30,690 km².

That gives:

The average square-grid spacing is:

Real life is not evenly distributed. Operators cluster around cities, towns, club areas, repeater sites and transport corridors. The Ardennes will not look like Antwerp. Brussels will not look like rural Luxembourg province.

But for a mesh, clustering is not automatically bad.

Mesh networks like density, as long as the RF channel is managed.

Neighbourhood density

Using the same average density, the number of potential radio amateurs within a given radius becomes:

Not all of them will be active. Not all of them will own MeshCore hardware. Not all of them will be in RF range because terrain, antenna height, building loss and local noise matter.

But the density is clearly sufficient.

The bottleneck is not “Belgium does not have enough hams”.

The bottleneck is using the wrong legal framework.

How many fixed relays would Belgium need?

A rough geometric relay-spacing estimate is also useful.

This is not a propagation prediction. It is only a planning sanity check.

For a triangular or hex-like mesh, the coverage area per relay site is roughly:

Using Belgium’s 30,690 km²:

That means a national Belgian MeshCore layer does not require 5,000 fixed relays.

It needs a coordinated backbone of tens to low hundreds of authorised fixed nodes, plus portable and attended user nodes around them.

That is a realistic amateur-radio project, especially if existing repeater sites, club sites, high buildings and well-sited private stations are used carefully.

The RFC already points in that direction with an initial Belgian test bed of authorised unmanned amateur stations.

Again, the GitHub RFC is here: https://github.com/Guru-RF/meshcore-rfc-iaru-r1.

Why it can hold without the SRD airtime wall

Without the SRD 10% airtime ceiling, the network no longer has to choose between usefulness and compliance.

That does not mean every node should transmit endlessly.

It means airtime can be engineered instead of artificially capped for the wrong service.

On a coordinated amateur channel, the limiting factors become:

• traffic discipline;
• hop limits;
• routing scope;
• relay density;
• transmitter cleanliness;
• power control;
• filtering;
• site coordination.

Those are engineering problems.

The 868 MHz 10% limit is not an engineering problem.

It is a regulatory boundary.

A real mesh needs bursts. It needs relays that can forward when forwarding is needed. It needs a backbone that can survive a busy period without every important relay hitting “duty cycle reached” and going silent.

It needs enough RF freedom to behave like infrastructure, while still being modest, narrow, clean and coordinated.

That is exactly the argument for moving ham-only MeshCore into the amateur service.

The real conclusion

The European 868 MHz mesh problem was never only about software.

Meshtastic can enforce 10%. MeshCore can be configured to 10%. A user can lower power, reduce adverts and be careful.

But if the goal is a real amateur mesh backbone, the 868 MHz SRD framework remains the wrong box.

Respect the 10% limit, and the network eventually hits the wall.

Bypass the limit, and the network may leave the legal framework.

That is why the next step is not another workaround, another default or another argument about whether 10% is enough.

The next step is the RFC.

A mesh that behaves like amateur infrastructure should live in the amateur service.

Not in the 868 MHz trap.

Interested in more technical content? Subscribe to our updates for deep-dive RF articles and lab notes.

Questions or experiences to share? Feel free to contact RF.Guru for practical RF and antenna support.