MeshCore and the Case for Communications You Actually Own

There is a version of self-sufficiency that most people think about: growing food, storing water, generating power. And then there is a version that gets far less attention: communication. The ability to send a message, coordinate with people nearby, and stay informed when the normal channels are unavailable.

Most of our communication infrastructure is borrowed. The phone in your pocket works because a mobile network operator has towers in range and the internet backbone is intact. The apps on it work because servers are running in data centres owned by companies that can, and occasionally do, go offline. When any part of that chain fails, your ability to communicate fails with it. You do not own the pipe. You just use it.

MeshCore is one answer to what happens when you decide to own the pipe instead.

What MeshCore is

MeshCore is open-source firmware for LoRa radio hardware that lets devices form a mesh network with no internet connection, no mobile coverage, and no central server required. Flash it onto an inexpensive LoRa board, pair it with your phone over Bluetooth, and you can send encrypted text messages to other MeshCore nodes within range, or across multiple hops to nodes further away.

The project was created in 2024 and developed rapidly after Cyclone Gabrielle devastated parts of New Zealand, where the practical need for infrastructure-independent communication was demonstrated with some force. It is published under the MIT licence and runs on hardware that costs between £15 and £40 per node.

It is worth distinguishing MeshCore from LoRaWAN, since this series covers both. LoRaWAN is the network protocol this series has been building infrastructure around: sensors, gateways, ChirpStack, Home Assistant. It is optimised for small, infrequent data packets from fixed or mobile sensors. MeshCore uses the same LoRa radio technology but operates differently: it is a peer-to-peer mesh protocol designed for human communication, not sensor telemetry. The two sit in the same frequency bands and run on overlapping hardware, but they are solving different problems. LoRaWAN is for getting data from things. MeshCore is for getting messages between people.

How it works differently from Meshtastic

If you have looked at off-grid LoRa communication before, you have probably encountered Meshtastic, which is the more established project in this space. MeshCore is worth understanding as a distinct thing rather than a Meshtastic clone.

Meshtastic uses broadcast flooding: every node rebroadcasts every message it receives, up to a configurable hop limit. The advantage is that it is simple and works without any infrastructure planning. The disadvantage is that in a dense network, every node spends a lot of radio time rebroadcasting messages it was not the intended recipient of, which creates congestion and drains batteries.

MeshCore takes a more structured approach. The network distinguishes between node types: companions are personal devices that send and receive messages but do not relay for others; repeaters are dedicated infrastructure nodes that extend coverage; and room servers provide store-and-forward group messaging for people who are temporarily offline. When a companion needs to reach another node, MeshCore floods once to discover a route, learns it, and uses that route for subsequent messages. The radio channel stays quieter, battery life is better, and the network scales more gracefully as it grows.

The tradeoff is that MeshCore requires intentional infrastructure. Handing someone a MeshCore companion node and telling them to go communicate works well if repeaters are in place. Without them, two companions need to be within direct radio range of each other. Meshtastic is more forgiving in truly ad-hoc situations. MeshCore rewards planning and investment in fixed infrastructure.

For a self-sufficiency context, that tradeoff tilts clearly toward MeshCore. If you are building a network deliberately, with repeaters placed at sensible high points and room servers providing resilient message storage, MeshCore gives you something more robust and more efficient than a flood-based network. The planning required is not a burden; it is the point.

The hardware

MeshCore runs on ESP32-based LoRa boards, the same family of hardware that LoRaWAN sensors and gateways in this series are built around. Boards from Heltec, LILYGO, RAK Wireless, and Seeed Studio are all supported. Flashing firmware is done from a browser via the MeshCore web flasher, which takes a few minutes per device.

Entry-level companion nodes, a Heltec WiFi LoRa 32 V3 for example, cost around £20 and run for days on a small battery. A solar-powered repeater node on a rooftop or hilltop runs indefinitely once placed. The LILYGO T-Deck gives you a standalone device with a small screen and keyboard that does not need a phone at all, which is the hardware equivalent of owning the whole stack.

The total cost of a useful network, a couple of companions and a repeater or two, is well under £200. The cost of the same capability from any commercial provider is a monthly subscription that disappears the moment you stop paying.

What self-sufficiency actually looks like

A network built on MeshCore has no single point of failure that someone else controls. The repeaters run on hardware you own. The firmware is open source. The messages are encrypted end-to-end with AES-128, and the keys are yours. There are no accounts to create, no terms of service to accept, no company that can shut it down or hand your message history to someone else.

More practically: it works when the mobile network does not. It works when the internet is down. It works when the power grid is out if your repeaters are on solar. It works across distances that make handheld radio impractical and at scales that make licensed spectrum unnecessary. In the unlicensed ISM bands, anyone can deploy it.

The range numbers are honest but context-dependent. Direct line-of-sight between two nodes can reach ten kilometres or more. In a town with buildings and terrain in the way, two to five kilometres is more realistic for a companion-to-repeater hop. With a well-placed repeater network, the effective range of the whole mesh is the sum of its hops, not the range of any single link.

Where it fits alongside the rest of this series

The LoRaWAN infrastructure this series has been building, the gateway, ChirpStack, Home Assistant, the sensor network, is about making a property smarter and more observable. MeshCore is a complement to that, not a replacement for it. One is for machine data; the other is for human communication. Both use LoRa. Both run without cloud dependencies if you build them correctly. Both are things you own outright once they are deployed.

The practical overlap is hardware and frequency. If you already have LoRa equipment in hand for sensor work, you are one firmware flash away from having a MeshCore node. The Heltec boards used in this series run both. Whether to run them as dedicated LoRaWAN sensor nodes or MeshCore companions is a question of what you need from each piece of hardware, and the answer can be both if you have enough boards.

The honest limits

MeshCore is not a replacement for all communication. It does not do voice. It does not do high-bandwidth data. Messages are short text, and the protocol is designed for that constraint rather than fighting it. If you need to stream video or make a phone call off-grid, you need different hardware entirely.

The project is also young. Documentation is improving but still patchy in places. The community is active and the development pace is fast, which means things change between firmware versions. This is the normal state of open-source projects at this stage of their life.

And repeater placement matters in a way that is easy to underestimate. A companion node in a valley with no line of sight to any repeater is a capable device that cannot reach anything. The network is only as good as its infrastructure, and infrastructure requires thought, physical access to high points, and some weatherproofing for anything that lives outside permanently.

None of these are arguments against building it. They are things to plan around.

A different kind of infrastructure

The version of self-sufficiency that most people pursue stops at the edge of their property. Food, water, power: all things that can be stockpiled or generated locally. Communication is different because it is inherently about connection to other people. You can be completely self-sufficient on a property and still entirely dependent on infrastructure you do not control the moment you want to tell someone you are okay.

MeshCore is not a complete answer to that. But it is a serious one, built on open hardware, open software, and a protocol designed from the start for the situation where the normal infrastructure is not there. For the cost of a few evenings and a handful of LoRa boards, it gives you something that belongs to you and works on your terms.