LOG-016 ·
The Neighborhood That Routes Around Me
Mesh networking explained from first principles, plus five months of data from the solar-powered radio node on my roof.
- Words
- 732
- Est. read
- 3.2 min
- Confidence
- 0.86
- Topics
- mesh, networks, explainer
There is a radio on my roof the size of a deck of cards. It cost $31, draws 0.4 W from a small solar panel, and it participates in a network with no owner, no subscription, no account, and no company. Today's entry explains the idea it runs on (mesh routing) from first principles, because mesh is the purest expression of the thesis this log exists to test, and then grades my five months of participation with real numbers.
The principle, without jargon
Every network must answer one question: how does a message get from A to B? The centralized answer is a hub. Everyone talks to the tower, the tower talks to everyone. Efficient, easy to bill, and possessed of exactly one failure that matters. When the tower dies (power cut, disaster, or a decision by its owner), every phone in the valley becomes a flashlight.
The mesh answer: every participant is also a relay. My radio hears my neighbor's radio, which hears one three streets over, which reaches the water tower node, which covers half the county. A message hops. If a node vanishes, messages route around the absence, the way water routes around a stone, and with about as much ceremony.
The trade is explicit and worth stating like an engineer: the hub buys efficiency with fragility; the mesh buys resilience with overhead. Every message that hops through my node costs airtime and battery that a tower would centralize away. Redundancy is never free. The question this log always asks is only: what does the resilience cost, and is the price fair?
What I actually run
The node is a LoRa radio running Meshtastic, an open-source mesh firmware. LoRa is slow by design: my measured throughput is on the order of one text message per second, not one video. This is not an internet replacement. It is a communications floor: the layer that still exists when the tower does not. Range per hop, measured from my logs: 1.1 km to a suburban neighbor at ground level, 14.7 km to a hilltop node, line of sight.
Five months of node statistics:
| Metric | Value |
|---|---|
| Nodes heard, cumulative | 63 |
| Nodes heard in a median week | 19 |
| Messages relayed for strangers | 4,112 |
| My messages delivered via others' relays | 1,006 |
| Node uptime (solar + battery) | 99.1% |
| Total cost, hardware and mounting | $54 |
| Subscription | none, structurally |
The ratio in rows three and four is the whole philosophy in two numbers. I carried four packets for my neighbors for every one they carried for me. In a hub network that asymmetry would be a billing problem. In a mesh it is just weather: the network does not meter contribution, it only requires that enough people make one, and 63 strangers in my county apparently do.
The honest limits
A skeptic's paragraph, as always. Mesh networks have a scaling ceiling: every relay repeats traffic, so a dense mesh can drown in its own helpfulness (the technical term is broadcast storm, and the firmware's hop limits exist because the problem is real). Coverage is participation-shaped, meaning it is excellent near enthusiasts and absent elsewhere; my map has a hole precisely where the golf course is, which tells you something about somebody. And LoRa's bandwidth will never carry your video call. The correct mental model is not "free internet." It is "the neighborhood can still pass notes when everything else is dark."
That capability was tested for real during last spring's windstorm, when 40,000 households in the county lost power and a number of cell sites went to battery and then went quiet. The mesh stayed up (solar nodes do not care) and carried, among other traffic, a genuinely useful stream of which roads were blocked. Sample size: one storm. I log it as anecdote, not proof. But the anecdote pointed the right direction.
The forecast
Node counts on the public Meshtastic map have roughly doubled yearly for three years. Extrapolating enthusiast curves is how people embarrass themselves, so I will make the narrow claim instead: I put 0.86 on my county's mesh being able to relay a message across town, without any commercial infrastructure, five years from now. It can today, and the trend line and the $31 price both push one way.
The tower is efficient. The mesh is unkillable. Own a piece of both, and know which one you are standing on at any given moment.