The physics
LoRa is long range and very low bandwidth. A mesh built on it — Meshtastic, which both OmilooS and PompoOS use — moves packets between nodes by radio. The default forwarding strategy is flooding: a node that hears a packet repeats it, so it reaches everyone.
Flooding has one property that kills scale: every node transmits every packet. Add nodes and you add repeaters, not capacity. The whole network shares one radio channel's worth of airtime, so a network of one node and a network of 7,000 both carry about 0.7 messages per second in total. Ten million users would get one message each every few months.
Routing
With routing, a packet travels only along a path toward its destination. Nodes not on the path stay silent. Each geographic cell's airtime carries that cell's traffic, so capacity grows with the network. The PompoOS analysis for a country the size of Kenya puts the difference at 6,928× — from 0.7 messages per second for everyone to 421 million messages a day.
Why it is not solved
Routing needs each node to know something about topology — which neighbour is closer to the destination — and topology in a mesh of phones and gateways changes constantly. Route discovery costs airtime. Routing tables cost memory on tiny devices. Meshtastic ships flooding with hop limits because it works at the scale of a hiking group; it does not have a routing layer that works at the scale of a city, and neither does anyone else on LoRa.
This is the one genuinely open problem in PompoOS, and it is shared with OmilooS's mesh: a solar-inverter gateway network across a region has the same airtime budget.
What the ecosystem brings to it
- Two products with the same problem, so a solution serves both.
- Fixed infrastructure nodes in OmilooS (inverter gateways do not move), which makes partial topology knowledge tractable.
- A registry —
pallet-device-registry— that could hold a node's declared position and role for nodes that do have connectivity, seeding routing for those that do not. Optional; the daily path must work offline.
What the ecosystem does not claim
A solution. A deployed network. Coverage figures. The previous version of this page described a mapping network that rewards contributors with tokens; PompoOS is explicitly tokenless so that the incentive problem does not compound the routing problem.
For a researcher
This is a real, bounded, measurable problem with a factor of ~7,000 on the table. If the grants programme opens with its infrastructure category, this is the kind of pilot it should fund.
Based on the PompoOS project analysis (2026-09-09) and the OmilooS mesh design.