In 2019, Rodolfo Novak sent Bitcoin transactions from Toronto to Michigan without using the internet or satellite. He used amateur radio, the 40 meter band, and the ionosphere for relays.
Nick Szabo called it “Bitcoin that can be sent across borders without the internet or satellites, using only the natural ionosphere.” Transactions were small, setup was cumbersome, and use cases bordered on absurdity.
But it proved something. The point is that the protocol doesn’t care about what carries the packets.
The experiment is part of a decade of stress tests that the Bitcoin community has been running quietly in the background, a decentralized research and development program that tests whether the network can function when normal infrastructure fails.
Satellites broadcast blocks to dishes across the continent. Mesh radios relay transactions between neighborhoods without the need for an ISP. Tor routes traffic around censors. The ham operator taps hexadecimal numbers on the shortwave.
These are not production systems. These are fire drills for scenarios that most payment networks treat as edge cases.
The question that’s driving it all is: If the internet is disrupted, how quickly can Bitcoin come back online?
Satellites give Bitcoin an independent clock
Blockstream Satellite broadcasts the complete Bitcoin blockchain 24/7 via four geostationary satellites that cover most populated areas.
Nodes equipped with inexpensive dishes and Ku-band receivers can synchronize blocks and maintain consensus even if the local ISP goes down.
This system is unidirectional and low bandwidth, but it solves a specific problem. During regional power outages or censorship, nodes require an independent and authoritative source of information about ledger status.
Satellite API extends this further. Anyone can uplink arbitrary data, including signed transactions, from a ground station for worldwide broadcasting. goTenna has partnered with Blockstream to allow users to create transactions on offline Android smartphones, relay them through a local mesh, and send them to satellite uplinks that broadcast without connecting to the broader internet.
Bandwidth is terrible, but independence is absolute.
This is important because satellites provide “out-of-band” channels. If normal routing fails, nodes scattered on different continents can still receive the same chain chips from space, providing a shared reference point to rebuild consensus after the terrestrial link returns.
Mesh and LoRa build Bitcoin backhaul at human scale
Mesh networks take a different approach. Instead of broadcasting from orbit, packets are relayed from device to device in short hops until one node with Internet access rebroadcasts them to the broader network. TxTenna, built by goTenna, demonstrated this in 2019.
Users send signed transactions over the mesh network from their offline phones, hopping from node to node until they reach an exit point. Coin Center has documented the architecture. Each hop extends your reach without requiring participants to access the Internet directly.
Long-range LoRa mesh takes this concept even further. Launched by Bitcoin Venezuela, Locha Mesh builds wireless nodes that form an IPv6 mesh over unlicensed spectrum.
The hardware, Turpial and Harpia devices, can also transmit messages, Bitcoin transactions, and perform block synchronization over kilometers without an internet connection.
Tests in disaster areas have proven successful cryptocurrency transactions over multi-hop networks where both mobile phones and fiber are down.
Darkwire fragments raw Bitcoin transactions into small packets and relays them hop-by-hop over LoRa radio. Each node will have a line-of-sight range of approximately 10 kilometers, turning an area of hobbyist radio into an ad-hoc Bitcoin infrastructure.
The range in urban areas drops to a range of 3-5 kilometers, but is enough to avoid localized power outages and censorship chokepoints.
Academic projects like LNMesh have extended this logic to Lightning Network payments and demonstrated offline micropayments over local wireless mesh during power outages.
Although the volume is small and the setup is weak, the principle is established that Bitcoin’s physical layer is fungible. As long as a path exists between the nodes, the protocol works.
Tor and amateur radio bridge the gap
Tor represents a middle ground between regular internet and exotic radio. Starting with Bitcoin Core 0.12, when the local Tor daemon is running, nodes automatically start hidden services and accept connections through .onion addresses even if your ISP blocks known Bitcoin ports.
Bitcoin Wiki and Jameson Lopp’s setup guide describe a dual-stack configuration in which nodes simultaneously route traffic on both Clearnet and Tor, complicating efforts to censor Bitcoin traffic at the ISP level.
Experts warn against running nodes solely on Tor due to the risk of Eclipse attacks, but using Tor as one of several routing options significantly increases the cost of blocking Bitcoin infrastructure.
Amateur radio is at one end of the spectrum. Beyond Novak’s ionospheric experiments, carriers have been relaying Lightning payments over amateur radio frequencies.
These tests involve manually encoding transactions, sending them over the HF band using protocols such as JS8Call, and then decoding and rebroadcasting them on the other side.
Throughput is laughable by modern standards, but efficiency isn’t the point. The key is to demonstrate that Bitcoin can travel across any medium that can transmit small data packets, including those that predate the Internet by decades.
What a global partition actually looks like
Recent modeling is investigating what happens during an extended global internet outage.
One scenario would be to divide the network into three regions: North and South America, Asia Pacific, and Europe and Africa, and set the hash rates to approximately 45%, 35%, and 20%, respectively.
Miners in each partition continue to produce blocks while adjusting the difficulty individually. Local exchanges build their own fee markets and create order books in branch chains.
Within each partition, Bitcoin continues to function. Transactions are confirmed, balances are updated, and local transactions proceed, but only within that island. Cross-border trade will be frozen. When the connection returns, the node is faced with multiple valid chains.
Consensus rules are definitive. That is, it follows the most cumulative proof-of-work chain. Weak partitions are reorganized and some recent transactions are removed from the global history.
If the outage lasts from a few hours to a day and the hash distribution is not highly skewed, the result is a temporary disruption that then converges as bandwidth recovers and blocks propagate.
A prolonged outage risks social coordination overriding the protocol’s rules and exchanges, or large miners choosing their preferred history. Yet even it remains visible and, unlike traditional economic settlements, bound by rules.
Banks don’t have fire drills for this.
Compare this to what happens when the payment infrastructure breaks down. A 10-hour outage of TARGET2 in October 2020 delayed SEPA file submissions and forced central banks to manually manage liquidity and collateral.
Visa’s pan-European outage in June 2018 completely halted 2.4 million card transactions in the UK and ran out of ATMs within hours of a single data center switch failing.
The ECB’s TARGET system also suffered a major failure in February 2025, prompting an external audit after the backup system failed to start up.
The IMF and BIS document on CBDC and RTGS resilience clearly warns that large-scale power outages or network outages can affect primary and backup data centers simultaneously, and that centralized payment systems require complex business continuity plans to avoid system disruption.
The architectural differences are important. Every Bitcoin node maintains a complete copy of the ledger and validation rules.
After any outage, as soon as you can communicate with other nodes via satellite, Tor, mesh, or your restored ISP, simply ask, “What is the heaviest valid chain?”
A protocol defines a resolution mechanism. There is no central operator to coordinate competing databases.
Banks rely on a layered, centralized infrastructure consisting of a core bank ledger, RTGS systems such as Fedwire and TARGET, card networks, ACH, and clearinghouses.
Recovery involves replaying queued transactions, reconciling mismatched snapshots, possibly manually rebalancing balances, and then bringing hundreds of intermediaries back into sync.
Visa’s 2018 outage took hours to diagnose, even with a full-time operations team. The ECB’s TARGET incident required an external review and a multi-month improvement plan.
Bitcoin practices to prepare for the worst-case scenario
Therefore, in times of crisis, plausible scenarios emerge. In the event of a fiber or mobile network failure, a subset of miners and nodes remain synchronized via satellite and radio to maintain the authoritative chain chip.
Once connectivity is restored with a patch, the local node retrieves the missing blocks and reorganizes its chain within minutes to hours.
Meanwhile, banks know which payment batches have cleared, reschedule missing ACH files, and wait for the RTGS system to complete end-of-day reconciliations before fully resuming.
This does not mean that Bitcoin will instantly “win”. Card rails and cash remain important to consumers. However, as a global payment layer, it is likely to reach a consistent state faster than a patchwork of national payment systems. That’s because we conduct continuous fire drills to prepare for global failure modes.
Ham operators extract trades on shortwave, mesh nodes in Venezuela route satellites to outage areas, and satellites broadcast blocks on dishes to the sky, but these are not production infrastructure.
These are proofs that if regular pipes break, Bitcoin has a Plan B, a Plan C, and a Plan D that includes the ionosphere.
Banking systems still treat infrastructure failures as rare edge cases. Bitcoin treats it as a design constraint.
(Tag translation) Bitcoin
