Summary

  • The 1971 ALOHAnet gave remote terminals access to University of Hawaiʻi computing over radio, with users sharing an inbound channel rather than holding a dedicated line open.
  • The satellite history has two separate paths: an operational INTELSAT IV channel on the ARPA network, and a NASA Ames experiment that used ATS-1’s VHF transponder as an ALOHA broadcast repeater.
  • The ATS-1 work tested a running system under channel noise, interference and limited received power. It widened the experiment’s reach without turning the hub into a distributed network.

The channel was shared; the network was not a mesh

In June 1971, the ALOHA packet-radio network began providing inter-island access to computing facilities at the University of Hawaiʻi. The basic economics differed from a leased circuit. Interactive users generated short bursts of input and then waited while the computer worked. A circuit dedicated to each terminal would sit idle through much of that interval.

ALOHAnet let remote terminals contend for a common radio uplink. A terminal sent a packet when it had one ready. If overlapping transmissions prevented the central station from receiving it, the terminal could not treat silence as delivery; acknowledgment and retry logic made the shared channel usable. A separate broadcast channel carried computer-to-terminal traffic. The arrangement made bursts share capacity, but it did not make every terminal a router.

The original design was a star around MENEHUNE, the central communications processor and multiplexer. Later repeaters extended radio coverage from Oahu to other islands, but they did not erase the central role of the hub. That distinction matters when the story moves into space: geographic reach and network control were separate questions.

Two satellite paths, not one

Franklin Kuo’s 1981 technical retrospective describes ALOHA’s December 1972 appearance as an operational satellite node on the ARPA network. That path used a commercial INTELSAT IV voice channel for data under a single-channel-per-carrier arrangement. Kuo says the service was billed as a voice channel, an alternative to the cost of a long terrestrial line. This was a working satellite connection, not the ATS-1 random-access experiment.

The second path began as a NASA Ames experiment. David Wax’s February 1974 status report describes University of Hawaiʻi and University of Alaska computing facilities linked to the ARPA computer network through an ATS-1 VHF link to Ames. ATS-1’s transponder acted as a broadcast repeater, and the satellite network operated in ALOHA random-access burst mode. In other words, the satellite carried packets broadcast from ground stations; the experiment was not a mesh of peer computers.

Kuo later compared the small ATS-1 VHF ground stations with the much larger standard INTELSAT earth stations then in use. He reported that ATS-1 antennas could be as small as ten feet and that the ground-station electronics cost less than $5,000, while a standard 97-foot INTELSAT station cost several million dollars. Those are Kuo’s period-specific ground-equipment figures, not a like-for-like estimate of total network cost. They nevertheless show why the experiment mattered: a satellite link could be tested with far smaller ground equipment than a conventional commercial earth station required.

The channel still had the final word

The 1974 report does not present satellite access as frictionless. During daylight tests with ATS-1 operating at full power, the project recorded an error rate on the order of 1 × 10⁻³ at 10 kbps. The report attributed remaining errors mainly to impulse noise and interference. That is one bounded measurement under stated test conditions, not a service-level guarantee. It shows that the protocol’s logic had to coexist with the radio channel’s limits.

The satellite also added timing and coordination demands. The project plan treated burst acquisition, coding, interference and the delay of interactive computer use as design problems. A packet format and protocol that worked on a local UHF link still had to be tested against a different path, transponder and ground-station arrangement. “Packet switching” described the method; it did not certify that every physical link would behave alike.

Analysis

The episode is useful because it separates reach from control. INTELSAT IV offered a commercial channel arrangement for an operational ARPA-network node. ATS-1 let the project study broadcast packet access with smaller VHF ground equipment and a different interference profile. Both extended the practical boundary of remote computing, but neither removed the organizations, gateway and central processor that decided where a packet could go.

The evidence supports a limited conclusion. ALOHA’s shared-access design could be carried into a satellite experiment, and the experiment exposed constraints that a diagram alone would miss: received power, error bursts, coding, burst acquisition and user delay. It does not establish that ATS-1 became a public service, that every remote site was connected, or that this single project caused later wireless systems. Kuo’s later discussion of Ethernet, PRNET and SATNET concerns subsequent systems with their own architectures and sponsors.

Sources