Skip to content

Paul Baran

Notes from class

RAND founded in 1946, for the airforce. a think tank. They all recognized that robust communications were necessary to survive (ha) a nuclear war/error. But how to achieve this was up in the air. This was the quesiton that engineer Paul Baran started exploring from 1959

Baran's original system was the simplest system: go/no go via am radio to commanders in the field: nuke or no nuke.

Military wants rather more than this; back to drawing board.

Regular Communication 'Distributed Communication'
hierarchy - call to local office <-> regional <-> national <-> regional and back nodes and switches
destroy local/regional/national office, comms cut off destroy node, still other paths

Controlling data flow - "message switching"

  • ­think 'post office': one integral message, label w/ origin and destination, passed from station to station
  • ­each station (post office) holds on to it until it can be forwarded to the next node/destination
  • ­telegraph system does this. At first manually, by 1960s, using computers to store/route messages
  • ­these also lets you even out the flow - messages can be stored until there's space to move it (makes the line more profitable)
  • ­Baran's system isn't hierarchical; key is that the switches can reroute as necessary
  • a distributed communications network, preceeding Baran's work! AT&T and the Army; military voice network built on top of the existing civilian network. Went online with 10 nodes in 1964. Eventually extended around the world. Continues in service until the 1990s

    AT&T 'hardened' exchanges, and connected things up in a grid, in contrast to hierarchical civilian lines. Distributed nodes, but those nodes centrally controlled, telephone switches. Like a traffic control system. Manually re-routing things as necessary. Baran's system control itself was distributed: the node was a computer making its own decision about where to send the message next.

    Computerizing the switches and giving them the autonomy to decide where to route things: that's the big idea. nodes along the line can regenerate the signal; in the analogue approach, this would actual degrade the signal.

    People trained in telephony scoffed at Baran.