Models of Internet congestion

[David Nicholls]

     Perhaps Linkers have seen this already.
     
     Huberman and Lukose report on a model of Internet congestion, in an article 
     last month in Science [Huberman, B.A. and Lukose R.M., Social Dilemmas and 
     Internet Congestion, Science, vol 277, 25 July 1997, p 535]
      
     They present a model based on assuming the public good will be consumed 
     greedily if users are not charged in proportion to their use, and that this 
     will lead to congestion patterns with predictable statistical properties.  
     Their model is borne out by measurements of latencies in ping packets 
     bewteen the US and UK.
     
     The model predicts that "congestion problems as presently experienced by 
     Internet users would disappear if individuals were charged in proportion to 
     their bandwidth consumption."
     
     This idea is not new, but it's nice to see a model that matches the real 
     world, and some predictions from it.
     
     DN
     
     ASTEC/DIST