Finding Low Diameter Low Edge Cost Networks

This paper describes a simulated annealing algorithm to compute k connected graphs that minimize a linear combination of graph edge cost and diameter Repli cas of Internet information services can use graphs with these properties to propagate updates among themselves We report the algorithm s performance on and connected node graphs where edge cost and diame ter correspond to physical distance in a plane For these graphs the algorithm nds solutions whose edge cost and diameter are and lower than our graph construction algorithm that returns feasible but unop timized solutions When optimizing edge cost only the resulting graphs show and reductions edge cost and diameter respectively