Summary: | 碩士 === 國立中興大學 === 應用數學系 === 84 === For two reasons , we adopt the fuzzy logic to control the
traffic flow .Conventionally , we linearly adjust the window to
control the traffic allowedinto the network .It may cause the
under utilization in the high speed networkdue to its
conservative control strategy. Another reason is the use of
binaryor limited number of explicit feedbacks is too sharp to
get the optimal perfo-rmance . So , we try to use the nonlinear
control ( fuzzy logic control ) themore aggressive strategy, to
acquire the more optimal performance. Using information from
both the router (the explicit congestion feed-back) , and the
sou-rce ( the window size ) ,the fuzzy control can easily
andquickly get the opti-mal operational point to improve the
performance . We use various topology to verify the fuzzy
ongestion control scheme and prove the scheme is roubst for
complex and changeable network eniroment. Moreover,by the
simulation we can get the better performance in the real-world
network model with large bandwidth-delay product and finite
source .
|