Performance Anomaly of b Martin Heusse, Franck Rousseau, Gilles Berger-Sabbatel, Andrzej Duda IEEE INFOCOM 2003
Outline ► Introduction ► Performance of IEEE b DCF access method ► Performance anomaly of IEEE b ► Simulation ► Performance measurements ► Conclusions
Introduction ► uses the CSMA/CA protocol to share the radio channel in a fair way ► Anomaly is the basic CSMA/CA method When a host detects repeated unsuccessful frame transmissions, it decreases its bit rate Least one host with a lower rate ► Throughput : higher rate host = lower rate host
Performance of IEEE b DCF access method No collisions proportion p t t r X = 0.7 t t r + t OV PHY p MAC p = 11 M/s * 0.7 = 7.74 Mb/s Collisions proportion p(N) = No collisions Throughput t t r t t r + t OV + t cont(N) = Fast host : T f Slow host : T s SdSd R
Performance anomaly of IEEE b U f = TfTf ( N -1) T f + T s + P c (N) * t jam * N collisions ( N -1) T f + T s + P c (N) * t jam * N TfTf TsTs N-1
Performance anomaly of IEEE b Throughput X f = U f * p f (N) * R =U f * Sd RT f * R = ( N -1) T f + T s + P c (N) * t jam * N Sd Xf = Xs = X
Performance anomaly of IEEE b ► When fast and slow hosts share the radio channel of a b cell, throughput ► The fair access to the channel provided by CSMA/CA ► Transmitting at 1 Mb/s to capture the channel eleven times longer than hosts emitting at 11 Mb/s N * X R * pf(N)Large number of fast hosts
Simulation
Performance measurements different rates, no mobility 11Mb/s 5.5 Mb/s 2 Mb/s 1 Mb/s
Performance measurements different rates, real mobility Keeps bit rate 11Mb/s
Conclusions ► b throughput is much smaller than the nominal bit rate ► Slow host may considerably limit the throughput of other hosts roughly to the level of the lower rate ► Deployment of access points