[ns] WLAN-Infrastructure simulation problem

[Joseph Cosmas]

Dear ns-user,

We are simulating 802.11b WLAN roaming with two wired nodes, two base
station and one MN (with 914MHz Lucent WaveLAN DSSS radio interface) as
follows:
                                                ____
                                               | W0 |
                                                ------
                                                   |
                                                   |
                                               ____
                                               | W1 |
                                               ------
                                            /             \
                                          /                 \
                                        /                     \
                                  ____                    ____
                                 | HA |                    | FA |
                                 ------                     -----
                               (175,150)              (275,150)

   ____
   | MN | ------> to (620,80)
   ------ 
 ( 20,80)

HA and FA are in different domain with MobileIP support using NOAH
routing agent. Initially MN is in HA-domain but it is created to move
between HA- and FA-domain.  We then setup a TCP connection between W0
(source) to MN (sink). The problem with our simulation is that once the
MN start moving it receive TCP packets through HA, then after some
seconds roam to FA, but awkwardly change back to HA, then again roam to
FA, then back ...; actually not stable as theories concerning effect of
distance against signal strength suggest. Does anyone experience same
problem before? What could be the possible source and solution to this
problem?

Thanks in advance,
Joseph.