1. When protocol is running duplicated packets are detected even without
any handoff. Why is that?
You need to turn off the "ip forwarding" on the nodes.
This function enables the nodes to run as router. Cellular IP uses its own
routing cache to forward packets. Just type in
sysctl -w net.inet.ip.forwarding=0
2. What handoff scheme is used if Mobile IP is used with Cellular
IP?
It's using Eager Cell Switching as in standard Mobile IP. The CIP gateway
implement FA functionalities to perform periodic operations such as broadcasting
Agent Advertisement.
However, the MIP functionalities is overlay on top of CellularIP routing,
therefore the actual handoff is done by cellular IP's handoff controller
implemented in the daemon that is running at the mobile host.
3. Then what IP address does the mobile host use? the FA or the
base station?
Mobile user use FA's IP address as the care-of-address just
as in standard Mobile IP.
4. Your web page says that any standard MIP implementation will run
on top of cellular IP. Please explain more about it.
Our implementation is just one of the many ways of the realization of the
protocol. Cellular IP can work with any standard MIP implementation. Our
released implementation demonstrates one of the simplest way to incorporate
mobile IP to achieve macro-mobility.
When macro-mobility support is turned on during the compilation of CIP, the GW
in Cellular IP is also a FA in the visiting network. Therefore it is not needed
to run another Foreign Agent (which provided by your Mobile IP implementation)
in the visiting access network. You need to have Home Agent and MH daemons
running though. Noticed that our current implementation of the GW only include
limited FA's functions such as Ads Advertisement and Detunneling, other
extensions such as security is not included.
When doing a macro-handoff, it fall back to MIP, CIP codes does not do anything
in setting up the new tunnel involving Home Agent. (it is transparent to MIP,
also keep in mind that Cellular IP is a local mobility solution to complement
but not to replace MIP in macro-mobility management)
5. If MIP implementation is to co-exist with your Cellular IP implementation,
then what will happen when my MIP implementation is trying to change the default
route when it receives the FA's advertisement.
Once again, Cellular IP is a generic local mobility solution optimized for
micro-cell environment. It does not tied to specific wireless technology. Our
released implementation, on the other hand, was designed to work with 802.11
wireless access network, therefore the implementation is L2 specific while the
protocol itself is independent of the wireless technology. In current
implementation, or I should call it our 802.11 implementation of CellularIP, the
code is working on the "Ad-hoc demo" mode (mostly for simplicity),
since
there are no association between MH and the Access Point in this mode, there is
really no actual L2 handoff when MH talk to different Access Point. Therefore
the handoff controller in CIP MH simply change the default route to force the
packet flow to the new BS as a means of "handoff". Noticed that in an
older version of wavlan card which can operate on different frequencies,
the handoff controller simply switch the card's frequency and does not need to
touch any L3 route in the MH.
So, back to your question, if you are using standard 802.11b DSS cards in
ad-hoc demo mode, then you may need to disable the changing-default-route in
your MIP code to avoid conflict since CIP code will handle it correctly. We'll
add this note on the web to avoid confusion. One of the reason that we provided
limited notes regarding working with MIP is that there are too many different
flavor of MIP implementations available today and almost all of them have more
or less different approaches. For example the one we used in our testbed does
not have this problem.
