[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

(1) mobile nodes and (2) possible protocol stack?



Hi,

I've got the following two questions on which I'd appreciate some input
(probably the ns-developers know about this).

1- I've read that the 'node' concept is being revised to deal with the
different types of nodes better. Second, in the latest snapshot, MobileIP
and the CMU wireless nodes are integrated, in the 2.1b5 (release) not yet.
Given my limited amount of time (thesis, 2 more months) and the fact that I
need to use wireless nodes and Mobile IP, would you advise me to use 2.1b5
and make my own adjustments, use 2.1b6-current or even wait for the 2.1b6
release? I also reckon there're still errors in the snapshot versions.

2- I don't know who is familiar with GPRS ('the packet data extension to
GSM') but the protocol stack I want to implement in ns looks like the
following (simplified):


TCP
TCP
-------
------------	
IP
IP	--- Internet--          IP
--------
------------	        	          -------------			
L3				       L3  / L3x               L3x

-----				   --------------------	        ------------
Ethernet
L2		L2     / L2x	       L2x  / IP                IP

-----		------------------        --------------------
------------
Radio		Radio / fixed            fixed / fixed           fixed


Mobile		         GPRS NETWORK
(fixed) IP-host
Station

My problem is the double IP-layer in part of the GPRS network. The lower
IP-layer is used to perform routing inside the GPRS network and is in no way
related to the 'normal' IP-layer. I must be able to model the upper-IP layer
and it's TCP agent in ns to investigate TCP's behaviour in the wireless
network (this is just the 'normal' ns-way). In addition, I also need to see
the IP-packets from the lower-IP layer (or the L3x packets) since they tell
me alot about mobility management in this network. Therefore, I'd like to
put this lower-IP layer in ns as well (instead of implementing it in my own
C++ code).

- Is this possible? Is it possible to use 2 IP-layers in ns, thereby not
'confusing' ns when it comes to routing ?. 

b.t.w. The upper-IP layer will be 'replaced' by Mobile IP and the L3x layer
will be removed in my 2nd scenario, thereby avoiding this problem. I tell
this because I know Mobile IP uses IPinIP encapsulation, so I figured
something like this should also be possible for 'IP over L3x over IP'


I'd be very glead with any sort of input,

regards, Sjoerd Janssen