Re: wireless chapter added to Greis' Tutorial

[Mickael Vera <Mickael.Vera@crm.mot.com>]

Hi, 
this mail is addressed to the new maintener of the tutorial.

It was a very good news to learn that you added these chapters to the
Tutorial. Sometimes I wonder if the NS team realizes how hard it is to
understand NS and that there is a real need of documentation as it
appears
every day in the mailing list.

I read your chapters and as you suggest it, I send you my suggestions.
They
are not critics and I hope they'll help you to improve your tutorial. I
told
you everythnig that may confuses the user (me with my little experience
of
wireless in NS) and everything that sould be precised. Many answers may
be
found easily but that doesn't mean they shouldn't be in this tutorial.

There are a lot of suggestions but I found the Tutorial very good !  

****************************************************************

Is it possible to download the tutorial for printing ? I don't know how
you
generated this tutorial but it would be great to have an html document
in a
single page.
 
It is very good that you give the complete API as it is difficult to
find it
elsewhere.

Maybe you should say which daily snapshot to use, maybe the date
release. It
would be a good thing to add identifier to the release of daily
snapshots,
they are all called ns2.1b6, they could be called ns2.1b6-date.

X.1. Creating a simple wireless scenario 

You use the "val" array variable, does the simulation works if I use a
variable
array called "myval" ? It is very confusing sometimes that objects uses
global
variables and the user doesn't know it. (for example the variable chan
for the
base-station in the previous snapshot)

Does it work if I change the name "ns_" to "myns" ?

Does it work if I change the name of "tracefd" (it was used by objects
some
time ago) 

What makes the command "set node_($i) [$ns_ node ]" build a mobile node
? Is
it just the -adhocRouting option ? Is it possible to create a node that
doesn't use any of the adhoc protocols, like a LAN for example ? With
the
previous version it was possible.

If I want to create a wired node after a wireless node how to do ? does 
"$ns_ node-config" sets default options for the node creation if it has
already been called with options?

Does the user need to use the "node_" array in the instruction
"set node_($i) [$ns_ node ]" ? I think it is used by the node movement
scenarios but is it used elsewhere ? 

No channel is created, where is it created and how do all mobile nodes
share
the same channel ? Is it possible to have two sets of mobile nodes, each
one
using its own channel (with this new API) ? It was possible with the old
one,
I managed to make two sets of dsr nodes with two sets of base stations
using
MIP (two HA and two FA). How to do thiz now as we don't create channelos
anymore ? (and you say it is not possible to use a base station with dsr
now
?) 

idem for the prop, only topo is shared by all modes ?

node movements : whatif the node leaves the topography ?

nodes reset : I think it is to difficult to see that the time at which
nodes
reset and the one at which ns stops are different, maybe use
150.000000001 and
150.000000002 . Maybe explain why you use such numbers

IX.2. Using node-movement/traffic-pattern files and other features in
wireless
simulations  

"$god_ set-dist 1 2 2" : How to define the shortest path for several 
pairs of
nodes. What is the default shortest path ?

It would be great to have a better documentation for traces, the one
included
in the ns Documentation and Notes is not good. The reader has to read
C++
code, it would be better to have each entry described with all possible
values. 

X.1. Creating a simple wired-cum-wireless scenario 

Just for information, I used to build base stations with the command 
create-base-station which could use the dsr routing protocol, this was
and
still is the example in ~ns/tcl/example/wired-cum-wireless.tcl. It
doesn't
work anymore ? Is that function still used for the base-station ?

One time you use the val array and opt one. It would be better to always
use
the same name.

The variables num_wired_nodes and num_bs_nodes are used only by the user
or
the ns librairies use them ? Can I not use them at all ?
  
################################################################
I think there is not enough documentation about addresses of base
stations,
which rules should respect base stations and mobile nodes for their
respectives addresses and domains ?

Extract from your Tutorial : 
"For mixed simulations we need to use hierarchical routing in order to
route
packets between wireless and wired domains (see ns Notes&Doc)."

Extracts from the ns Notes&Doc: 
<< So a node called BaseStationNode is created which plays the role of a
gateway for the wired and wireless domains. >>
What are domains ? Is it the same as the 1st number of hierarchical
addresses
? All wired nodes should be in the same domain (address) ?

<< The basestation node is responsible for delivering packets into and
out of
the wireless domain.>>
Which domain, the set of wireless nodes or the address domain number ?

<< Each wireless domain along with its base-station would
have an unique domain address assigned to them. >>
A number ?

If you could answer me this question, I'd be very pleased.
################################################################

After you explained the addresses structures you say to look above for a
figure but there is not any figure about addresses.

When you create wired nodes, you use this command :
          "set W($i) [$ns node [lindex $temp $i]]"
you should use "ns_" instead of "ns"
I don't understand how it can work ?
Maybe "ns" is created in a sub-function but it is not a nice behaviour
of OTcl
about identifiers viewability

X.2. Running MobileIP in a simple wired-cum-wireless topology

Use of opt() instead of val()

In a previous version of ns-2.1b6 there was this code :
<< 
   ## Default settings
   set num_wired_nodes    2
   set num_bs_nodes       2
   set num_wireless_nodes 3
   set opt(nn)            5            ;# total number of wireless nodes
>>

now you write:
<< set opt(nn)             1           ;# just one MH >>

Something has changed, or something is wrong

<< # note address of MH indicates its in the same domain as HA. >>
                               ^^^
Is it necessary to give the mobile node the same address domain as the 
HA ? (this is not really a question but I think it should be in the
tutorial)

>From the settings of the mobile node destination to the end of the
script "ns"
is used instead of "ns_"

It is not possible to download the files  wireless*.tcl. (but it was
possible
earlier today)

I hope you'll answer me,

-- 
Vera Mickael Stagiaire