Re: [manet] bandwidth calculation at a node

["Dr Manthos Kazantzidis" <kazantz@cs.ucla.edu>]

Hi Harpreet and Christian,

 While I agree with the email, I wanted to add that there
is an issue of available bandwidth definition and measurement
method. For example if there is a local measurement
that takes into account the MAC mechanisms then one
can say that the available bandwidth can be predicted at some
value - of course this does not imply when you sent a packet
it will be received but just that under some scheduling (similar to such of
the past
if you will) if no additional flows enter the link etc the available
bandwidth will be that.

At

End-to-end versus Explicit Feedback Measurement in 802.11 Networks -ISCC02

downloadable from http://www.cs.ucla.edu/~kazantz/

I am attempting a per-neighbour measurement of available bandwidth
using the ACK times.

I am considering the neighbour and neighbour's neighborhood by combining
a source, destination specific throughput measurement
(which indirectly depends on 802.11 timers) with the node utilization.
If the paper is not very clear I can post additional slides and some
results that show that -in simulation- works very well for multimedia
adaptation, if interested.

Regards,
Manthos

PS: I have not read that paper but will do.

----- Original Message ----- 
From: "Christian de Waal" <dewaal@cs.uni-bonn.de>
To: "Harpreet Arora" <ha33@drexel.edu>; <manet@ietf.org>
Sent: Thursday, July 24, 2003 1:02 AM
Subject: Re: [manet] bandwidth calculation at a node


> Hi Harpreet,
>
> QoS assurances are not possible in 802.11 DCF networks. Period. To make
> assurances, you need something like a slotted MAC layer, because
> "simply" knowing the available bandwidth at the sending and at the
> receiving station is not enough: The stations also have to negotiate at
> what times the medium is free at _both_ stations. If half of the link
> bandwidth is available at both stations, but one station is blocked when
> the other is not and vice versa, then these stations cannot communicate
> unless some re-scheduling is performed.
>
> I can once again point you to our ASWN'03 paper which describes these
> difficulties in detail.
>
> http://web.informatik.uni-bonn.de/IV/Mitarbeiter/dewaal/aswn03.html
>
> Best regards,
> Christian
>
>
> Harpreet Arora wrote:
>
> > Thanks Philippe. This is certainly useful. However, the paper assumes
that a
> > channel is divided into time slots and reservations are made by
reserving
> > time slots. I am trying to figure out if it is possible to make QoS
> > assurances in an 802.11-DCF   based network.  In such a network, each
node
> > has to keep track of the amount of bandwidth which is being used by the
> > neighboring nodes (nodes which are transmitting within the interference
> > range of  this node) before it can make any reservations or admission
> > control decisions. Is there any work done which discusses how bandwidth
can
> > be calculated in this scenario?
> >
> > thanks,
> > Harpreet
>
>
> -- 
> ____________________________________________________________________
> Christian de Waal
>
> University of Bonn                            Phone: +49 228 73-4548
> Institute of Computer Science IV                Fax: +49 228 73-4571
> Roemerstr. 164                         E-mail: dewaal@cs.uni-bonn.de
> 53117 Bonn, Germany                    http://www.cs.uni-bonn.de/IV/
>
>
> _______________________________________________
> manet mailing list
> manet@ietf.org
> https://www1.ietf.org/mailman/listinfo/manet


_______________________________________________
manet mailing list
manet@ietf.org
https://www1.ietf.org/mailman/listinfo/manet