[ns] Throughput in presence of hidden node
raysaikat at lycos.com
Tue Jan 27 11:46:55 PST 2004
Easiest way of creating hidden node is to set CSThrehold=RXThreshold.
----------- Original Message ---------
DATE: Mon, 26 Jan 2004 11:14:13
From: Joshua Robinson <jpr at is.rice.edu>
To: Lalit Kotecha <lkotecha at sbcglobal.net>
Cc: ns-users at ISI.EDU
>I've had this same problem and here's what I figured out.
>The carrier sense threshold is about 550m, while the receive threshold is
>250m. A hidden terminal is one-hop away, so it's approx between 250 and
>500m from the node it's hidden from. This means the hidden terminal is
>always within carrier sense range. The hidden terminal will then sorta
>hear the data packet from the node it's hidden from - it'll sense carrier
>and therefore a busy medium and it won't send.
>Consider this topology:
> N1 N2 N3
>where N1 and N2 are 250m apart, same with N2 and N3. Assume N1 sends to
>N2 and N3 sends to N2. Without RTS/CTS you'd expect collisions at N2
>because N1 and N3 don't know when the other is sending data. But when N1
>sends a packet to N2, N3 will sense carrier and then defer - so the
>collision won't actually occur.
>Is this really what happens? As far as I could tell the problem is with
>the way the propagation model is implemented - basically the carrier sense
>range implemented in NS is an unrealistic model. Because it works
>perfectly in NS, you have no need for RTS/CTS. To get meaningful
>results, you therefore need to implement a channel fading model so that
>the carrier sense doesn't work so damn well. See here for information
>about a channel fading model:
>Hope this helps,
>On Fri, 23 Jan 2004, Lalit Kotecha wrote:
>> Hello All,
>> I am trying to compare impact on Throughput with/without use of RTS/CTS
>> in scenario where there are some hidden nodes (in BSS - infrasture
>> mode). I am taking a simple scenario where one node is AP and there are
>> 2 nodes located at extreme boundary.
>> I do not get any sensible results. I expect that there is some
>> performace improvement by using RTS/CTS which avoids collison from
>> hidden nodes. At the same time RTS/CTS is overhead and decreases TP
>> (worst when dataRate_ is 11Mb).
>> Has anyone done similar analysis and would like to share their
Get advanced SPAM filtering on Webmail or POP Mail ... Get Lycos Mail!
More information about the Ns-users