[ns] Fading and interference effect in NS2 802.11b

[Matthias Wellens]

Hi Jee,

>1. So is the CPThresh always there to be compared by every received packet?
>And the value of it is pre-defined and remains constant during the
>simulation? These make the simulation not quite accurate, isn't it?
>
No, the capture-threshold CPThresh will only be used, if a collision
occurs. For exmaple, during the reception of frame A another frame B
reaches the node. NS calculates the ratio of the received power of frame
A to the received power of frame B. If this ratio is below CPThresh, the
reception process of frame A is corrupted and thus the frame will be
marked as erroneous at the end of its reception.
Altogether CPThresh is only used for modelling collisions. Aspects such
as attenuation are modelled by the propagation model and the two other
thresholds CSThresh and RXThresh.
The value of CPThresh is constant for the whole simulation. You can
compare it to a needed Signal to Interference Ratio. It can be set by
the respective Tcl-parameter.
You're right. This implementation is not very realistic especially
because only one interfering frame is considered at the same time.

>2(non-NS2 question). So is the 802.11 series link technology reliable (I
>mean frame loss will not be detected by upper layer, whether or not FEC is
>adopted)? If there is no FEC, the sending STA will re-send a frame if it
>doesn't receive the ACK for a certain period. If there is FEC adopted, the
>receiving STA will correct an error frame anyway (If it fails to correct,
>retransmission from the sending STA will be invoked later).
>
No, a 802.11 connection is not reliable. For example node A transmits a
frame to node B. If node B can receive the frame, perhaps because of
FEC, it will send an ACK back and everything will be OK. If node B
cannot receive the frame, node A will recognise this by the missing ACK.
It will retransmit the frame for a certain number of times. If these
retransmissions cannot be received, too, node A will simply drop the
frame and will inform the routing agent about the broken link. However,
higher layer protocols such as TCP don't get informed and can only
recognise this by e.g. duplicate ACKs or a timeout.
The number of retransmissions is set by ShortRetryLimit and
LongRetryLimit. ShortRetryLimit is used for frames that are shorter than
RTSThresh, LongRetryLimit is used for longer frames, respectively.

I hope this helps,
regards,
Matthias

>>>Could anybody give me some clue that whether NS2 integrates/considers
>>>channel fading and interference effect in the 802.11b model?
>>>
>>ns2 does not consider fading. There is an extension available from CMU
>>that can be ported to the actual version of ns2:
>>http://www.ece.cmu.edu/wireless/
>>
>>Additionally, only one interfering packet is considered. If the original
>>received power is high enough, the frame will not be discarded, if it's
>>too low, the frame will be discarded. The ratio of received to
>>interfering power is compared to CPThresh and thus it's decided, whether
>>the frame is discarded or not.
>>
>>Check the mailinglist-archives for some more details.
>>    
>>
>>>Another question on 802.11 is: when there are bit errors in a received
>>>frame, what will the receiving node do? Just discards the frame and does
>>>      
>>>
>not
>  
>
>>>send the ACK? This is not a NS2 question, but I hope somebody can provide
>>>      
>>>
>me
>  
>
>>>some information. Thanks a lot in advance!
>>>
>>>      
>>>
>>A node following 802.11 or 802.11b will do this, yes. In 802.11a or
>>802.11g FEC is adopted so that the errors might be corrected.
>>    
>>

-- 
Matthias Wellens
Research Student
Department of Wireless Networks
Aachen University
Germany

Email:   mwe at mobnets.rwth-aachen.de
Address: Kackertstrasse 9
         D-52072 Aachen
         Germany