Re: [Madwifi-devel] 802.11b behavior without RTS-CTS

[Mesut A. E. <er...@wi...>]

> Sumit Rangwala a écrit :
>> Hi, 
>>    I had a question regarding 802.11b (wifi) protocol and I 
>> am not sure if this is the right mailing list for the 
>> question. In case my email is misguided please let me know. 
>>
>> 802.11b specifies the following parameter values 
>> slot time = 20us 
>> contention window range = 31-1023 
>>
>> I was trying to see what would happen in case of a hidden 
>> terminal if RTS-CTS is disabled in 802.11b. Consider a 
>> topology 
>>       A--------B-------C
>>
>> (A and C cannot hear each other else everyone can hear each 
>> other.) 
>>
>> If A and C transmits simultaneously to B the packets 
>> will collide at B. Both A and C will exponentially 
>> increase there contention window to 63 (starting from an 
>> initial value of 31). Now, in the *absence* of RTS-CTS it is 
>> required that
>>    transmission time of a packet < avg. wait time of A or C 
>>
>> i.e. if in the next round C starts transmitting before A 
>> then (on an average) it should be able to transmit a packet 
>> before A initiates a transmission again. Assume a 1500 byte 
>> packets, and assume that C start transmitting immediately 
>> after the collision, it will take 2084us for C to transmit 
>> the packet [1]. Now, irrespective of what value A chooses 
>> from 0-63, the wait time of A (0-63 * 20us) will alway be 
>> less that 2084us. So inevitably a packet collision will 
>> occur again.
>>
>> The only explanation I can guess from these parameter 
>> settings is the 802.11b assumes that RTS-CTS is used all 
>> the time and the correctness of the protocol is not 
>> guaranteed without RTS-CTS in case of a hidden node.s
>>   

Plausible, if you define performance as cumulative system throughput. 
Physical-layer capture effect (in a nutshell, a radio's ability decode 
stronger of the two colliding signals) is observed quite often in 
practice, yielding to what you mentioned. Per node throughput is a 
different ball game though (see http://tinyurl.com/27ey6b for more details).

Best,

--
MAE


> You are indeed right. 802.11b only works if all nodes can hear each 
> other. Following your theory, I'll like to know if rts-cts would resolve 
> this problem at all. Real field tests did not show much improvements.
>> Any help/feedback is greatly appreciated.
>>
>> Thanks,
>> Sumit 
>>
>> [1] 
>> http://tim.oreilly.com/pub/a/wireless/2003/08/08/wireless_throughput.html
>>
>>   
> Regards,
> Benoit
> 
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >> http://get.splunk.com/
> _______________________________________________
> Madwifi-devel mailing list
> Mad...@li...
> https://lists.sourceforge.net/lists/listinfo/madwifi-devel
>