Sample Rate Implementation

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Sample Rate Implementation

Bishal Thapa
Hello,
  I have a small question regarding the implementation of Sample Rate in "madwifi version 0.9.4" concerning Sample Rate adaptation algorithm's coupling with MRR (Multi Rate Retry). My question is, if we leave MRR enabled (by default) and Sample Rate chosen as the rate adaptation algorithm (by default), and lets say all transmissions at 54Mbps fails, then MRR retries at rate 1Mbps and is successful, does the transmission_time calculation for 54Mbps bit rate now include the time it took to retransmit using 1Mbps? I would really appreciate if someone could answer this question for me. 

Thank you very much,
Sincerely,

------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Madwifi-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/madwifi-users
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Madwifi-devel] Sample Rate Implementation

Derek Smithies

On Wed, 14 Jul 2010, Bishal Thapa wrote:

> Hello,  I have a small question regarding the implementation of Sample
> Rate in "madwifi version 0.9.4" concerning Sample Rate adaptation
> algorithm's coupling with MRR (Multi Rate Retry). My question is, if we
> leave MRR enabled (by default) and Sample Rate chosen as the rate
> adaptation algorithm (by default), and lets say all transmissions at
> 54Mbps fails, then MRR retries at rate 1Mbps and is successful, does the
> transmission_time calculation for 54Mbps bit rate now include the time
> it took to retransmit using 1Mbps? I would really appreciate if someone
> could answer this question for me. 

Hi, Some years ago I tried the different rate algorithms on offer then.
Sample, amrr and onoe.

They all performed badly. Take the following simple test.

Two nodes running madwifi+some_rate_algorithm.

put two nodes 30cm apart, measure bandwidth.

the measured datarate is good - real good

Insert barrier between minders - the panel of a computer is ideal.
measure throughput. The datarate is bad - as expected.

30 seconds later, remove that barrier.
With all the rate algorithms, the response time is incredibly slow. It may
take a minute to get back to where it was.

=====>
So I wrote Minstrel, based on an idea from Andrew McGregor.

The basis for the minstrel code was the sample code. I injested much
coffee, and had many discussions with Andrew, but could not work out the
reasons for the accounting in sample.

When Minstrel is put to the above test - measured throuhput is back to
perfect in under 5 seconds - every time.

let me be more definate.
Abandon Sample. The idea behind sample (to use ewma) is good.
The implementation in sample is poor.

The evidence on the net is conclusive. Use Minstrel.

Cheers,
  Derek.

--
Derek Smithies Ph.D.
IndraNet Technologies Ltd.
ph +64 3 365 6485
Web: http://www.indranet-technologies.com/

"How did you make it work??"
       "Oh, the usual, get everything right".
------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Madwifi-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/madwifi-users
Loading...