[drats_users] Simultainious voice and data

[Dave Cooley]

Okay understood. I think what you are saying is that it is better to
increase the number of pipelined blocks than the size of the block? This
would accomplish as much of the primary purpose without increasing the
re-transmit time for bad blocks by increasing the block size.

Ultimately there are trade offs in either scenario.
If you want a transparent data transfer (during voice comms) then use
broadcast. The trade off is you don't know if the file reached its intended
target. The problem with this scenario is that forms are not broadcast.  

The only real way to send data with some assurance of quality of delivery is
to use forms or file transfer and adjust the number of blocks pipelined to
fit the file size. This would keep retransmit times at a minimum.  

Any suggestions for the block size???

73

Dave Cooley
N4DIB

-----Original Message-----
From: drats_users-bounces at lists.danplanet.com
[mailto:drats_users-bounces at lists.danplanet.com] On Behalf Of Dan Smith
Sent: Friday, December 12, 2008 8:27 AM
To: Discussion of D-RATS
Subject: Re: [drats_users] Simultainious voice and data


> Yes that is what happens. When the block size is too short DRATS sends 
> the first part of the transfer and then waits for block status from 
> the distant end. If the transmission time is enough to accomodate the 
> entire file then all is well. Confirmation is received from the 
> distant end and a file transfer complete is sent.  If not, the 
> handshake occurs and any bad blocks are re-transmitted along with the 
> next pipeline. This will continue until all of the data is transmitted 
> and confirmed.

Well, this isn't quite accurate.  Even if the entire file fits in the first
round, there is still negotiation before and after the actual file
transmission that needs to happen.  A couple of back-and-forths to setup the
session, exchange the file information at the beginning.  After the file,
another couple to confirm receipt of the file (you don't want the sender to
think it was successful when it wasn't) and then to close the session.

> 	Net control wants to send a radiogram to a DRATS station. The form
is 
> filled out and ready to be sent. A DRATS equipped station states they 
> are ready to take the traffic. Net control acknowledges and during the 
> voice conversation, sends the radiogram form. If the block size and 
> pipeline blocks are set to the size of the entire form,  a single 
> transmission will pass the radiogram to the receiving station in a 
> single transmission. The receiving DRATS station will check the blocks 
> for errors and request a resend of bad blocks. If no bad blocks were 
> revceived the transmission is complete.

Not quite, as stated above.

> It is understood that the larger the block size the greater chance for 
> errors and this must also be taken into consideration.

That's why I pipeline.  You can keep the block size small, but increase the
pipeline length.  This means that a single bit error will affect a smaller
amount of data, thus requiring less to be re-sent.  The smaller the block
size, the more the overhead, of course, but on a lossy channel it will pay
off.

-- 
Dan Smith
dsmith#danplanet.com, s/#/@/
www.danplanet.com
KK7DS
_______________________________________________
drats_users mailing list
drats_users at lists.danplanet.com
http://lists.danplanet.com/mailman/listinfo/drats_users

Internal Virus Database is out of date.
Checked by AVG - http://www.avg.com 
Version: 8.0.173 / Virus Database: 270.8.0/1717 - Release Date: 10/9/2008
4:56 PM