headers
BABONNEAU Gérard FTRD/DMI/REN | 25 Jul 2003 13:45

PGMCC window manager computation

Hi,
 
Algorithm in section 2.2 seems ambiguous.
For "ACK RECEPTION, LOSS DETECTED", order is important: if I suppose there are only one T variable and one W variable:
 
  • If I compute first T = T - W/2, I use the previous value of W
  • If I compute T after W, I use the new value of W
 
The same problem exist also in the token computation without losses.
 
Is there something missing in the draft, or do I misunderstand?
 
Regards,

Gérard BABONNEAU

france telecom R&D
4, rue du Clos Courtel - BP 59 - 35512 Cesson Sevigné Cedex
Tel : +33 2 99 12 40 08  -  Fax : +33 2 99 12 40 98
e-mail : gerard.babonneau <at> francetelecom.com



 
Permalink | Reply |

Gmane