Hi Aldebaro,
It seems HDF is what you want. HDF is basically FSM + SDF combined, where
states refined to SDF models. In terms of efficiency, i think HDF is not
bad, since it is composed with SDF directors (where things are statically
scheduled). PSDF basically allows run-time reconfiguration of a SDF model
based on variable parameters. However, it seems like you want to specify the
reconfiguration using FSM's in your case, so PSDF is probably not expressive
enough (i may be wrong about this...). DDF is the most expressive and
generic out of the three, as it allows any dataflow actors w/o requiring
rate (production/consumption) signatures. It recalculates the firing
schedule for every iteration, so it's also the most inefficient out of the
three.
~mankit
----- Original Message -----
From: "Aldebaro Klautau" <[EMAIL PROTECTED]>
To: <ptolemy-hackers@bennett.EECS.Berkeley.EDU>
Sent: Saturday, May 17, 2008 2:51 PM
Subject: PSDF, DDF or HDF: which is faster?
We are evolving in the simulation of a modem using Ptolemy II and want to
make it rate-adaptive.
The idea is to use an estimate of the signal to noise ratio (SNR) and
change the modem constellation on the fly.
However, it seems that using only SDF is not possible and we need
"reconfiguration" to do that.
Our original modem transmitter was a simple LineCoder that would fire with
B input bits and output 1 complex number (one QAM symbol).
Speculating: maybe using a FSM we can drive the modem in a "retraining"
state, in which the constellations (in transmitter and receiver modems)
would be changed.
Reading ptIIdesign3-domains.pdf and looking at the demos
(BlindCommunication in PSDF, RandomWalk in DDF and AdaptiveCoding in HDF),
it seems that PSDF, DDF and HDF are all capable of implementing the
retraining functionality via reconfiguration.
Is that true? If yes, which one would be faster or even more appropriate?
Thanks in advance,
Aldebaro
--
==========================
Aldebaro Klautau
www.laps.ufpa.br
Phone:
+55 (91) 3201-7674
Address:
UFPA - PPGEE - LaPS
CP: 8619 - CEP: 66075.110
Belem - Para - Brasil
----------------------------------------------------------------------------
Posted to the ptolemy-hackers mailing list. Please send administrative
mail for this list to: [EMAIL PROTECTED]
----------------------------------------------------------------------------
Posted to the ptolemy-hackers mailing list. Please send administrative
mail for this list to: [EMAIL PROTECTED]
