Random number of output items

Hi,

I’m making a block which takes bit from a bit slicer and output packets
as
input comes in. My block will output bytes so it can emulate a usb
adapter
that receives the RF signal and output a packet stream through an FTDI.
That way I can use the stack that comes with the adapter without owning
one. I’ll use a FIFO file so other than not issuing the serail
configuration the stack should be used pretty much as is.

However, I’m not sure what I should do about the the number of outputs.
Let
say I’m waiting for the preamble, I won’t output anything. When I get
the
preamble and the sync I’ll send a sync byte of my own. From here every 8
inputs I’ll output a byte. So basically my block will output 0 or 1
output.

Can someone help me a little with the use of forecast and noutput_items
in
my case? Also do I need to use the consume_each function?

Thanks

On Tue, May 20, 2014 at 2:17 AM, Francois Gervais
<[email protected]

wrote:

Let say I’m waiting for the preamble, I won’t output anything. When I get
the preamble and the sync I’ll send a sync byte of my own. From here every
8 inputs I’ll output a byte. So basically my block will output 0 or 1
output.

Can someone help me a little with the use of forecast and noutput_items in
my case? Also do I need to use the consume_each function?

If your block emulates a USB adapter, defines it as a source block, then
you don’t need to touch forecast().
If your block takes input from another block, then it is not source
block.
I don’t really understand your requirements.

The number of outputs (referred as noutput_items) is determined by the
scheduler, not yourself.
Says, when you have X bytes to send out,
if X > noutput_items: Send out noutput_items number of output, and
return noutput_items
if X < noutput_items: Send out X number of output, and return X
if X == noutput_items: (either one of above)

When you send out a sync byte, add that to the output count.

When you are waiting for the preamble, you may want to send out a series
of
zeros, rather than just producing no output.
Producing no output may cause the downstream blocks to become
unresponsive.

Thanks Activecat you actually answered quite well to my question. I
thought
it might be better to send 0s, i’m glad you confirmed that. And thanks
for
the output algorithm.

Could you tell me more about forecast? Most of the time I need 8 input
samples to produce one byte output so I set the forecast like so:

ninput_items_required[0] = noutput_items*8;

It seems pretty straight forward. Is this correct?

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Francois,

as Activecat, I’m kind of having a hard time understanidng your
requirements.
If you’re emulating a hardware signal source, go for the source
approach.
If you’re basically taking input from anywhere and packing it into
packets of varying length, that’s exactly what the PDU block
infrastructure is for:
http://gnuradio.org/doc/doxygen/page_tagged_stream_blocks.html

Greetings,
Marcus

On 20.05.2014 15:59, Francois Gervais wrote:

It seems pretty straight forward. Is this correct?

I’m making a block which takes bit from a bit slicer and output
byte of my own. From here every 8 inputs I’ll output a byte. So
don’t really understand your requirements.
When you are waiting for the preamble, you may want to send out a

_______________________________________________ Discuss-gnuradio
mailing list [email protected]
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTe2FKAAoJEBQ6EdjyzlHtrvsH/3EhTVbESphbUfNeWmZe/rNU
YYOVD7XVYVndHzstznIwvYlNVRLugjJw74pPJ0DS050b1ggc9mDK/mW4975BICau
hQQktHxN3QReWk5qKwpAx6Y/+3bVpC+phyFzZO0+1TBwHNYexsVA+Zw0mkGrOuvA
3pInPREkJxqxcMrbZZhMTYVVDfOpB5MEjxHSKOMWyAQqop2fg1ahlEwpKjVqDmmV
9NhjSbIy29kpayTcEq525ha0QPMb5bRkRiP1sw4GqHDJZSHyUR4RWYYHiVfD3CvP
/Bmx74UyS69gnP6NmMxun3OjWngpNHkJNC0lN/GGHVhz7YjwVXINuXNkbYuEVaM=
=x2qL
-----END PGP SIGNATURE-----

Hi,

PDU blocks are a type of blocks. Basically, you tag your sample stream
so
that the blocks downstream know how long your packet is.
The tagged stream infrastructure is an innovation meant to simplify the
design of blocks dealing with packetized data.
Lool in the gr-digital/examples subfolder for how some implementations
of
that principle are used.

Greetings,
Marcus

On Tue, May 20, 2014 at 4:23 PM, Francois Gervais
<[email protected]

Thanks I’ll take a closer look.

Sorry about that I’ll try to clarify thing.

I’m using an rtl-sdr adapter to receive an RF signal. I demodulate it
and
send it through the MM clock recovery and bit slicer. Then the binary
signal enters the block I’m talking about here.

This block find a valid packet by matching the preamble and the sync
pattern and translates the packet into another format that is understood
by
a software stack designed for these type of packets.

Normally this stack would take it’s input from a serial port but in my
case
I output the packets in the correct serial protocol through a file sink
that is binded to a FIFO.

I’ll take a look at the PDU block but this translation needs to be done
between two very specific protocol. I don’t think a generic block can
achieve that.

Right now the development is mostly finished. All I need to make sure is
that my block interface correctly with the scheduler since I don’t have
a
fixed input to output number relationship. It’s mostly 8:1 but that not
always exactly the case.

Hope it makes sense.

Hi Marcus,

I’m not sure about the steps required to translate the bit stream from
the
RF receiver into a tagged stream.

I looked at the ofdm_rx example and from what I understand I’ll need a
first block that takes the output of the demod/Clock Recovery/bit slicer
and find the packets inside the stream and tag it so the other blocks
can
work with the tagged stream. This first block is a normal one that
inherit
from the generic Block class it that it? Does this first block need to
output anything while waiting for the preamble of a packet to prevent
the
downstream block from freezing?

Once the stream is packetized and tagged it seems pretty straight
forward.

Thanks

On Tue, May 20, 2014 at 10:37 AM, Francois Gervais <

On Tue, May 20, 2014 at 10:23 PM, Francois Gervais <
[email protected]> wrote:

A picture is worth a thousand words. If you attached your existing
flowgraph then it helps a lot.

I’ll take a look at the PDU block but this translation needs to be done

between two very specific protocol. I don’t think a generic block can
achieve that.

Please be reminded that PDU and tagged stream are two different things.
There are “PUD to Tagged Stream” and also “Tagged Stream to PDU” blocks,
to
perform conversions.
“Tagged Stream Block” is a special block that can only take input of
tagged
stream; the tag of the first element must consists the packet-length
information, here the default tag key is “packet_len”

PDU is used for Message Passing, I doubt you really need this in your
flowgraph.

If you are sure that your custom block needs to produce output in
packets
(not solely bytes), then you may want to make the block output as tagged
stream. If applicable, in alternative, you may consider to use the
“Stream
to Tagged Stream” block. In this case, the downstream blocks may be
inherited from Tagged Stream Block.

In cases other than that, you don’t need tagged stream block at all.

This forum is not affiliated to the Ruby language, Ruby on Rails framework, nor any Ruby applications discussed here.

| Privacy Policy | Terms of Service | Remote Ruby Jobs