Packet Header Generator

GNU Radio
1

I’m trying to use the Packet Header Generator block
(not the Default one) from within GRC to generate a
packet header formatted to my own specification in
order to match an existing protocol.

I assume that an appropriate “formatter object” can
be submitted to this block to instruct it to build an
arbitrary header, but I’m having difficulty finding
any documentation or example of using a formatter object.
All I found points to the packet_header_default::header_formatter
method, which composes a fixed format that is of no use to me.

Any help point me in the right direction? Am I just missing
it or is there no “generic” header formatter capability?

@(^.^)@ Ed

2

On Fri, Nov 14, 2014 at 3:14 PM, Ed Criscuolo
[email protected]
wrote:

method, which composes a fixed format that is of no use to me.

Any help point me in the right direction? Am I just missing
it or is there no “generic” header formatter capability?

@(^.^)@ Ed

Hey Ed,

Funny enough, I was just working on this today. I’m adding a similar
packet
header generator for async (PDU-based) packets. The concept is similar
with
a default base class that you’d overload for your own purposes for
packet
formatting. I’ll be documenting this behavior more in the blocks
themselves, and this should eventually become the basis for another
level
of our tutorials.

I have it working here in simulation, but I know there’s going to be
some
trouble going over the air. I suspect I’ll have something that’s is
worth
sharing and reasonably documented early next week.

Tom

3

On 11/14/14 4:49 PM, Tom R. wrote:

arbitrary header, but I'm having difficulty finding

some trouble going over the air. I suspect I’ll have something that’s is
worth sharing and reasonably documented early next week.

Tom

OK. Meanwhile, is there some clever Python hack I can use in the
interim? I’m trying to get a timetag in a specific format
(64-bit integer, LSB=250 picoSeconds) into the header I’m generating.
It has to increment by 0x0000000000028000 for each packet.

For example, for my sequence number field, I needed a 16-bit unsigned
number that skipped over any value where the last 5 bits are ‘1’,
and rolled over at 2**16. I used a repeating short vector source
with the Python list comprehension

[s for s in range(0,32768)+range(-32768,0) if s%32 != 31]

as the vector entry in the GR block. This got merged with the
other header fields via stream mux block. It worked great.

But for the timetag, I don’t have the rollover which allowed
me to pregenerate a a fixed (large) vector. Any ideas?

@(^.^)@ Ed

4

On 11/14/2014 10:49 PM, Tom R. wrote:

Funny enough, I was just working on this today. I’m adding a similar
packet header generator for async (PDU-based) packets. The concept is
similar with a default base class that you’d overload for your own
purposes for packet formatting. I’ll be documenting this behavior more
in the blocks themselves, and this should eventually become the basis
for another level of our tutorials.

I have it working here in simulation, but I know there’s going to be
some trouble going over the air. I suspect I’ll have something that’s is
worth sharing and reasonably documented early next week.

Ed,

if you check out the OFDM code, is uses its own custom packet header
generator. Follow those examples to see how it’s done (or wait until Tom
has his stuff ready).

Cheers,
M

5

On Mon, Nov 17, 2014 at 3:47 PM, Martin B. [email protected]
wrote:

Ed,

if you check out the OFDM code, is uses its own custom packet header
generator. Follow those examples to see how it’s done (or wait until Tom
has his stuff ready).

Cheers,
M

Hey Ed,

You can check out my work-in-progress branch here:

https://github.com/trondeau/gnuradio/tree/packet_handling

It has what you need to add headers to PDUs in two different styles. The
default is our normal access code plus the frame length repeated twice.
The
second does the same header plus a 16 bit frame counter. The second is
partly to show how to manage the inheritance. Note that this is a WIP
and
some commits will likely get squashed and reworked to clean up the
history
before merging.

This concept is based off of Martin’s OFDM packet header
generators/formatters, but these are now classes designed around PDUs.
The
default async formatter works well inserted directly into our chat app
from
the Guided Tutorials after the CRC32.

As an aside, one project that I have for 3.8 is to clean up our digital
modulation support. We have a lot of blocks that have a lot of overlap
in
their capabilities, but are designed for very small differences in the
application. I’d like to try to reduce the overlap and remove some
duplication.

Tom