Re: [Discuss-gnuradio] Better approach for FEC

[Tom Rondeau]

On Wed, Aug 26, 2015 at 2:14 PM, Marcus D. Leech <address@hidden> wrote:

I think many of the folks who would have comments on this are at the Gnu Radio conference this week....

On 08/26/2015 02:09 PM, bob wole wrote:

On Tue, Aug 25, 2015 at 11:36 PM, bob wole <address@hidden> wrote:

Hi,

I have a burst system in gnuradio which is working fine with usrps. I have used tags (tx_time, tx_eob and tx_sob) to define packet boundaries and transmission time. Tags are inserted by a block which is connected before the usrp sink. I did not use any "tagged stream block" in the burst system. All the blocks used are streaming blocks.

Now I want to insert FEC in my system. After going through the FEC API, I realized that I can use any of FEC deployments (Streaming, Tagged stream, Asynchronus). Are there any difference(s) performance wise? Which one should is better for my system? 

I have attached a picture of my current system and identified where I want to insert FEC in tx and rx.

Comments are appreciated.

--

Bob

Comments please ?

--

Bob

Bob,

The various models of handling packets through a GNU Radio system are dependent on a few things, none of which are easy black/white issues to describe or offer support on. Partly, it can just be your opinion on where the lines are between the different models; or based on what blocks are available in the different models. You don't want to go back and forth, though, so when you move from a stream into a PDU-based system, you'll want to stay there.

Performance is going to be dependent on the rest of your system as well. We can't tell you one is better than the other, but we've put in a lot of hooks into the system to help you understand those issues for yourself, like using the performance counters and naming each thread with the block name so top/htop can help.

Tom