Re: Some misconceptions about the "peak_detector2" block

GNU Radio
#1

Ive also been looking for an appropriate fix for peak_detector2. When I
review this thread and the issue tracker, Im uncertain how the block is
supposed to behave. I think most of the developers have looked at the
documentation in the header file, and have tried to make fixes in
accordance with it. Specifically, that the peak search should only be
restricted to the range within the look_ahead parameter. If so, then
Achilleas is very close to an appropriate fix, needing only some
additional calls to set_output_multiple to prevent hanging. The block
would also work fine with a sine wave, as long as the look ahead value
was appropriate, like half the period.

As mentioned previously, the QA test may not necessarily be useful,
given that the input signal is much smaller than the window. Perhaps
the test file can also be modified to get better results. Im willing to
contribute to a fix and help make the peak detector block more stable,
but I would have to know more about how the block should behave. Any
comments or insights are appreciated.

Frank Fu

From: Tom R.
Subject: Re: [Discuss-gnuradio] Some misconceptions about the
“peak_detector2” block
Date: Mon, 13 Apr 2015 16:54:47 -0400
Achilleas,

I think you’ve completely failed to understand the issue from my
perspective. I do NOT disagree that there is a bug in the code. I also
do NOT disagree that most of what you’ve tried to do in the rewrite is
the correct way to rethink the block. What I have a problem with is that
you’ve provided me with a fix that breaks applications that used to run
fine, including the QA test.

As for the applications, there’s a really simple test you can perform.
Apply the peak detector to a sine wave. In the current code, it finds
the peak of the sine wave. With the new version, it does not just find
that as the peak, but it outputs as though it’s found the peak for every
length of the look-ahead value. This could be a valid design choice in a
peak detector where given a window, always emit the highest value in the
window. That is not what this block is supposed to do, nor is the new
block behaving consistently in this manner.

As for the QA test, the current tests presents a vector to the block and
the block finds the correct peak. With the new code, it doesn’t
complete. It just hangs. While it is true that stream-based blocks in
GNU Radio expect a continuous stream of data, any block that simply
fails to complete its processing when the rest of the flowgraph is done
is a bug. Instead, we have hooks like set_output_multiple and
overloading the forecast function that help us work with the scheduler
to make sure everyone gets the right amount of data they require. In
this case, you make a good argument that the block should look beyond
it’s current window to see if the max is in fact reached. If that’s the
case, then we need to have the block tell the scheduler this. If less
data is passed because there is no more data to process and the
flowgraph is shutting down, this block too must shut down. It will then
do so without providing the right answer. So the QA test will fail –
but it will complete and report failure in the data.

Please understand the above points when reworking your fix.

Thanks,

Tom

On Fri, Apr 10, 2015 at 4:22 PM, Achilleas A.
[email protected] wrote:
Hi all,

recently there has been some discussion regarding the peak_detector2
block, both in the github/gnuradio (pull request 404) as well as in the
issue tracker (issue 783).

It is now well accepted that this block is buggy: there are cases the
work function returns -1, which is a bug (see issue 783 on how to
recreate this bug).

I believe however that there is a DEEPER misconception about how this
block works/should work that has resulted in some frustration on what an
appropriate fix should be.
In particular there is an insistence that an appropriate bug fix should
pass the qa_test of this block and it should be [in the spirit] of the
existing algorithm.
In the following I will explain why passing the qa_test is a consequence
of the buggy behaviour of this block and NOT its feature.
In addition I will suggest what a proper behaviour of this block should
be, so that others who may want to write their own version of a peak
detector find it useful.

So the peak_detector block is very reasonable in its conception and its
name is very informative and appropriate. It works as follows:

Reads the input and keeps track of a running average (through a
single-pole iir filter)

When the current input crosses a threshold (= average * a user-defined
factor) upwards the block enters a search state, where it looks for the
maximum value of the input over a window of user-defined length.

This is clearly the intended behaviour of the block according to the
documentation (I don’t know who the original author is…).

#2

Thanks for your response, Tom. So basically, the documentation needs to
be updated to better reflect the desired behavior of the block. My
interpretation of the behavior would be something like The look_ahead
value creates a minimum search window. The peak is first searched
within this window. If the last sample in this window is the peak, the
block continues searching for the peak until the input flattens or falls
down.

So, I can see this block containing three states: peak not found, peak
search in window, and extended peak search. The current trouble, as I
see it, is during the window search phase. A lot of buggy things happen
when the window search isnt fully completed within a work call. If
set_output_multiple is not desired, how about creating a saved output
buffer of length look_ahead? During the window search phase, the output
state could be saved until enough inputs come in. The block would have
to use general_work instead, because there would be a temporary point
where input_samples are consumed, but no outputs are written.

Finally, I hope you would consider creating a third peak detector block
that would perform the finite window search. It would be extremely
useful, as it solves many kinds of signal detection problems for a
receiver.

Frank

#3

On Mon, Apr 20, 2015 at 12:47 PM, Frank Fu [email protected] wrote:

As mentioned previously, the QA test may not necessarily be useful, given
that the input signal is much smaller than the window. Perhaps the test
file can also be modified to get better results. I’m willing to contribute
to a fix and help make the peak detector block more stable, but I would
have to know more about how the block should behave. Any comments or
insights are appreciated.

Frank Fu

So I pretty much agree with this now. In the current form of the block,
the
function of the look_ahead doesn’t behave the way that you described
here,
Frank. However, looking at the (sparse) documentation for this block, I
can
now see why everyone thinks that’s how it’s supposed to behave. From
[1]:

look_ahead: The look-ahead value is used when the threshold is found
to
locate the peak within this range.

And if that’s how everyone expects it to behave, which is what I’m
hearing,
then we should correct that.

Here are a couple of points that I’d like to consider for reworking this
block. First, the flowgraph must finish no matter how many items are
passed
to it. The block was initially designed for an OFDM preamble detector,
so
we can’t have it just waiting around for more samples to come in, which
will add to latency.

I think that the variable “look_ahead” should probably be renamed to
“window” or something. The documentation for this block also needs a lot
of
work to avoid ambiguity or confusion about what it’s really supposed to
do
and how it works. Better QA code and an example would be really useful,
too.

Finally, I don’t think that set_output_multiple is the right answer
here.
I’d rather it save state between calls to work if it’s looking in the
current window. We shouldn’t have a block wait for thousands of samples
(the default look_ahead is 1000) before making a decision. And using
set_output_multiple with large values has, in my experience, a
performance
hit on a flowgraph.

Tom

[1]
http://gnuradio.org/doc/doxygen/classgr_1_1blocks_1_1peak__detector2__fb.html

#4

On Tue, Apr 21, 2015 at 2:23 PM, Frank Fu [email protected] wrote:

the window search isn’t fully completed within a work call. If
set_output_multiple is not desired, how about creating a saved output
buffer of length look_ahead? During the window search phase, the output
state could be saved until enough inputs come in. The block would have to
use general_work instead, because there would be a temporary point where
input_samples are consumed, but no outputs are written.

Frank,
See the discussion in http://gnuradio.org/redmine/issues/783

I’ve created a pull request to add Achilleas’ version to maint to
replace
the current one. There are still some potential issues that we might run
in
to, but this is still a much better answer, and my concerns might be
completely unfounded. The set_output_multiple issue is something that
I’ve
seen, but it’s anecdotal evidence. We’d need benchmarking before trying
to
develop a different solution.

Finally, I hope you would consider creating a third peak detector block

that would perform the finite window search. It would be extremely useful,
as it solves many kinds of signal detection problems for a receiver.

Frank

I would consider receiving a pull request for such a block :slight_smile:

But if anyone works on this, please don’t call it peak_detector3! Give
it a
more meaningful name for what it’s doing.

Thanks!
Tom