osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Tom Tsou]

Patch Set 1: Code-Review-1

We can't reduce the detection window unless the timing concerns in e90c24c8 are 
considered and/or the underlying issue is addressed. One possibility is 
disabling 4/1 SPS support and/or defaulting to 4/4 SPS (which has more stable 
timing), but those are much more drastic solutions.

-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Harald Welte 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: No

osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Tom Tsou]

Patch Set 1:

> Btw, if I understand correctly what you're talking about - it's not
 > "jitter", i.e. it doesn't change in time, there is an error which
 > is static after PLL has been locked.
 
Correct. It's timing 'jitter' in the sense that it changes run-to-run, but the 
timing offset static once running.

 > I think we should deal with it by setting max_toa accordingly
 > instead of changing these fundamental values.

The 'tail' side can be addressed with max_toa, but the problem that came up in 
e90c24c8 was not delay but excessive advance, the 'head' portion, so max_toa 
does not apply.

-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Harald Welte 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: No

osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Alexander Chemeris]

Patch Set 1:

> > Tom, could you explain why did you set search window to -/+6
 > > symbols? My experiments showed that at least `tail` should be set
 > > to 3 in order for `max_toa` to actually represent maximum TOA.
 > And
 > > same for `head` - from my experiments I think it should be 3.
 > 
 > Previous pre-search or 'head' settings on normal have come from
 > experimentation.
 > 
 > In number of symbols:
 > 
 > 5 - Original OpenBTS / Range Networks
 > 4 - Reduced during filter & SSE rewrite
 > 6 - Expanded in e90c24c8 to account for timestamp jitter in certain
 > configurations
 > 
 > With 3 symbols, have you checked TSC timing stability in the 4/1
 > sps mode where jitter is/was most severe?

That's an interesting point. I so far has tested in a receive only application 
where there is no Rx/Tx difference issues - we always do blind detection.

Btw, if I understand correctly what you're talking about - it's not "jitter", 
i.e. it doesn't change in time, there is an error which is static after PLL has 
been locked.

I think we should deal with it by setting max_toa accordingly instead of 
changing these fundamental values.

-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Harald Welte 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: No

osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Alexander Chemeris]

Patch Set 1:

(2 comments)

https://gerrit.osmocom.org/#/c/2765/1//COMMIT_MSG
Commit Message:

PS1, Line 7: W
> I this is still WIP, I am not sure it should be submitted to gerit? If it's
I thought it's better to discuss it here with the code/patch in front of us. I 
can probably move the discussion to the mailing list if you think this is not 
the right place.


Line 9: Otherwise we detect bursts with search window far beyond specified.
> specified by whom / where? Would be good to have a spec reference here.
Specified by the max_toa variable.


-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Harald Welte 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: Yes

osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Tom Tsou]

Patch Set 1:

> Tom, could you explain why did you set search window to -/+6
 > symbols? My experiments showed that at least `tail` should be set
 > to 3 in order for `max_toa` to actually represent maximum TOA. And
 > same for `head` - from my experiments I think it should be 3.

Previous pre-search or 'head' settings on normal have come from experimentation.

In number of symbols:

5 - Original OpenBTS / Range Networks
4 - Reduced during filter & SSE rewrite
6 - Expanded in e90c24c8 to account for timestamp jitter in certain 
configurations

With 3 symbols, have you checked TSC timing stability in the 4/1 sps mode where 
jitter is/was most severe?

-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Harald Welte 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: No

osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Harald Welte]

Patch Set 1:

(2 comments)

https://gerrit.osmocom.org/#/c/2765/1//COMMIT_MSG
Commit Message:

PS1, Line 7: W
I this is still WIP, I am not sure it should be submitted to gerit? If it's not 
WIP, it should not say so.


Line 9: Otherwise we detect bursts with search window far beyond specified.
specified by whom / where? Would be good to have a spec reference here.


-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Harald Welte 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: Yes

osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Alexander Chemeris]

Patch Set 1:

Tom, could you explain why did you set search window to -/+6 symbols? My 
experiments showed that at least `tail` should be set to 3 in order for 
`max_toa` to actually represent maximum TOA. And same for `head` - from my 
experiments I think it should be 3.

-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris 
Gerrit-Reviewer: Alexander Chemeris 
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Tom Tsou 
Gerrit-HasComments: No

[PATCH] osmo-trx[master]: WIP:sigProcLib: Reduce burst detection window for NB.

[Alexander Chemeris]

Review at  https://gerrit.osmocom.org/2765

WIP:sigProcLib: Reduce burst detection window for NB.

Otherwise we detect bursts with search window far beyond specified.

Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
---
M Transceiver52M/sigProcLib.cpp
1 file changed, 2 insertions(+), 2 deletions(-)


  git pull ssh://gerrit.osmocom.org:29418/osmo-trx refs/changes/65/2765/1

diff --git a/Transceiver52M/sigProcLib.cpp b/Transceiver52M/sigProcLib.cpp
index c51d094..e47c985 100644
--- a/Transceiver52M/sigProcLib.cpp
+++ b/Transceiver52M/sigProcLib.cpp
@@ -1849,8 +1849,8 @@
 return -SIGERR_UNSUPPORTED;
 
   target = 3 + 58 + 16 + 5;
-  head = 6;
-  tail = 6 + max_toa;
+  head = 3;
+  tail = 3 + max_toa;
   sync = gMidambles[tsc];
 
   rc = detectGeneralBurst(burst, threshold, sps, amplitude, toa,

-- 
To view, visit https://gerrit.osmocom.org/2765
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: newchange
Gerrit-Change-Id: If3cb40d2311504a13c03e1fbccad663ac201d9a4
Gerrit-PatchSet: 1
Gerrit-Project: osmo-trx
Gerrit-Branch: master
Gerrit-Owner: Alexander Chemeris