First step improving and reworking TA loop:
Move trigger of the loop to similar place done by BS/MS Power Control
Loop, that is, upon receivial of UL SACCH block, which contains
information about the TA used to transmit the block encode in L1SACCH
Header. Hence, from computed received TOA and TA used when transmitting
from the MS, we can infer the desired TA to be used by the MS, which
will send back to it later during DL SACCH block.
The values taken are actually the ones calculated for the previous SACCH
block and stored in lchan->meas.ms_toa256. That's because L1SACCH
contains the TA used for the previous reporting period, similarly to
what's specified for MS Power Control loop (TS 45.008 sec 4.2):
"""
The MS shall confirm the power control level that it is currently employing
in the SACCH L1 header on each uplink channel. The indicated value shall
be the power control level actually used by the mobile for the last burst
of the previous SACCH period.
"""
(The reader may observe that currently this is not properly done for MS
Power Control loop when calling lchan_ms_pwr_ctrl(): this is a bug.)
This new method also permits changing TA quicker, since we have more
confidence that the TA we request is aligned with the one used to
transmit, and we don't simply increment/decrement based on the value we
request to transmit.
Related: SYS#5371
Change-Id: I2d0f128c8dcac93ee382283a1c91fca76623b8fc
The timeslot number ts->nr is set in gsm_bts_trx_alloc() and can
never be greater than 7. Regarding the physical channel type
returned by ts_pchan(), it's also unlikely to be greater than
_GSM_PCHAN_MAX.
Change-Id: I6d20d7cba49cc8b6d1dc2192598ca372b7d2c5bf
The only reason why it was not 'const' is that in lchan_new_ul_meas()
we may need to overwrite 'ulm->is_sub'. This can still be done after
memcpy()ing a new set of samples to the destination buffer.
Change-Id: I0cabf75f8e0bf793c01225a4a8433e994c93f562
Related: OS#5024
GCC-4 doesn't like static const struct foo bar = (struct foo)...:
measurement.c:33:1: error: initializer element is not constant
Related: OS#5004
Change-Id: Iec5a4006dfc90abe240fcf8b6c2fefd4f7071a70
The variables num_meas_sub_expect - num_meas_sub must not be subtracted
without prior checking. Depending on the input (which might be
errornous), num_meas_sub might be greater then num_meas_sub_expect. This
eventually leads into odd behavior, which can be difficult to debug.
Change-Id: I381cc637d1c125f279ccf88db114609946fe24fe
Related: OS#4799
The logic in measurement.c checks the amount of collected measurement
values. This is done for the total amount of measurements and the amount
of SUB blocks measurements.
The functions that return the expected number of measurement values
currently do not take into account that the mode of a TCH/F or TCH/H has
an effect on the number of expected SUB blocks. (In signalling channels
all blocks count as SUB). Also a TCH/H in signalling mode generates only
half the amount of measurements because the blocks in signalling mode
are sepreded over 6 bursts instead of 4. This also needs to be taken
into account.
Change-Id: I01c7b6cc908c647263ab88f6b6281c4732f88779
Related: OS#4799
SUB frames exist only in voice (or CSD) channels. When a TCH/F is in
signalling mode, all blocks must be counted as SUB blocks. (for TCH/H
the current implementation is correct.)
Change-Id: I04be21200afa1d03afa0d7e476c66fa79cf42249
Related: OS#4799
The amount of SUB frames that may occur in AMR is not fixed, nor can it
be determined by some formular or lookup table. The reason for this is
that the DTX period may negotiated dynamicly. This means that the lower
layers must make the decision if an AMR sub frame is a SUB frame early
and tag the repective measurement / frame they hand up to the upper
layers accordingly. However, regardles of the occurrence of DTX periods
the amount of SUB frames in AMR must be always 1 or more because SACCH
frames always count as SUB frames as well. Lets make sure that this is
respected in the debug log as well as in the logic that tries to
substitue missing SUB frame measuremnets.
Change-Id: I1fd91b576ff7274caa6d4356bcd7a4fa4311219d
Related: OS#4465
A debug line is printed whenever a measurement is added and it also
displays if the added measurement is a sub frame or not, lets display
all the other properties too, especially inv_rssi would help a lot as it
shows us the rssi values which are reported by the lower layers
Related: OS#2978
Change-Id: I0299b75e99661e8dd22ad6604a897db4533c3b2b
The function ts45008_83_is_sub rougly decides if a frame is a SUB frame
or not. This works by checking the frame number against against lookup
tables. This works fine for codecs where the occurrence of SUB frames is
fixed. However for AMR this is not the case as the DTX periods are
dynamic. Here it is the responsibility of the lower layers (phy,
frame decoding) to tag SUB frames early since making the decision later
based on the frame number is not possible.
The parameter is_amr_sid_update was probably added as a placeholder. It
is set to falls by the callers of the function. Lets remove this
parameter as a late decision if an AMR frame is a SUB frame will never
work.
Change-Id: I125d5ff592218a9e98130a6a7b6bbc6378ce4132
Related: OS#2978
The variable irssi_full_sum is not populated with a dummy value when we
are not able to compute irssi_full_sum. Instead we mistakenly write
MEASUREMENT_DUMMY_IRSSI to ber_full_sum, which is wrong
Change-Id: I44d7cb48e3c68ab1b48c78cceb9381ce3e39d7e8
Related: OS#2987
The timing advance controller that is implemented in loops.c of
osmo-bts-trx only works for osmo-bts-trx and not for any of the phy
based bts. Lets move the timing advance controller into the common part
and make it available for every bts. Also lets add a unit-test.
Change-Id: If7ddf74db3abc9b9872abe620a0aeebe3327e70a
Related: SYS#4567
The variable ta256b_sum is int32_t and num_ul_meas_actual is unsigned
int. When ta256b_sum is negative the division produces the wrong result.
This is beacuse the division is performed unsigned as the usual
arithmetic conversions promote to unsigned where both both operands are
the same width.
Lets fix this by casting num_ul_meas_actual to signed.
(Note that in the same function there are various other averages
computed in the same pattern, but they have unsigned operands and so are
correct.)
Related: SYS#4728
Change-Id: I37e3f69109c5ca2948bd4cdb7aa017bf2fcb8172
The unit test that tests lchan_meas_process_measurement() only inputs
test data to lchan_meas_process_measurement() but it is not checked if
the interval end could be detected or not.
- Add a return code to lchan_meas_process_measurement()
- Ensure that the return code is checked in the unit-test
Change-Id: I9e00ce683e8c44528804f65181dbfed9e85e3aed
Related: OS#2975
The function is_meas_overdue() was introduced to allow
lchan_meas_process_measurement() to detect when the end of a measurement
interval has been missed. Interval ends may be missed when the SACCH
block of the related measurement interval gets lost. This is due to the
fact that the SACCH block is used as a trigger to start the measurement
result computation.
The idea behind is_meas_overdue() was to check the frame number of the
current measurement against the frame number of the previous measurement
in order to see if there was a measurement for SACCH in between or not.
Unfortunately SACCH and TCH Voice data is not necessarly processed in
order by each phy. Depending on the phy there may be a jitter between
the timing of SACCH and TCH Voice. Depending on the phy this jitter may
be enough to mess up the timing so that we see a SACCH block earlier
than expected. So we can not use the current frame number of TCH Voice
measurements to check for missed SACCH blocks.
Change-Id: Idfdbf64c1f965f35c12559b3995e2b746c74ee9e
Related: OS#3502
Related: OS#2975
When adding a new measurement also log the frame number by the modulus
of the measurement interval to simplify debuggung
Change-Id: I77a4d947dab32de0d5717ebf13bd8de6179dfe6a
At the moment the measurement calculation of osmo-bts works by
collecting the measurement reports the phy emits during a measurement
interval. Normally one would expect a well defind fixed number here, but
some phys will not emit a measurement report for lost blocks. Also
blocks and their reports may get lost because of cpu overload etc.
The computation that is executed at the end of the measurement interval
computes over all received measurement. This evenutally means that
missing measurements will not taken into account and the result will
look better than it is in reality.
To fix this, the interval must be of a defined size and in cases where
less measurements as expected were collected, the algorithm must assume
they have been received with a 100%BER and take that into account.
However, all RSSI and TA/TOA related computations should continue to
rely on actual measurement data.
- make sure the algorithm works over a fixed interval
- replace missing measurements with 100%BER
- fix and extend unit-tests
Change-Id: Idd30fc07603ad7d042c1fb416e247c3bf7d35c8b
Related: OS#2987
The lookup table that control the measurement interval endings do not
make clear what their indexes refer to. Lets give them more distinct
names.
rename sdcch8_meas_rep_fn102 to sdcch8_meas_rep_fn102_by_ss
rename sdcch4_meas_rep_fn102 to sdcch4_meas_rep_fn102_by_ss
rename tchf_meas_rep_fn104 to tchf_meas_rep_fn104_by_ts
rename tchh0_meas_rep_fn104 to tchh0_meas_rep_fn104_by_ts
rename tchh1_meas_rep_fn104 to tchh1_meas_rep_fn104_by_ts
Change-Id: I3dc891e1860109f803c1bfa46445e8fef35586d9
Related: OS#2975
The tests TC_meas_res_sign_sdcch4 and TC_meas_res_sign_sdcch8 are
failing mainly because lchan->ts->nr is confused with lchan->nr.
There is also a small problem with one of the formulas that compute
fn_missed_end.
- Add explainatory comment to the lookup tables on what the index
is refering to
- use lchan-nr instead of lchan->ts->nr when dealing with SDCCH/4/8
- simplfy and fix the formula
- increase the testcoverage of the unit tests, give SDCCH/4/8 also
a thorough check.
Change-Id: I5d555a21003943bf720c53f3a611029ba45339a9
Related: OS#2975
The function ts45008_83_is_sub() is an integral part of the measurement
calculation as it automatically tags incoming measurements as SUB
measurements. This is important in the context of DTX. Unfortunately
there is no unit test for this function yet.
- Add unit test for ts45008_83_is_sub()
Change-Id: Ia26774859f4bf31baee075896905079368bddd42
Related: OS#3502
The Table that lists the fn%104 frame number masks that define which of
the incoming measurements to be recognized as SUB measurements contains
one wrong number.
For comparison see also:
3GPP TS 05.08, chapter 8.3 Aspects of discontinuous transmission (DTX)
- Change 29 to 20 in ts45008_83_tch_hs1[]
Change-Id: Id3698551d94866f418e662b9fe81429b16e2621c
Related: OS#3502
We do not test is_meas_complete() individually yet, but it is an
integral part of the measurement processings since this function decides
where a measurement interval ends.
- Add unit tests that test TCH/F, TCH/H, SDCCH/4 and STDCH/8
Change-Id: I8f89d9e7092cd65ba4d5c5649140692dcc20bdd6
Related: OS#2987
for SDDCH4 channels, the detection is not working correctly since the
function uses the lookup table for SDCCH8 interval endings there. This
needs to be corrected. Also there are two unnecessary assignments in
the code which should be removed.
- use correct table (sdcch4_meas_rep_fn102 instead of
sdcch8_meas_rep_fn102.
- remove unnecessary assignments to last_fn_mod
Change-Id: If8a269ecd3f9fa4eeadf379114db816ef5c77d77
Related: OS#2975
the measurement interval end is detected by using the measurement
indication that is related to the SACCH block as a trigger to start the
computation. If the measurement indication for the SACCH gets lost
because the block could not be received then the processing is not
executed. This may cause wrong results or when it happens condecutively
an overflow of the measurement sample buffer.
- Store the frame number of the last received measurement indication
- Use the stored frame number to check if an interval was crossed when
the next measurement indication is received. If we detect that we
missed the interval, catch up by running the computation and
start the next interval.
Change-Id: I3a86cd8185cc6b94258373fe929f0c2f1cf27cfa
Related: OS#2975
At the moment only lchan_meas_reset is reset on channel activation.
All other states are not reset. This may lead to irretations in the
first measurement interval if there are still leftover messages from
a previous connection. Lets ensure everything is reset to zero by
zeroing out the whole .meas struct in struct lchan.
- Add a centralized function that does the reset
- Call that function from rsl_tx_chan_act_ack() in rsl.c
Change-Id: I880ae3030df6dcd60c32b7144c3430528429bdea
Related: OS#2975
Related: OS#2987
In l1sap.c we call lchan_new_ul_meas() and lchan_meas_check_compute()
directly in sequence. Lets unify thos two steps inside measurement.c so
that we only need to call one function from l1sap.c.
Change-Id: If48bc7442dfaab8c36b93949f741de6e836e792a
Related: OS#2975
This patch adds extended processing of the high-resolution TOA256
measurement values. It adds reporting of the following values
for each RSL MEAS REP for uplink measurements:
* minimum TOA256 value during reporting period
* maximum TOA256 value during reporting period
* standard deviation of TOA256 value during reporting period
Change-Id: Iea4a4781481f77c6163d82dcd71a844a5be87bf2
At the end of a measurement processing window, we currently compute
the ToA / timing offset at 1/256th symbol accuracy, but we only print
it to the log. Let's store the value in the lchan to make it usable
by other code in follow-up patches.
Change-Id: I5f00a16ac966b627d9452a98b8fa70984bed684a
The existing code contained an ugly hack that if we didn't have any
"SUB" measurements we would simply use the "FULL" values. That's wrong
as TS 45.008 contains quite detailed rules on how the "SUB" values are
to be computed. In some cases, they are identical to "FULL", but in
most they are not.
Let's remove the hack and replace it with an ERROR message, as clearly
something is wrong if we ever encounter a measurement period end in
which no single "SUB" measurement was received. The only situation in
which this can occur is if the related uplink burst/block was missing,
so let's set BER to 100% and level to lowest possible.
Change-Id: I358f7b97fd4ea19264a77eff7abef13da7d5fbcd
The rules on how to compute RX{LEV,QUAL}-SUB are rather convoluted, and
depend on the detailed channel type and mode.
For SDCCH and TCH/H in signalling mode, it's easy: No DTX is allowed,
and all measurements are used in SUB.
For non-AMR (TCH/F and TCH/H in non-signalling mode), we need to count
the SACCH block measurements, as well as any
SID/SID_UPDATE/L3_FILL/DUMMY blocks received in the blocks of table
8.3 of TS 45.008.
Only AMR (TCH/AFS + TCH/AHS) are more difficult, as there are no fixed
blocks/bursts/frames that always contain uplink messages, but the L1
will have to determine if a valid SID_UPDATE was received or not.
This patch implements the above rules (except AMR related) in the common
part of OsmoBTS. The AMR specific bits will have to follow as a later
patch, likely in a BTS specific way, i.e. separate changes to
sysmo/lc15/octphy/trx code.
Related: OS#2978
Change-Id: I16eb3747a1c23df935a4c50dafe46abce512a474
Since Change-Id: I23fba50f48415314da40cf5bf86fce2ed3e66af6 we were not
reporting measurements for SDCCH channel types due to the wrong
encoding of the sdcch{4,8}_meas_rep_fn102 table.
Let's fix the table by encoding the needed information:
"What is the modulo-102 remainder of the first burst of the last block
before fn%102 reaches 37?" (SDCCH/4)
"What is the modulo-102 remainder of the first burst of the last block
before fn%102 reaches 12?" (SDCCH/8)
The TS 45.002 Clause 7 tables have to be consulted carefully to
determine this information.
Change-Id: Icf02354872670126ab3297b787b216981ca6c309
Related: OS#2965
This is currently only used for logging, but will be needed for proper
RX{LEV,QUAL}-SUB reporting in upcoming patches.
Related: OS#2978
Change-Id: I07fd06e8a379cab7c0c2eb111c3f5600037d3c9e
This reverts commit d5fdcfe6d9, which
introduces a new function lchan_meas_update_ordered_TA whose
functionality I still haven't yet managed to fully understand. It
appears to be adjusting the requested timing advance (lchan->rqd_ta) but
outside osmo-bts-trx/loops.c code. This is odd, as rqd_ta is a state
variable of that loops.c code.
So for one part, it is a failure of encapsulation. The TA loop code
should be self-contained, particularly as it is only used for
omso-bts-trx, and not for the other BTS models. The new
lchan_meas_update_ordered_TA() function is used from common code,
applicable to all BTS models.
The resulting interaction between loops.c code and this new (now
reverted) function cause the TA value to only ever grow, despite the MS
never moving at all.
Change-Id: I5a5adac6f18f94a5b51758a5ace8ef6ddfd23e80
Related: OS#2989
There's no point in either having full verbosity in DEBUG level
and not logging any measurement related information in INFO. Let's
at least print the results at the end of each measurement period in INFO
level.
Change-Id: I2c870531478c05ce31cc1015597a068a4a76cf99
The end of the measurement reporting period is not aligned with the
SACCH block where the results are reported. The tables that are
used to detect the end of the measurement period are therefore
wrong. The frame number of the SACCH block must be used and not the
TDMA frame number (modulo 104) of the measurement reporing interval.
The tables are oriented to the frame number of the first SACCH block,
at the beginning of an interval. However, when a SACCH block is received
it will always contain the result of the recently passed measurement
reporting period.
To match the tables, introduce another lookup table to remap each
SACCH block that ends to the matching beginning block number.
Change-Id: I1eef894e6f15b4449fc8926bebb118624efc7924
The code that receives the uplink measurement data from L1 does not
print the number of already received uplink measurements. Since
this is a valuable information when debugging the log output will
now print this information as well.
(Patch by Octasic Inc.)
Change-Id: I79926f25de088571fcc2c14388c72fc968c2d382
We used to have trx_meas_check_compute() and call that from the
bts-specific code in order to iterate over all timeslots and all lchans
in the timeslots if we have to send measurement reports. This was
executed once per frame, and created unequal CPU load over time, which
in turn might increase different per-ts jitter.
Since 2f028c4e2c in April 2017 we have
lchan_meas_check_compute(), which performs this on a per-lchan basis,
and hence CPU load is distributed over all active timeslots.
Change-Id: I6308cefe4a51e55719ea4ed4d613d3782b805c08
The debug log does not print much information about the measured
rxlev and rxqual values. This commit adds debug output to make
measurement debugging simpler
Change-Id: Ic871eed6dcbc7d10aca6cd11dbc803b3e6da449f
The offsets used in the lookup tables may be phy dependand. So we
will have to do the alignment in the layer 1 interface of the
phy dependand code before we report the frame number up to the
higher layers.
This reverts commit e0fb3ae52d.
Change-Id: I8acab50cc1bb1cb133831c6a145f4d790e99176c
Pau Espin
Vadim Yanitskiy
Oliver Smith
Philipp Maier
Alexander Couzens
Martin Hauke
Harald Welte