Currently we have 2 out parameters, but in forthcoming commits will add
a third one. All those functions already have too many parameters, so
let's put together all the output params in a struct to pass them easily
and make it easier to understand they are the estimated output values.
Related: OS#4006
Change-Id: I05cfa0ceaa2e633a5e6e404e2eae497ff4442dea
Make the interface using trx_ul_burst_ind more implementation agnostic
as well as easier to use. For instance, we don't care about SoftVector
size one returned from pullRadioVector(); we want to use nbits instead.
As a result, we no longer spend time normalizing guard periods. While at
it, change vectorSLicer to return void since it always returns true.
Change-Id: I726e5a98a43367a22c9a4ca5cbd9eb87e6765c7a
- Those are not used any where
- Those are not supported by the sse/neon accelerated versions
- And I see very little use cases for those.
Change-Id: Ic850269a0ed5d98c0ea68980afd31016ed555b48
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Original issue: In order to use SSE instructions, 16-byte aligned memory
chunks are needed, and C++ version < C++11 doesn't provide for a native
new/delete store. For that reason, memalign() must be used in the
implementation of convolve_h_alloc() for some buffers.
On the other side, The C++ code relies on C++ "new T[]" operator to
allocate a chunk of memory containing an array of class instances. As
classes are complex types, they cannot be allocated through C structures
(calling malloc). Experimentally can be seen too that it's unreliable
and the process will crash during startup if malloc() is used and then a
Complex<> deferred from it.
Previous implementation allowed for use of convolve_h_alloc or new[]
based on how the (signal)Vector is called, because then the buffer is
not going to be managed internally. But that's unreliable since resize()
calling resize() on it could use "delete" operator on a malloc'ed
buffer, and end up having a new new[] allocated buffer. It was also
found that some of the callers were actually leaking memory through ASan (because the
buffer is not managed by the Vector instance).
IMHO best option would be to rewrite all this code using C structures
and malloc/free exclusively, since it would make all this cod eeasier to
maintain.
But for now, let's extend the Vector class to allow specifying an
external alloc/free function and let the Vector instance take care of
the ownership of the buffer in all scenarios.
Change-Id: Ie484a4762a7f77fe1b105188ea03a6f025730b82
This is a preparatory change that enables a possibility to choose
the amount of synch. sequences to be used for Access Burst (RACH)
detection. The VTY flag will be introduced in further changes.
There are two correlation types now:
- RACH (default) - TS0 only;
- EXT_RACH - all TS0, TS1, and TS2 together.
Change-Id: Ia4f20524350bb8c380d7e10360758eddae8b03e9
Related: OS#3054
According to 3GPP TS 05.02, section 5.2.7, there are three
synch. sequences for Access Bursts:
- TS0: GSM, GMSK (default),
- TS1: EGPRS, 8-PSK,
- TS2: EGPRS, GMSK.
Let's prepare everythyng to be able to detect all TS0-3 synch.
sequences, but keep detection of both TS1 and TS2 disabled
until the corresponding VTY option is introduced.
Change-Id: I838c21db29c54f1924dd478c2b34b46b70aab2cd
Related: OS#3054
This change allows to remove some wrong use of code as per compilation
warning:
osmo-trx/Transceiver52M/sigProcLib.cpp:1266:40: error:
‘void* memcpy(void*, const void*, size_t)’ writing to an object of non-trivially copyable type ‘class Complex<float>’; use copy-assignment or copy-initialization instead [-Werror=class-memaccess]
midMidamble->size() * sizeof(complex));
Change-Id: Id446711349bec70fa4e7c8efe0f7f9faf7e4f277
In commit a3dce85f
"sigProcLib: Use explicit NaN check in sinc table generation"
Use of std::isnan(double) was added without namespace specifier,
which may cause build issues depending on whether the C version
isnan() call is available. Add standard namespace to force C++
call usage and potential build issues.
Change-Id: I49328c43fdd690a4e6a2b2e949411aaf5674ead1
Using "x < 0.01" is a crude check for detecting NaN condition, which
occurs in a sinc call when x = 0 due to divide-by-zero. Use stdlib
isnan() call for this purpose. Also, as the table is created only
once during initialization, use double floats for table value
generation.
Change-Id: I3a838fe3139fa977dfe906246020a14451185714
Trigonometric sin/cos tables are unused after initialization.
There is no benefit to implementing lookup tables for run-once
operations. Also perform initial calculations in double width
because there is no penalty for doing so.
Change-Id: I45bba5daf8610cbba6af95b92c2142f2256491ff
The osmo-trx internals rely heavily on dynamic alloction of
I/Q signal vectors. In a number of cases there is no reason
to to use dynamic rather than stack based allocation. Convert
these cases accordingly.
Change-Id: If53da1bf77b5944b6117765fa98ce12e1ccdeede
The modulator vector to be shaped by Laurent C1 pulse is complex,
but was set as real. The error does not affect behaviour because
we only support complex-complex and complex-real calculations;
real-real convolution is not supported. So in this case the data
vector was already assumed to be complex despite the improper
flag setting.
Change-Id: I03afc6a93a01fde7a9a02e4eb9d201d3ee37d21a
Also remove entirely completely unused calls. Most of these
calls have been around since OpenBTS conception. Nearly a
decade is long enough time for deprecation.
Change-Id: Ifc122aaff23414c363b4b00f99061eed8a6902d0
This bug only affects generation of normal bursts filled with random bits which
are used in test mode. It doesn't affect operation of osmo-trx during normal
operation. That's why it has stayed unnoticed for so long.
Each Normal Burst has 3 tail bits, not 4.
Also it's better to set stealing bits to 0 for maximum compatibility. We may want to
introduce a selector for each bit whether to set it to 0, to 1 or to a random number.
Change-Id: I0377029556c8b681b3ba3b635bf19572b34546ea
The 'diversity' option was an experimental 2 antenna receiver
implementation for UmTRX. The implementation has not been
maintained and current working status is unknown.
In addition to code rot, Coverity is triggering errors in the
associated code sections.
Removal of code cleans up many cases of special handling that
were necessary to accommodate the implementation.
Change-Id: I46752ccf5dbcffbec806081dec03e69a0fbdcdb7
Convert negative value check on unsigned value to zero check
to avoid potential divide-by-zero error condition.
Change-Id: Ib0d7d1bceb5fe66e69345db93a74e3e0773a2257
Fixes: Coverity CID 165059
Improper length values will cause the polyphase resampler
rotation to fail. Check return and return NULL on error.
Change-Id: I3ad22f9fd7a20754f589c04258dcca3770474a9b
Fixes: Coverity CID 165235
demodCommon() used to scale input vector in place which changed original data.
That's a bad practice and is not really necessary, so I've changed the code to
scale burst after it's copied to a new vector during a delay operation.
Change-Id: Ic45f71b634e48808356d68925bb9f5783e0bf0d3
vectorSlicer() converts soft-bits from -1..+1 to 0..1 while we want
to keep SoftVector in -1..+1 mode until the last minute, because at some
point we'll want to transmit -1..+1 to osmo-bts instead of converting it
from 0..1 back to -1..+1 on the osmo-bts side.
Plus it removes code duplication - we call it once instead of twice.
Change-Id: Idd6ddd7ac219afb0df055a692632678b66373764
This makes it similar to 8-PSK demod and also saves a bit of lines ofcode and
should give us a tiny improvement in performance.
Ideally we need to remove vector slicing at all, because in osmo-bts-trx
we convert back to +-1.0 again (actually to +-127, but it doesn't mater).
So we should rather transmit +-1.0 values to avoid double conversion.
Change-Id: If9ed6f0f80fbe88c994b2f9c3cae91d0d57f4442
Documentation in sigProcLib.h was noticeably out of sync with the actual
implementation - e.g. not all arguments were documented and arguments
which are already removed are still in the documentation. Also argument
names were different between declaration in .h and implementation in .cpp
which was confusing.
I've fixed this for detect*Burst() functions.
Change-Id: I4dfd07125d9a1e9a42a78b79faff539f003deb16
Invert the initial and trailing diffential bits in the 4 sps
GMSK modulator to fix power-time mask irregularity at start
and end of burst. GSM tail bits are always zero, so we can
safely use fixed bit values.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Agilent E4406A measurement of TSC synchronized RF envelope measurement.
showed failing power levels at the tail end of EDGE bursts. Regression
traced back to following commit.
Commit d2b070369d
"uhd: Correct timing alignment in 8-PSK and GMSK downlink bursts"
In the patch, the EDGE burst was delayed one symbol, but erroneously
truncated at the tail end causing the failing RF envelope measurement.
The missing tail symbol did not appear to affect end-to-end EGPRS tests.
This patch corrects the truncation.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Reduce the burst detection threshold to pass more bursts to upper
layers, but force stricter requirements on the computation itself.
For the latter, we now require at least 5 samples (rather than 2)
to compute a peak-to-average value.
End result is increased burst detection at low SNR conditions with
a small increase in false positive bursts when no signal is present.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Add new radio interface "radioInterfaceMulti" for multi-carrier
support.
Only USRP B200/B210 devices are supported because of sample
rate requirements (3.2 Msps).
Only 4 SPS operation Tx/RX is supported.
8-PSK is supported.
Other options may be added at a later time
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Two buffers, inner and outer, are used in the transceiver
implementation. The outer buffer interfaces with the device receive
interface to guarantee timestamp aligned and contiguously allocated
sample buffers. The inner buffer absorbs vector size differences between
GSM bursts (156 or 157 samples) and the resampler interface (typically
fixed multiples of 65).
Reimplement the inner buffer with a ring buffer that allows fixed size
segments on the outer (resampler) portion and variable lengths (GSM
side) on the inner side. Compared to the previous stack-like version,
this implementation removes unnecessary copying of buffer contents.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Slightly widen the search range to accommodate timing jitter
on certain classes of devices. The expanded range minimizes
the possibility of missing bursts that arrive too early or
too late due to timing error.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
EDGE 8-PSK soft slicer was receiving input from the output of the
downsampler. Equalization and derotation were missing causing the
soft symbol output to be invalid.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Delay the EDGE downlink bursts by one symbol in order to match GMSK
pulse shaping group delay. The difference in group delay arises from
the dual pulse filter combination of the GMSK Laurent represenation
whereas 8-PSK uses a single pulse linear filter.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Previously MAXDLY value was applied to Normal Bursts, which was nice
when working with sloppy test equipment like CMD57, but useless for
real world usage. At the same time documentation and de facto usage
of MAXDLY in OsmoBTS and OpenBTS assumed that it actually applies to
Access Bursts (RACH). So this patch changes osmo-rx behavior to apply
MAXDLY to RACH bursts and introduces a new command MAXDLYNB for the
old behavior.
Timing recovery and single tap channel compensation are identical
in both GMSK and EDGE receivers. This is the section ahead of and
including the optional 4-1 downsampler. GMSK and EDGE specific
sections operate at 1 SPS.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
Setup generators for empty, random, and dummy bursts. This moves error
prone burst length handling out of the Transceiver and into the signal
processing core.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>
At 4 samples per symbol, we don't need to maintain the 156/157 sample
slot structure to account for the GSM 156.25 sample burst length.
Set the 4 SPS Laurent modulator to ignore the guard interval setting
and always output 625 sample sized bursts. The EDGE 8-PSK modulator
already has this behavior.
Signed-off-by: Tom Tsou <tom.tsou@ettus.com>