was not computationally feasible with my 3rd Gen i7 processor running
and airspy at 10 MHz. Restructured the polyphase channelizer to use
vector iteration structures that the JDK 9 JIT compiler can translate
into SIMD instructions. Separated the IFFT processing from the filter
process so that they can be executed on separate threads and this allows
the processing to occur real time.
P25 C4FM decoding at 25 kHz channel sample rate using an airspy
10MHz sample stream seems to be working nicely with ~25% CPU usage and
~300 to 500 MB memory usage. Incremental (parallel) channel decoding
should have a minimal increase in CPU loading since all of the 12.5 kHz
channels are available in the polyphase channelizer and the only
additional workload is the decoding of the 25 kHz sample stream.
There is still an issue with stopping channels and the channelizer keeps
processing, and occasionally when starting traffic channels something
deadlocks and causes out of memory.
adds Feed forward gain support to the C4FM DQPSK demodulator which seems
to improve output from the interpolating sample buffer at each symbol
period. Updates the symbol timing gain for C4FM to decrease the gain a
bit since the feed forward gain is providing a better interpolated symbol.
updates the LSM decoder to follow the same design pattern. Adds LSM
decoder support to the demodulator viewer application. The current
symbol timing gain for LSM is set too high -- is using the C4FM gain
and causes too much fluctuation in the timing error.
error calculator. After increasing the gain for symbol timing, this
new symbol evaluator class seems to track the signal very accurately
despite lots of noise and ISI.
based on the P25 LSM decoder using a costas loop and a gardner detector.
Updates the Demodulator Viewer FX application to add in a symbol chart
and a phase error chart. The decoder won't lock on yet ... still a wip.
when the tuner controller indicates that there is a center DC spike
blocked region or when the tuner has clipped outer spectral regions
resulting in less than 100% usable bandwidth.
produces a significant reduction in memory usage, especially for tuners
like the airspy and hackrf that produce 10/20 MSPS sample rates. This
is observed to be around 66% reduction in peak memory usage and provides
a more stable memory utilization across time reducing stress on the
garbage collector.
Use of these reusable memory buffers will be pushed further down into
the processing chain modules to ensure we keep new object creation
to the lowest levels possible.
calculator must be updated immediately any time the tuner's center
frequency changes so that a channel can be allocated immediately. So,
I changed source event handling to two parts where the channel calculator
is immediately updated and the channelizer configuration is deferred so
that it can happen on the buffer processor execution thread and we can
avoid any sync() locking on the channelizer or polyphase channel
sources.
messages and events can be accurately timestamped, and more importantly,
allocated traffic channels can request to be pre-loaded with buffered
samples to ensure that traffic channels don't miss any samples that
are currently missed due processing latency inherent in current buffer
sizing. This should reduce the clipping of starting messages or voice
on the traffic channels.
Restructures processing path for tuner frequency and sample rate change
events. The tuner now embeds these events in the sequence of buffers
that are generated immediately after the event occurs since those are
the buffers that will be impacted by the change. Previously, the change
event would be broadcast throughout the system and would impact any
buffered/cached buffers. This also reduces the number of source event
change listeners that have to be managed. Note: this change is not yet
implemented in all tuners ... only in the TestTuner implementation.
Updates the polyphase channelizer to now buffer channel results and then
dispatch the buffer to the channel output processors to improve efficiency
of using float arrays versus single value complex float processing.
