CVS log for emstar/devel/loc/ar/ar.h
	Help

Request diff between arbitrary revisions / Display revisions graphically / Jump to LXR

Added a box version check (/etc/box_version) which should == 2 for new
sensor boards
Turns on and off the audio out amp correctly (in conjunction with
ensbox_sb)

A bit of a hack to clean up just now

fixed bug in treatment of new array geometry.
updated to assume new geometry.
fixed annoyances in plotit.

lots of changes for v2 boxes.

many changes for co tests.
* updates to emsbox cgis.  added some new support, some still to go
* improved netrec, added support for continuous recording
* fixed up the testtabs
* neihbors null term
* emproxy enalbes "control laptop mode"
* merged wavpack support in to record_from, integrate with netrec

several changes for CO trip.
* extended range for ranging processing (still not tested)
* added support for new arry geom and made it configurable (cmd line/run time)
* added support to set sample rate before starting ranging

extended timeout for mlat computation holdoff to 40 sec
added additional log message for "restart"

modified ar to precompute chirp and keep it for reference in selfrange
added selfrangelate fault
speeded up selfrange by cutting window by factor of 4
removed alberto's logging from default ar.run
removed mactime stuff from ar.run

still observe
* some problems with vxp server, possible driver/hw lockup
* timesync appears to get stuck.. reason unclear
* inconsistency in ssync

more fixes.. getting closer to working!

seems to mostly work...

changes to ar and mlat to support auto mode... not tested yet.

modified ar_doa to use the array geometry instead of being hard-coded
modified to optionally accont for the excess lag from the self-range
speaker

added prorustes to multilat, but commented out.
added delta estimatin, but it did not work, commented out..
fxed studentized, but it seemed to perform worse, strangely.. still in
though

checking in the GSL dependencies for mutlilat and ar

added 1 sec after, fixed the selfrange warning

added several features

* fixed estimation of noise floor, added normalization after clip extraction
* fixed confidence values to be against the better noise estimates
* recombination scaled by SNR.. (is this good?)
* added code to get noise stats

* added a bunch of upgraded scripts for experiments etc
* added fault reporting for failed self-range
* fixed huge error in code generation (wasnt using my selected codes)
* set up push for fault info via state sync

adjusted the offset for calibration..

added support for timestamp logging in hop-by-hop conversion in the
flood module.  added support to log those timestamps in acoustic ranging

moved some of the sound emitting code out to a separate file.
created new binary soundtest that continuously emits a pn code

implemented find first swing algorithm.
made this the default as it seems to work quite well.
calibrated in the lab.

changes to AR system to improve self-ranging.  uses crossing point
still, eventually use largest swing as metric..
offline supports reprocessing self range offsets.
added some utility scripts etc.

updated offline code to parse new file naming scheme
modified algorithms based on info gathered in initial experiments.
changed from finding zero-crossing to finding first peak over threshold.
may need to adjust threshold alg.  also changed to do first peak finding
on self-ranging.

remaining issues:  detect mic malfunction.  thresh estimation?
   codec offset between codecs?

limited code space to only those codes that work the best..

*** empty log message ***

added max_value, seed command arg

* changes to ranging code to support experiment mode
* new offline computation program
* fixes to the command interface to make it more sensible

support for card index other than 0

added amp enable/disable to ar

a few minor mods.. updated TODO

added various data collection features, fixed the confidence vals for angles

changed sqrt to sqrtf
changed fsqr to sqrf
minor improvments to ranging dsp algs.. correcting for fractional
samples in timesync etc

cool new DOA2 algorithms implemented

upgraded ar code
* add correction for rate differences
* modified code to work with new vxpc server

bunches of changes to ar..
added new logging/stats
added ability to enable dumps of time series data
fixed a bunch of weirdness

* fixes to ar, trying to get rid of memory leak
* added timeconversion to ssync measuremnet

added statesync connector, new stats, new queue status dev

* fixed problem with not using whole sequence
* doubled length of seq
* added high pass but not used yet
* fixed offset problem

first semi-working version of ar subsystem...
still needs to interface to statesync etc.
seems to have some mysterious bug in lag time provision..