First attempt at a smoothing loop inside avg65. Needs testing!

git-svn-id: svn+ssh://svn.code.sf.net/p/wsjt/wsjt/branches/wsjtx@6546 ab8295b8-cf94-4d9e-aec4-7959e3be5d79
This commit is contained in:
Joe Taylor 2016-03-21 19:11:38 +00:00
parent ceb72eb8e9
commit e84ce2bad1
3 changed files with 76 additions and 12 deletions

View File

@ -15,8 +15,6 @@ subroutine decode65b(s2,nflip,mode65,ntrials,naggressive,ndepth, &
do i=1,64
s3(i,j)=s2(i+2,k)
enddo
k=mdat2(j) !Points to data symbol
if(nflip.lt.0) k=mdat(j)
enddo
nadd=mode65

View File

@ -3,7 +3,7 @@ program fer65
! End-to-end simulator for testing JT65.
! Options
! jt65sim jt65sim
! jt65sim jt65
!----------------------------------------------------------------
! -a aggressive
! -d Doppler spread -d depth

View File

@ -224,6 +224,7 @@ contains
call avg65(nutc,nsave,sync1,dtx,nflip,nfreq,mode65,ntol, &
ndepth,ntrials,naggressive,nclearave,neme,mycall, &
hiscall,hisgrid,nftt,avemsg,qave,deepave,nsum,ndeepave)
nsmo=param(9)
if (associated(this%callback) .and. nsum.ge.2) then
call this%callback(nutc,sync1,nsnr,dtx-1.0,nfreq,ndrift, &
@ -311,10 +312,12 @@ contains
integer nfsave(MAXAVE)
integer listutc(10)
integer nflipsave(MAXAVE)
! real s1a(-255:256,126)
! real s1save(-255:256,126,MAXAVE)
real s1b(-255:256,126)
real s1save(-255:256,126,MAXAVE)
real s2(66,126)
real s3save(64,63,MAXAVE)
real s3b(64,63)
real s3c(64,63)
real dtsave(MAXAVE)
real syncsave(MAXAVE)
logical first
@ -326,8 +329,8 @@ contains
nfsave=0
dtdiff=0.2
first=.false.
s3b=0.
s3save=0.
s1save=0.
nsave=1 !### ???
endif
@ -335,22 +338,24 @@ contains
if(nutc.eq.iutc(i) .and. abs(nhz-nfsave(i)).le.ntol) go to 10
enddo
! Save data for message averaging
! Save data for message averaging
iutc(nsave)=nutc
syncsave(nsave)=snrsync
dtsave(nsave)=dtxx
nfsave(nsave)=nfreq
nflipsave(nsave)=nflip
! s1save(-255:256,1:126,nsave)=s1
s1save(-255:256,1:126,nsave)=s1
s3save(1:64,1:63,nsave)=s3a
10 syncsum=0.
dtsum=0.
nfsum=0
nsum=0
s1b=0.
s3b=0.
s3c=0.
do i=1,64 !Consider all saved spectra
do i=1,MAXAVE !Consider all saved spectra
cused(i)='.'
if(iutc(i).lt.0) cycle
if(mod(iutc(i),2).ne.mod(nutc,2)) cycle !Use only same (odd/even) seq
@ -358,6 +363,7 @@ contains
if(abs(nfreq-nfsave(i)).gt.ntol) cycle !Freq must match
if(nflip.ne.nflipsave(i)) cycle !Sync type (*/#) must match
s3b=s3b + s3save(1:64,1:63,i)
s1b=s1b + s1save(-255:256,1:126,i)
syncsum=syncsum + syncsave(i)
dtsum=dtsum + dtsave(i)
nfsum=nfsum + nfsave(i)
@ -412,12 +418,72 @@ contains
nadd=nsum*mode65
nftt=0
! nexp_decode=0 !### not used, anyway
df=1378.125/512.0
call extract(s3b,nadd,mode65,ntrials,naggressive,ndepth,mycall, &
hiscall,hisgrid,nexp_decode,ncount,nhist,avemsg,ltext,nftt,qual)
! Do the smoothing loop
qualbest=0.
minsmo=0
maxsmo=0
if(mode65.ge.2) then
minsmo=nint(width/df)
maxsmo=2*minsmo
endif
nn=0
do ismo=minsmo,maxsmo
if(ismo.gt.0) then
do j=1,126
call smo121(s1b(-255,j),512)
if(j.eq.1) nn=nn+1
if(nn.ge.4) then
call smo121(s1b(-255,j),512)
if(j.eq.1) nn=nn+1
endif
enddo
endif
do i=1,66
jj=i
if(mode65.eq.2) jj=2*i-1
if(mode65.eq.4) then
ff=4*(i-1)*df - 355.297852
jj=nint(ff/df)+1
endif
s2(i,1:126)=s1b(jj,1:126)
enddo
do j=1,63
k=mdat(j) !Points to data symbol
if(nflip.lt.0) k=mdat2(j)
do i=1,64
s3c(i,j)=4.e-5*s2(i+2,k)
enddo
enddo
call extract(s3c,nadd,mode65,ntrials,naggressive,ndepth,mycall, &
hiscall,hisgrid,nexp_decode,ncount,nhist,avemsg,ltext,nftt,qual)
if(nftt.eq.1) then
nsmo=ismo
param(9)=nsmo
exit
else if(nftt.eq.2) then
if(qual.gt.qualbest) then
decoded_best=decoded
qualbest=qual
nnbest=nn
nsmobest=ismo
endif
endif
enddo
if(nftt.eq.2) then
decoded=decoded_best
qual=qualbest
nsmo=nsmobest
param(9)=nsmo
endif
900 continue
return
end subroutine avg65