w4kek
/
WSJT-X
mirror of https://github.com/saitohirga/WSJT-X.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Experimental tweak to FT8 decoder. Try a second symbol metric if the first one fails - currently configured to use max-amplitude and max-log. 

git-svn-id: svn+ssh://svn.code.sf.net/p/wsjt/wsjt/branches/wsjtx@8093 ab8295b8-cf94-4d9e-aec4-7959e3be5d79
This commit is contained in:
Steven Franke 2017-09-17 15:34:32 +00:00
parent 78c257b4f0
commit b6740eb437
1 changed files with 91 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

119
lib/fsk4hf/ft8b.f90
View File

@ -10,16 +10,17 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
character*6 mygrid6
logical bcontest
real a(5)
real s1(0:7,ND),s2(0:7,NN)
real s1(0:7,ND),s2(0:7,NN),s1sort(8*ND)
real ps(0:7)
real bmeta(3*ND),bmetap(3*ND)
real llr(3*ND),llra(3*ND),llr0(3*ND),llrap(3*ND) !Soft symbols
real bmeta(3*ND),bmetb(3*ND),bmetap(3*ND)
real llr(3*ND),llra(3*ND),llr0(3*ND),llr1(3*ND),llrap(3*ND) !Soft symbols
real dd0(15*12000)
integer*1 decoded(KK),apmask(3*ND),cw(3*ND)
integer*1 msgbits(KK)
integer apsym(KK)
integer mcq(28),mde(28),mrrr(16),m73(16),mrr73(16)
integer itone(NN)
integer indxs1(8*ND)
integer icos7(0:6),ip(1)
integer nappasses(0:5) ! the number of decoding passes to use for each QSO state
integer naptypes(0:5,4) ! (nQSOProgress, decoding pass) maximum of 4 passes for now
@ -27,6 +28,7 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
complex ctwk(32)
complex csymb(32)
logical first,newdat,lsubtract,lapon,nagain
equivalence (s1,s1sort)
data icos7/2,5,6,0,4,1,3/
data mcq/1,1,1,1,1,0,1,0,0,0,0,0,1,0,0,0,0,0,1,1,0,0,0,1,1,0,0,1/
data mrrr/0,1,1,1,1,1,1,0,1,1,0,0,1,1,1,1/
@ -153,10 +155,15 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
s1(0:7,j)=s2(0:7,k)
enddo
call indexx(s1sort,8*ND,indxs1)
xmeds1=s1sort(indxs1(nint(0.5*8*ND)))
s1=s1/xmeds1
do j=1,ND
i4=3*j-2
i2=3*j-1
i1=3*j
! Max amplitude
ps=s1(0:7,j)
r1=max(ps(1),ps(3),ps(5),ps(7))-max(ps(0),ps(2),ps(4),ps(6))
r2=max(ps(2),ps(3),ps(6),ps(7))-max(ps(0),ps(1),ps(4),ps(5))
@ -167,6 +174,34 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
bmetap(i4)=r4
bmetap(i2)=r2
bmetap(i1)=r1
! Max log
ps=log(ps)
r1=max(ps(1),ps(3),ps(5),ps(7))-max(ps(0),ps(2),ps(4),ps(6))
r2=max(ps(2),ps(3),ps(6),ps(7))-max(ps(0),ps(1),ps(4),ps(5))
r4=max(ps(4),ps(5),ps(6),ps(7))-max(ps(0),ps(1),ps(2),ps(3))
! Metric for Cauchy noise
! r1=log(ps(1)**3+ps(3)**3+ps(5)**3+ps(7)**3)- &
! log(ps(0)**3+ps(2)**3+ps(4)**3+ps(6)**3)
! r2=log(ps(2)**3+ps(3)**3+ps(6)**3+ps(7)**3)- &
! log(ps(0)**3+ps(1)**3+ps(4)**3+ps(5)**3)
! r4=log(ps(4)**3+ps(5)**3+ps(6)**3+ps(7)**3)- &
! log(ps(0)**3+ps(1)**3+ps(2)**3+ps(3)**3)
! Metric for AWGN, no fading
! bscale=2.5
! b0=bessi0(bscale*ps(0))
! b1=bessi0(bscale*ps(1))
! b2=bessi0(bscale*ps(2))
! b3=bessi0(bscale*ps(3))
! b4=bessi0(bscale*ps(4))
! b5=bessi0(bscale*ps(5))
! b6=bessi0(bscale*ps(6))
! b7=bessi0(bscale*ps(7))
! r1=log(b1+b3+b5+b7)-log(b0+b2+b4+b6)
! r2=log(b2+b3+b6+b7)-log(b0+b1+b4+b5)
! r4=log(b4+b5+b6+b7)-log(b0+b1+b2+b3)
bmetb(i4)=r4
bmetb(i2)=r2
bmetb(i1)=r1
if(nQSOProgress .eq. 0 .or. nQSOProgress .eq. 5) then
! When bits 88:115 are set as ap bits, bit 115 lives in symbol 39 along
@ -209,12 +244,14 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
enddo
call normalizebmet(bmeta,3*ND)
call normalizebmet(bmetb,3*ND)
call normalizebmet(bmetap,3*ND)
ss=0.84
llr0=2.0*bmeta/(ss*ss)
llra=2.0*bmetap/(ss*ss) ! llr's for use with ap
apmag=4.0
scalefac=2.83
llr0=scalefac*bmeta
llr1=scalefac*bmetb
llra=scalefac*bmetap ! llr's for use with ap
apmag=scalefac*(maxval(abs(bmetap))*1.01)
! pass #
!------------------------------
@ -227,35 +264,36 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
! 7 ap pass 4, etc.
if(lapon) then
npasses=3+nappasses(nQSOProgress)
npasses=4+nappasses(nQSOProgress)
else
npasses=3
npasses=4
endif
do ipass=1,npasses
llr=llr0
if(ipass.eq.2) llr(1:24)=0.
if(ipass.eq.3) llr(1:48)=0.
if(ipass.le.3) then
if(ipass.eq.2) llr=llr1
if(ipass.eq.3) llr(1:24)=0.
if(ipass.eq.4) llr(1:48)=0.
if(ipass.le.4) then
apmask=0
llrap=llr
iaptype=0
endif
if(ipass .gt. 3) then
iaptype=naptypes(nQSOProgress,ipass-3)
if(ipass .gt. 4) then
iaptype=naptypes(nQSOProgress,ipass-4)
if(iaptype.ge.3 .and. (abs(f1-nfqso).gt.napwid .and. abs(f1-nftx).gt.napwid) ) cycle
if(iaptype.eq.1 .or. iaptype.eq.2 ) then ! AP,???,???
apmask=0
apmask(88:115)=1 ! first 28 bits are AP
apmask(144)=1 ! not free text
llrap=llr
if(iaptype.eq.1) llrap(88:115)=apmag*mcq/ss
if(iaptype.eq.2) llrap(88:115)=apmag*apsym(1:28)/ss
if(iaptype.eq.1) llrap(88:115)=apmag*mcq
if(iaptype.eq.2) llrap(88:115)=apmag*apsym(1:28)
llrap(116:117)=llra(116:117)
llrap(142:143)=llra(142:143)
llrap(144)=-apmag/ss
llrap(144)=-apmag
endif
if(iaptype.eq.3) then ! mycall, dxcall, ???
apmask=0
@ -263,8 +301,8 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
apmask(116:143)=1 ! hiscall
apmask(144)=1 ! not free text
llrap=llr
llrap(88:143)=apmag*apsym(1:56)/ss
llrap(144)=-apmag/ss
llrap(88:143)=apmag*apsym(1:56)
llrap(144)=-apmag
endif
if(iaptype.eq.4 .or. iaptype.eq.5 .or. iaptype.eq.6) then
apmask=0
@ -272,10 +310,10 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
apmask(116:143)=1 ! hiscall
apmask(144:159)=1 ! RRR or 73 or RR73
llrap=llr
llrap(88:143)=apmag*apsym(1:56)/ss
if(iaptype.eq.4) llrap(144:159)=apmag*mrrr/ss
if(iaptype.eq.5) llrap(144:159)=apmag*m73/ss
if(iaptype.eq.6) llrap(144:159)=apmag*mrr73/ss
llrap(88:143)=apmag*apsym(1:56)
if(iaptype.eq.4) llrap(144:159)=apmag*mrrr
if(iaptype.eq.5) llrap(144:159)=apmag*m73
if(iaptype.eq.6) llrap(144:159)=apmag*mrr73
endif
if(iaptype.eq.7) then ! ???, dxcall, ???
apmask=0
@ -283,8 +321,8 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
apmask(144)=1 ! not free text
llrap=llr
llrap(115)=llra(115)
llrap(116:143)=apmag*apsym(29:56)/ss
llrap(144)=-apmag/ss
llrap(116:143)=apmag*apsym(29:56)
llrap(144)=-apmag
endif
endif
@ -297,7 +335,7 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
if(ndepth.eq.3 .and. nharderrors.lt.0) then
ndeep=3
if(abs(nfqso-f1).le.napwid .or. abs(nftx-f1).le.napwid) then
if((ipass.eq.2 .or. ipass.eq.3) .and. .not.nagain) then
if((ipass.eq.3 .or. ipass.eq.4) .and. .not.nagain) then
ndeep=3
else
ndeep=4
@ -315,8 +353,8 @@ subroutine ft8b(dd0,newdat,nQSOProgress,nfqso,nftx,ndepth,lapon,napwid, &
if(any(decoded(73:75).ne.0)) cycle !Reject if any of the 3 extra bits are nonzero
if(nharderrors.ge.0 .and. nharderrors+dmin.lt.60.0 .and. &
.not.(sync.lt.2.0 .and. nharderrors.gt.35) .and. &
.not.(ipass.gt.1 .and. nharderrors.gt.39) .and. &
.not.(ipass.eq.3 .and. nharderrors.gt.30) &
.not.(ipass.gt.2 .and. nharderrors.gt.39) .and. &
.not.(ipass.eq.4 .and. nharderrors.gt.30) &
) then
call chkcrc12a(decoded,nbadcrc)
else
@ -363,3 +401,28 @@ subroutine normalizebmet(bmet,n)
bmet=bmet/bmetsig
return
end subroutine normalizebmet
FUNCTION bessi0(x)
! From Numerical Recipes
REAL bessi0,x
DOUBLE PRECISION p1,p2,p3,p4,p5,p6,p7,q1,q2,q3,q4,q5,q6,q7,q8,q9,y
SAVE p1,p2,p3,p4,p5,p6,p7,q1,q2,q3,q4,q5,q6,q7,q8,q9
DATA p1,p2,p3,p4,p5,p6,p7/1.0d0,3.5156229d0,3.0899424d0,1.2067492d0, &
0.2659732d0,0.360768d-1,0.45813d-2/
DATA q1,q2,q3,q4,q5,q6,q7,q8,q9/0.39894228d0,0.1328592d-1, &
0.225319d-2,-0.157565d-2,0.916281d-2,-0.2057706d-1, &
0.2635537d-1,-0.1647633d-1,0.392377d-2/
if (abs(x).lt.3.75) then
y=(x/3.75)**2
bessi0=p1+y*(p2+y*(p3+y*(p4+y*(p5+y*(p6+y*p7)))))
else
ax=abs(x)
y=3.75/ax
bessi0=(exp(ax)/sqrt(ax))*(q1+y*(q2+y*(q3+y*(q4 &
+y*(q5+y*(q6+y*(q7+y*(q8+y*q9))))))))
endif
return
end function bessi0