mirror of
https://github.com/saitohirga/WSJT-X.git
synced 2024-11-18 10:01:57 -05:00
109 lines
3.4 KiB
Fortran
109 lines
3.4 KiB
Fortran
|
subroutine gf64_osd(mrsym,mrprob,mr2sym,mr2prob,cw)
|
||
|
|
use jt65_generator_matrix
|
||
|
|
integer mrsym(63),mrprob(63),mr2sym(63),mr2prob(63),cw(63)
|
||
|
|
integer indx(63)
|
||
|
|
integer gmrb(12,63)
|
||
|
|
integer correct(63)
|
||
|
|
integer correctr(63)
|
||
|
|
integer candidate(63)
|
||
|
|
integer candidater(63)
|
||
|
|
logical mask(63)
|
||
|
|
data correct/ & ! K1ABC W9XYZ EN37
|
||
|
|
41, 0, 54, 46, 55, 29, 57, 35, 35, 48, 48, 61, &
|
||
|
|
21, 58, 25, 10, 50, 43, 28, 37, 10, 2, 61, 55, &
|
||
|
|
25, 5, 5, 57, 28, 11, 32, 45, 16, 55, 31, 46, &
|
||
|
|
44, 55, 34, 38, 50, 62, 52, 58, 17, 62, 35, 34, &
|
||
|
|
28, 21, 15, 47, 33, 20, 15, 28, 58, 4, 58, 61, &
|
||
|
|
59, 42, 2/
|
||
|
|
|
||
|
|
correctr=correct(63:1:-1)
|
||
|
|
call indexx(mrprob,63,indx)
|
||
|
|
! do i=1,63
|
||
|
|
! write(*,*) i,correctr(indx(i)),mrsym(indx(i)),mr2sym(indx(i))
|
||
|
|
! enddo
|
||
|
|
|
||
|
|
nhard=count(mrsym.ne.correctr)
|
||
|
|
nerrtop12=count(mrsym(indx(52:63)).ne.correctr(indx(52:63)))
|
||
|
|
nerrnext12=count(mrsym(indx(40:51)).ne.correctr(indx(40:51)))
|
||
|
|
write(*,*) 'nerr, nerrtop12, nerrnext12 ',nerr,nerrtop12,nerrnext12
|
||
|
|
|
||
|
|
! The best 12 symbols will be used as the Most Reliable Basis
|
||
|
|
! Reorder the columns of the generator matrix in order of decreasing quality.
|
||
|
|
do i=1,63
|
||
|
|
gmrb(:,i)=g(:,indx(63+1-i))
|
||
|
|
enddo
|
||
|
|
! Put the generator matrix in standard form so that top 12 symbols are
|
||
|
|
! encoded systematically.
|
||
|
|
call gf64_standardize_genmat(gmrb)
|
||
|
|
|
||
|
|
! Add various error patterns to the 12 basis symbols and reencode each one
|
||
|
|
! to get a list of codewords. For now, just find the zero'th order codeword.
|
||
|
|
call gf64_encode(gmrb,mrsym(indx(63:52:-1)),candidate)
|
||
|
|
! Undo the sorting to put the codeword symbols back into the "right" order.
|
||
|
|
candidater=candidate(63:1:-1)
|
||
|
|
candidate(indx)=candidater
|
||
|
|
|
||
|
|
!write(*,'(63i3)') candidate
|
||
|
|
!write(*,'(63i3)') correctr
|
||
|
|
!write(*,'(63i3)') mrsym
|
||
|
|
nerr=count(correctr.ne.candidate)
|
||
|
|
write(*,*) 'Number of differences between candidate and correct codeword: ',nerr
|
||
|
|
if( nerr .eq. 0 ) write(*,*) 'Successful decode'
|
||
|
|
return
|
||
|
|
end subroutine gf64_osd
|
||
|
|
|
||
|
|
subroutine gf64_standardize_genmat(gmrb)
|
||
|
|
use gf64math
|
||
|
|
integer gmrb(12,63),temp(63),gkk,gjk,gkkinv
|
||
|
|
do k=1,12
|
||
|
|
gkk=gmrb(k,k)
|
||
|
|
if(gkk.eq.0) then ! zero pivot - swap with the first row with nonzero value
|
||
|
|
do kk=k+1,12
|
||
|
|
if(gmrb(kk,k).ne.0) then
|
||
|
|
temp=gmrb(k,:)
|
||
|
|
gmrb(k,:)=gmrb(kk,:)
|
||
|
|
gmrb(kk,:)=temp
|
||
|
|
gkk=gmrb(k,k)
|
||
|
|
goto 20
|
||
|
|
endif
|
||
|
|
enddo
|
||
|
|
endif
|
||
|
|
20 gkkinv=gf64_inverse(gkk)
|
||
|
|
do ic=1,63
|
||
|
|
gmrb(k,ic)=gf64_product(gmrb(k,ic),gkkinv)
|
||
|
|
enddo
|
||
|
|
do j=1,12
|
||
|
|
if(j.ne.k) then
|
||
|
|
gjk=gmrb(j,k)
|
||
|
|
do ic=1,63
|
||
|
|
gmrb(j,ic)=gf64_sum(gmrb(j,ic),gf64_product(gmrb(k,ic),gjk))
|
||
|
|
enddo
|
||
|
|
endif
|
||
|
|
enddo
|
||
|
|
enddo
|
||
|
|
|
||
|
|
return
|
||
|
|
end subroutine gf64_standardize_genmat
|
||
|
|
|
||
|
|
subroutine gf64_encode(gg,message,codeword)
|
||
|
|
!
|
||
|
|
! Encoder for a (63,12) Reed-Solomon code.
|
||
|
|
! The generator matrix is supplied in array gg.
|
||
|
|
!
|
||
|
|
use gf64math
|
||
|
|
integer message(12) !Twelve 6-bit data symbols
|
||
|
|
integer codeword(63) !RS(63,12) codeword
|
||
|
|
integer gg(12,63)
|
||
|
|
|
||
|
|
codeword=0
|
||
|
|
do j=1,12
|
||
|
|
do i=1,63
|
||
|
|
iprod=gf64_product(message(j),gg(j,i))
|
||
|
|
codeword(i)=gf64_sum(codeword(i),iprod)
|
||
|
|
enddo
|
||
|
|
enddo
|
||
|
|
|
||
|
|
return
|
||
|
|
end subroutine gf64_encode
|
||
|
|
|