項目一：LDPC在碼刪除信道下譯碼
作者：互联网
clc;clear;
for loop = 1:1000
load PEGH_3_6_1024.mat
[m,n] = size(H);
% H_simple = zhu_rank(H); %H化为行最简阵
%***********************高斯消元开始****************************************
a = H;
for j = 1:m
for i = j:m-1
for k = i+1:m
if a(i,j)==0 && a(k,j)==1
a([i,k],:)=a([k,i],:);
end
end
end
for i = j+1:m
if a(i,j)==1 &&a(j,j)==1
a(i,:)=xor(a(i,:),a(j,:));
end
end
end
%F_F=a
%*********************调整顺序*****************
%********统计******
for i = 1:m
for j = i:n %最大值只能统计到n，不能统计到m+n，m为单位矩阵
if a(i,j)==1
order(i) = j;
break;
else
order(i) = m+1;%全零视为越界
end
end
end
order_1 = order;%读取顺序 %未整理前每行第一个不为0数字所在位置组成的向量
%*********第一次排序*****
for i = 1:m-1 %i，j是冒泡前后排序，不是行列排序
for j = i+1:m
if order_1(i)>order_1(j)
t = a(i,:);
a(i,:) = a(j,:);
a(j,:) = t;
t = order_1(i);
order_1(i) = order_1(j);
order_1(j) = t;
end
end
end
order_2 = order_1;
%调整顺序 % 整理后没行第一个不为0数字所在位置组成的向量
a_begin = a; %整理后的排序
%***************第一次行最简化简*******************************
for j = m:-1:1 %列元素排序，和上面行元素排序不同，不可替换
for i = j:-1:1 %****确定位置***
if a(i,j)~=0
countt(j) = i; %找到每一列最后一个非零元素所在的位置
break; %找到每一列从下至上第一个非零元素后，跳出本轮循环
else
countt(j) = m+1;
end
end
end
countt;
for j = m:-1:1
for i = countt(j)-1:-1:1 %全零列不影响
if a(i,j)~=0
a(i,:) = xor(a(i,:),a(countt(j),:));
end
end
end
a_simple = a;
%********************************************************
%********第二次统计&排序*******
for i = 1:m
for j = i:n %最大值只能统计到n，不能统计到m+n，m为单位矩阵
if a(i,j)==1
count = j;
break;
else
count = m+1;
end
end
order(i)= count;
end
order_3 = order; %读取顺序,便于行变换
for i = 1:m-1 %i，j是冒泡前后排序，不是行列排序
for j = i+1:m
if order_3(i)>order_3(j)
tt = a(i,:);
a(i,:) = a(j,:);
a(j,:) = tt;
t = order_3(i);
order_3(i) = order_3(j);
order_3(j) = t;
end
end
end
order_4 = order_3; %最终顺序
a_third = a ;
%*********第二次行最简化简:消除第一次化简的影响（非行阶梯）*******************
for j = m:-1:1
for i = j:-1:1
if a(i,j)~=0
record(j) = i;
break;
else
record(j) = m+1;
end
end
end
record;
for j = m:-1:1
for i = record(j)-1:-1:1
if a(i,j)~=0
a(i,:) = xor(a(i,:),a(record(j),:));
end
end
end
a_last = a;
%*********第二次化简完成************************************************
%G_G = trace(a(1:m,1:m));
BB= a(1:m,n-m+1:n); %变换后后半部分矩阵
H_simple=a(1:m,1:n); %变换后前半部分矩阵
%***********************高斯消元结束****************************************
%***********************生成矩阵运行开始************************************
Hs = H(1:m,1:n-m);
Hr = H(1:m,n-m+1:n);
H_H = [Hr Hs]; %只进行列变换
G = zhu_rank(H_H); %%%%%%
%*******先考虑列排序，再考虑行变换******
for j = m:-1:1
for i = j:-1:1
if G(i,j)~=0
G_order(j) = i;
break;
else
G_order(j) = m+1;
end
end
end
G_order ; %记录列向量中从下至上首个非零元素
%****************预备工作完成*********************
%****************开始第一次列变换***********************
for i = 1:m-1
for j = i+1:m
if G(i,i)==0 %和冒泡排序的区别
if G_order(i)>G_order(j) %G_order元素大小进行排序，对应G的列元素进行
t = G_order(i); %顺序
G_order(i) = G_order(j);
G_order(j) = t;
tt = G(:,i); %化简
G(:,i) = G(:,j);
G(:,j) = tt;
ttt = H_H(:,i);
H_H(:,i) = H_H(:,j); %影子
H_H(:,j) = ttt;
end
end
end
end
G_order_1 = G_order; %从小至大依次排序
G_1 = G;
%**************第一次列变换完成*************************
%******************消除1 2 2 3 4 6 6影响**********
for i = 1:m
if G_order_1(i)~=i
for j = i:m-1
for k = i+1:m
if G_order_1(k) == i
t = G_order_1 (i);
G_order_1 (i) = G_order_1 (k);
G_order_1 (k) = t;
tt = G_1(:,k);
G_1(:,k) = G_1(:,j);
G_1(:,j) = tt;
ttt = H_H(:,k);
H_H(:,k) = H_H(:,j);
H_H(:,j) = ttt;
end
end
end
end
end
G_order_2 = G_order_1; %重新排序
G_2 = G_1;
%******************前m列消除工作完成***********************
%******************与m列后面的列向量进行置换开始************
Hr_ji = trace(G_2(1:m,1:m));
if Hr_ji < m
for i = 1:m
if G_2(i,i)==0 %Hr 定位非满秩列
for j = m+1:n %G矩阵进行变换
G_2(:,[i,j])=G_2(:,[j,i]); %列变换
H_H(:,[i,j]) = H_H(:,[j,i]);
kkk = zhu_rank(G_2);
k = trace(kkk(1:m,1:m));
if k==Hr_ji+1
Hr_ji = Hr_ji+1 ; %秩加1
break;
else
G_2(:,[i,j])=G_2(:,[j,i]); %换回来
H_H(:,[i,j]) = H_H(:,[j,i]);
end
end
end
end
H_H_finished = H_H;
test = zhu_rank(H_H_finished);
else
%disp('原Hr矩阵满秩');
test = G_2;
end
middle_G = test(1:m,m+1:n);
G =[eye(n-m) middle_G'];%eye(n-m)
%***********************生成矩阵运行结束************************************
Cs = rand(1,m)<0.5; %随机生成信息位
C = mod(Cs*G,2); %利用生成矩阵G生成完整码字
W = zeros(8,n);
%***********************randperm随机删除码字开始***********************************
% % % for del_rate = 36:43
% % % C1 = C;
% % % rand_del = randperm(n,fix(del_rate*n*0.01));
% % % for bb = 1:length(rand_del)
% % % C1(rand_del(bb)) = 8; %将删除码字赋值8
% % % end
% % % C_del = C1;
%***********************randperm随机删除码字结束************************************
%***********************概率随机删除码字開始************************************
for del_rate = 40%36:43
C1 = C;
rand_del = rand(1,n)<del_rate*0.01;
find_1 =find(rand_del==1);
length(find_1);
for bb = 1:length(find_1)
C1(find_1) = 8;
end
C_del = C1;
%***********************概率随机删除码字结束************************************
%*********************删除信道译码开始**************************************
HH = H;
b = zeros(1,n);
while(length(find (C_del(1,1:n)==8))~=0) %&&length(find(HH(1:m,1:n)==1))~=0)
sect_first = zeros(1,m);
for i = 1:m
sect_first(i) = length(find(HH(i,1:n))==1);
end
for j = 1:n
for i = 1:m
if HH(i,j)==1&&C_del(j)~=8
b(i) = xor(b(i),C_del(j));
HH(i,j) = 0; % 刪除邊
end
end
end
for i = 1:m
hop = find(HH(i,1:n)==1);
if length(hop)==1
if C_del(hop)==8
C_del(hop) = b(i);
HH(i,hop) = 0;
else
HH(i,hop) = 0;
end
end
end
sect_last = zeros(1,m);
for i = 1:m
sect_last(i) = length(find(HH(i,1:n))==1);%%HH notH
end
% %************判斷環開始*******************
sect_check = (sect_first == sect_last);
sum_jud = 0;
for i = 1:m
sum_jud = sum_jud + sect_check(i);
end
if sum_jud == m
break;
end
%************判斷環結束*******************
end
%*********************删除信道译码完成**************************************
W(del_rate-35,1:n)=C_del; %Y_1译码输出码字
end
%*************duo個碼字檢驗開始*****************
for i = 1:8
sum = 0;
for j = 1:n
if W(i,j)~=C(1,j)
sum = sum +1;
end
end
BER(i,loop) = sum/n;
end
end
BER
for i=1:8
sum = 0;
for j = 1:loop
sum = sum + BER(i,j);
end
BER_averg(i) = sum/n;
end
BER_averg
t = 0.36:0.01:0.43;
y = 0,0,0,0,0.0002,0.0098,0.0802,0.1964;
plot(t,y)
%*************duo個碼字檢驗結束*****************
%*********************删除信道译码结束**************************************
%***************************误码率计算开始**********************************
% sum = 0;
% for i = 1:n
% if W(i) == C(i)
% count(i) = 0;
% else
% count(i) = 1;
% end
% sum = sum + count(i);
% end
% Ber(x-19,loop) = sum/n;
%**************************误码率计算结束************************************
%end
%end
%
% %*************统计平均开始*****************
% for i = 1:8
% BER = 0;
% for j = 1:loop
% BER = BER + Ber(i,j);
% end
% BER_1(i) = BER/loop;
% end
% BER_1
% %*************统计平均结束*****************
%**********************錯誤譯碼開始（無迭代）********************************
% % % for j = 1:n
% % % for i = 1:m
% % % if HH(i,j)==1&&(C_del(j)==0||C_del(j)==1)
% % % b(i) = xor(b(i),C_del(j)); %更新
% % % HH(i,j) = 8; %删除
% % % C_yi(j) = C_del(j); %记录
% % % k = find(HH(i,1:n)==1); %统计
% % % if length(k)==1 %传送
% % % if C_del(k)==8
% % % C_del(k) = b(i);
% % % HH(i,k) = 8 ; %删除
% % % if k<j %子程序
% % % for i_son = 1:m
% % % if HH(i_son,k)==1
% % % b(i_son) = xor(b(i_son),C_del(k)); %子更新
% % % HH(i_son,k) = 8;
% % % end
% % % end
% % % end
% % % else
% % % HH(i,k)=8;
% % % end
% % % end
% % % end
% % % end
% % % end
%****************************無迭代運算結束*********************************************
%Hr_ji = trace(G); %利用最简阵的迹的和判断Hr是否满秩
% ber_1 = 0;
% ber_2 = zeros(1,m);
% for x = 1:7
% for xx = 1:1000
% %*******************************************************************
% % Hr通过列变换变为满秩矩阵
% %*******************************************************************
% if Hr_ji < m
% for i = 1:m
% if G(i,i)~=1 %Hr 定位非满秩列
% for j = 1:m %Hs 扫列
% tt = Hs(:,j); %列交换
% Hs(:,j) = Hr(:,i);
% Hr(:,i) = tt;
% k = trace(zhu_rank(Hr));
% if k==Hr_ji+1
% Hr_ji = Hr_ji+1; %退步
% break %返回上一个循环
% end
% end
% end
% end
% CC = zhu_rank(Hr); %Hr最简矩阵
% Hr_ji_1 = trace(CC ); %Hr经过列变换后的迹
% end
% %**********************************************************************
% Hr_1 = zhu_nijuzhen(Hr); %Hr_1为Hr的逆矩阵
% yz = mod(Hr_1*Hr,2); %验证Hr确实是Hr的逆矩阵
% Cs = rand(1,m)<0.5; %系统码字，rand生成
% %*********Cr = Hr_1*Hs*Cs*********************************************
% HH = mod(Hr_1*Hs,2);
% Cr = mod(HH*Cs',2); %Cr校验码字计算得到
% C = [Cs Cr']; %生成码字C
% %**********************************************************************
% %****************此时H矩阵已经发生变化
% H = [Hs,Hr]; %校验矩阵H
% %*********************************************************************
% a = zeros(1,m); % 生成验证码字a
% k = fix(n*((36+x)/100));
% % k = fix(0.1*x*n);
% suiji_shan = randperm(n,k); %生成随机删除码字的地址，是一个向量
% for bb = 1:length(suiji_shan)
% C(suiji_shan(bb)) = 8; %将删除码字赋值8
% end
% Y = C; %经过删除信道删除后的码字
% k = delete_decode(H,a,Y); %经过删除信道译码后的码字
% ber(x,xx) = sum(xor(C,k)); %错误译码数量
% end
% % Ber = ber(x,:);
% end
标签：項目,end,Hr,tt,else,下譯碼,刪除,排序,order 来源： https://blog.csdn.net/zujiusheng/article/details/95371163