close all;
clear
echo off
Tx_n = 4;
Rx_n = 4;
index = 2;
frame_length=1200;
SNRindB=0:2:20;
for i=1:length (SNRindB),
    SNR(i)=10^(SNRindB(i)/10);
    a=rand(1,frame_length);
    for L=1:frame_length
        if a(L)>.5
            a(L)=1;
        else  a(L)=0;
        end
    end

    a1=modulation(a,index);
    a2=reshape(a1,Tx_n,frame_length/index/Tx_n);
    sigma = 1/sqrt(2*Rx_n*SNR(i));
    a5=zeros(4,frame_length/index/Tx_n);
    AWGN_noise = sigma*(randn(Rx_n,frame_length/index/Tx_n)+j*randn(Rx_n,frame_length/index/Tx_n));
  for k=1:frame_length/index/Tx_n
        H=(randn(Rx_n,Tx_n)+j*randn(Rx_n,Tx_n))/sqrt(2);
        r=H*a2(:,k)+AWGN_noise(:,k);
    % ML檢測算法
      
        count=0:1:15;
        count1=zeros(4,256);
       a4=zeros(4,256);
         for  n1=1:16
             for n2=1:16
       
              d=[-1,1];
              c1=de2bi(count(n2),4);
              c1=d(c1+1);
              c2=de2bi(count(n1),4);
              c2=d(c2+1);
              c=c1+j*c2;
              count1(:,(n1-1)*16+n2)=c.';
             a4(:,(n1-1)*16+n2)=r-H*count1(:,(n1-1)*16+n2);
             end;
         end;

             [gk k0]=min(sum(abs(a4).^2,1));
               a5(:,k)=count1(:,k0);
  end
       a6=reshape(a5,1,frame_length/index);
       a7=demodulation(a6,index);
    NumErr1 =sum(a7~=a)
    Ber1(i)=NumErr1/frame_length;
  
end

semilogy (SNRindB,Ber1,'-o');
grid on
xlabel('Eb/No in dB');
ylabel('error probability');