# Problem with rcosflt(Raised cosine filter) in Matlab 6.5 version

Started by June 1, 2004
 Hi, In a LAB experiment of digital communication system where rcosflt is used as channel, the error is produced due to extended length of output of the filter. When I ran the program in Matlab 5.1, its giving perfect result but when running in 6.5, its giving error due to group delay of the rcosflt. Can anyone suggest how can I remove this error? i tried to truncate, but its giving much error without noise which should be zero. Here is my program %Typical Answer for lab exp 3 % 1. M; modmap('qask',M); % 2. N 0; x=randint(N,1,M); %3 Fd=1; Fs; y = dmodce(x, Fd, Fs,'qask',M); figure(2); plot(y); title('QASK modulated signal'); % 4 y1=rcosflt(y,Fd,Fs,'Fs/fir'); %5 n=[0:1/Fs:N-1/Fs]; figure(3); subplot(2,1,1); plot(n,real(y),'r',n,real(y1),'b'); title('The comparision of real part of channel input and output signal'); subplot(2,1,2); plot(n,imag(y),'r',n,imag(y1),'b'); title('The comparision of imaginary part of channel input and output signal'); % 6 --- Channel output --- Dpoint; offset=Fs; figure (4); subplot(2,1,1); eyescat(y1,Fd,Fs,offset,Dpoint); Dpoint='*'; subplot(2,1,2); eyescat(y1,Fd,Fs,offset,Dpoint); % 7 --- Demodulated signal --- figure(5); y2modce(y1,Fd,Fs,'qask/eye',M); % 8 --- symbol error rate --- [numb,SER]=symerr(x,y2) % --- Gaussian noise --- noise_var=0.2; if noise_var>0 noise=(randn(length(y1),1)+j*randn(length(y1),1))*(noise_var)^0.5; y1=y1+noise; end %5(Noise) n=[0:1/Fs:N-1/Fs]; figure(6); subplot(2,1,1); plot(n,real(y),'r',n,real(y1),'b'); title('The comparision of real part of channel input and output signal with noise'); subplot(2,1,2); plot(n,imag(y),'r',n,imag(y1),'b'); title('The comparision of imaginary part of channel input and output signal with noise'); % 6(Noise) --- Channel output --- Dpoint; offset=Fs; figure (7); subplot(2,1,1); eyescat(y1,Fd,Fs,offset,Dpoint); Dpoint='*'; subplot(2,1,2); eyescat(y1,Fd,Fs,offset,Dpoint); % 7 --- Demodulated signal --- figure(8); y2modce(y1,Fd,Fs,'qask/eye',M); % 8 --- symbol error rate --- [numb,SER]=symerr(x,y2) numb SER __________________________________