Signals and Systems II

Lecture "Signals and Systems II"

Basic Information

	Lecture overview
	Lecturers:		Gerhard Schmidt (lecture), Patricia Piepjohn and Johannes Hoffmann (exercise)
	Room:		Digital format (videos and live question time)
	Language:		German
	Target group:		Students in electrical engineering and computer engineering
	Prerequisites:		Signals and Systems, Part 1
	Contents:		This course teaches basics in systems theory for electrical engineering and information technology. This basic course is focused on stochastic signals and linear systems. Topic overview: Modulation State-space description Stochastic signals and corresponding spectra Linear systems and stochastic signals Ideal systems Add-ons (transforms)
	References:		E. Hänsler: Statistische Signale, 3.Auflage, Springer, 2001 H. D. Lüke: Signalübertragung, Springer, 1995 H. W. Schüßler: Netzwerke, Signale und Systeme II: Theorie kontinuierlicher und diskreter Signale und Systeme, Springer, 1991

News

Results of the exam from 07.09.2020 are online.

The exam inspection has been scheduled for 30.09.2020 at 15:00 Uhr. You will receive an email with further information. Please note that a pre-registration is required.

Lecture Slides

	Link		Content
			Slides of the lecture "Modulation" (Linear and nonlinear modulation, digitalization)
			Slides of the lecture "State-space description" (State-space representation of linear, time-invariant systems)
			Slides of the lecture "Stochastic signals and corresponding spectra" (Random processes, power spectral densities)
			Slides of the lecture "Stochastic signals and linear time-invariant systems" (Random processes, power spectral densities)
			Slides of the lecture "Ideal systems" (Lowpass filter, magnitude and phase distortions)
			Slides of the lecture "Extensions on transformations" (Spectra of sampled continuous and discrete signals, convergence conditions and inverse transformations)

Lecture Videos

	Video		Content		Please watch until
					xx.xx.2020

Exercises

Details will follow soon.

	Link		Content		Please try until
					xx.xx.2020

Solutions for the Exercises

	Link		Content

Exercise Videos

	Video		Content		Please watch until
					xx.xx.2020

Evaluation

	Link		Content		Link		Content
			Current evaluation				Completed evaluations

Matlab demos

Central limit theorem

%**************************************************************************
% Parameter fuer die Generierung von gleichverteilten Zufallsvariablen
%**************************************************************************
m_v   =      0.0;        % Mittelwert
var_v =      1.0;        % Varianz
N     = 400000;          % Anzahl der Variablen pro Experiment
M     =      4;          % Anzahl der summierten Zufallsvariablen

%**************************************************************************
% Erzeugen der Zufallszahlen
%**************************************************************************
V = ((rand(M,N) - 0.5) / sqrt(12) * sqrt(var_v)) + m_v;

%**************************************************************************
% Nicht-normierte Histogramm-Analyse der ersten Zufallsvariablen
%**************************************************************************
fig = figure(1);
set(fig,'Units','Normalized');
set(fig,'Position',[0.1 0.1 0.8 0.8]);

[h,x] = hist(V(1,:),20);
bar(x,h,0.7);
ylabel('Haeufigkeit');
xlabel('Variable v_1');
grid on

%return;

%**************************************************************************
% Histogramm-Analyse der einzelnen Zufallsvariablen
%**************************************************************************
A_min    =   -0.6;
A_max    =    0.6;
IntWidth =    (A_max - A_min) / 50;

IntVec   = A_min:IntWidth:A_max;
HistRes  = zeros(M,length(IntVec));

for k = 1:M
    HistRes(k,:) = hist(V(k,:),IntVec) / (N * IntWidth);
end;

%**************************************************************************
% Darstellung der einzelnen Histogrammanalysen
%**************************************************************************
fig = figure(2);
set(fig,'Units','Normalized');
set(fig,'Position',[0.1 0.1 0.8 0.8]);

for k = 1:min(4,M)
   subplot(2,2,k)
   bar(IntVec,HistRes(k,:),0.7)
   grid on;
   xlabel(['Variable v_',num2str(k)']);
   axis([A_min A_max -0.05*max(HistRes(k,:)) 1.05*max(HistRes(k,:))]);
   ylabel('Relative Haeufigkeit');   
end;

%return;

%**************************************************************************
% Summation der Zufallsvariablen
%**************************************************************************
v = zeros(1,N);
for k = 1:M
    v = v + V(k,:);
end;

%**************************************************************************
% Histogrammanalyse der Summation
%**************************************************************************
histRes = hist(v,IntVec) / (N * IntWidth);

%**************************************************************************
% Darstellung der Histogrammanalyse der Summe
%**************************************************************************
fig = figure(3);
set(fig,'Units','Normalized');
set(fig,'Position',[0.1 0.1 0.8 0.8]);

bar(IntVec,histRes,0.7)
grid on;
xlabel('Variable y');
axis([A_min A_max -0.05*max(histRes) 1.05*max(histRes)]);
ylabel('Relative Haeufigkeit');   

%**************************************************************************
% Darstellung einer Gauss-Dichte
%**************************************************************************
f_g = 1/(std(v)*sqrt(2*pi)) * exp( -0.5 * ((IntVec-mean(v))/std(v)).^2);
hold on
plot(IntVec,f_g,'r','LineWidth',2);
hold off

Matlabdemo for exercise ''Amplitudenmodulation''

%% Example for single-sideband modulation and "double"-sideband modulation
%
% -> general concept
% -> advantage of singe-sibeband modulation (reduced error caused by frequency shift)
%
% Jens Reermann
%% Reset
clear all;
close all;
clc;
%% Parameter
f_s=44.1e3;
PLAY_SIGNAL=0;
AUDIO_ENABLED = 0;

%%
fc=16e3;
f_error=1500;
phi_error = 0; %pi*0.48;

%% Load signal
if AUDIO_ENABLED == 1
    [u, fs_sig] = audioread( 'signals/interview_heino.wav' );
    n=0:length(u)-1;
else
    fsig=2000;
    fs_sig=100e3;
    T=10;
    Nsig=T*fs_sig;
    n=0:Nsig-1;
    u = sin(2*pi*fsig/fs_sig.*n)'+randn(Nsig,1);
end


%% LP 3kHz | 4kHz
f_l=3e3;
f_h=4e3;
f = [ f_l f_h];
a = [1 0];
rp = 1;           
rs = 120;     
dev = [ (10^(rp/20)-1)/(10^(rp/20)+1) 10^(-rs/20)]; 
[n_ord,fo,ao,w] = firpmord(f,a,dev,f_s);
b = firpm(n_ord+1,fo,ao);
u_f = filter(b,1,u);

%% Carrier signals

c=cos(2*pi*n*fc/f_s)';
c_e=cos(2*pi*n*(fc+f_error)/f_s+phi_error)';

%% Modulation

% DSB
m=c.*u_f;

% BP: DSB->SSB
f_bp = [ fc fc+1000 fc+f_l fc+f_h];
a_bp = [0 1  0]; 
dev_bp = [10^(-rs/20) (10^(rp/20)-1)/(10^(rp/20)+1) 10^(-rs/20)]; 
[n_ord_bp,fo_bp,ao_bp,w] = firpmord(f_bp,a_bp,dev_bp,f_s);
b_bp = firpm(n_ord_bp+1,fo_bp,ao_bp);

m_bp = filtfilt(b_bp,1,m);    %SSB

%% Demodulation

% *cos
u_d=m.*c;
u_d_bp=m_bp.*c;
u_de=m.*c_e;
u_de_bp=m_bp.*c_e;
% LP
u_df = filter(b,1,u_d);
u_df_bp = filter(b,1,u_d_bp);
u_def = filter(b,1,u_de);
u_def_bp = filter(b,1,u_de_bp);

%% Play signalss
if PLAY_SIGNAL==1
    sound(u_f,f_s)
    pause
    sound(u_df,f_s)
    pause
    sound(u_def,f_s)
    pause
    sound(u_def_bp,f_s)
end

%% PSD estimation
N_FFT=2^12;
[Puu, F] = pwelch(u,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pufuf, F] = pwelch(u_f,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pmm, F] = pwelch(m,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pmm_bp, F] = pwelch(m_bp,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pudfudf, F] = pwelch(u_df,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pudefudef, F] = pwelch(u_def,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pudfudf_bp, F] = pwelch(u_df_bp,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);
[Pudefudef_bp, F] = pwelch(u_def_bp,hann(N_FFT),N_FFT-N_FFT/4,N_FFT,f_s);

%% Plotting (Modulation)
[val1 ind1] =max(10*log10(Puu));
[val2 ind2] =max(10*log10(Pmm));
[val3 ind3] =max(10*log10(Pmm_bp));

if AUDIO_ENABLED
    y_lims=[-160 -50];
else
    y_lims=[-100 10];
end

figure

plot(F, 10*log10(Puu));
text(F(ind1),val1,[num2str(val1) 'dB']);
legend('S_{uu}');
xlabel('f / Hz');ylabel('LDS / dB');xlim([0 f_s/2]);ylim(y_lims);
pause
hold on;
plot(F, 10*log10(Pufuf),'r');
legend('S_{uu}','S_{ufuf}');
xlabel('f / Hz');ylabel('LDS / dB');xlim([0 f_s/2]);ylim(y_lims);
pause
plot(F, 10*log10(Pmm),'g');
text(F(ind2),val2,[num2str(val2) 'dB']);
legend('S_{uu}','S_{ufuf}','S_{mm}');
xlabel('f / Hz');ylabel('LDS / dB');xlim([0 f_s/2]);ylim(y_lims);
pause
plot(F, 10*log10(Pmm_bp),'k');
text(F(ind3),val3,[num2str(val3) 'dB']);
legend('S_{uu}','S_{ufuf}','S_{mm}','S_{mm_{BP}}');
xlabel('f / Hz');ylabel('LDS / dB');xlim([0 f_s/2]);ylim(y_lims);
pause
hold off;

%% Plotting (Demodulation)

[val3 ind3] =max(10*log10(Pufuf));
[val4 ind4] =max(10*log10(Pudfudf));
[val5 ind5] =max(10*log10(Pudefudef));
[val6 ind6] =max(10*log10(Pudefudef_bp));

figure
xlim([0 f_h+1000]);
ylabel('LDS / dB');
xlabel('f / Hz');
if AUDIO_ENABLED
    ylim=[-160 -50];
else
    ylim=[-100 10];
end

hold on;
plot(F, 10*log10(Pufuf));
text(F(ind3),val3,[num2str(val3) 'dB']);
legend('S_{ufuf}');
pause
plot(F, 10*log10(Pudfudf),'g');
text(F(ind4),val4,[num2str(val4) 'dB']);
legend('S_{ufuf}','S_{ZSB}');
pause
plot(F, 10*log10(Pudefudef),'r');
text(F(ind5),val5,[num2str(val5) 'dB']);
legend('S_{ufuf}','S_{ZSB}','S_{eeZSB}');
pause
plot(F, 10*log10(Pudefudef_bp),'k');
text(F(ind6),val6,[num2str(val6) 'dB']);
legend('S_{ufuf}','S_{ZSB}','S_{eeZSB}','S_{eeESB}');
pause

hold off;

disp( [ num2str(20*log10(cos(phi_error))) ] );

Matlab examples from the exercises as a zip file

Current Exam

Previous Exams

	Link		Content		Link		Content
			Exam of the summer term 2020				Corresponding solution
			Exam of the winter term 2019/2020				Corresponding solution
			Exam of the summer term 2019				Corresponding solution
			Exam of the winter term 2018/2019				Corresponding solution
			Exam of the summer term 2018				Corresponding solution

On Wednesday, 2nd of September, Liv Preßer, interviewed us about or research on Parkinson speech analysis and corresponding therapy. The story appeared in Liv's podcast called "Was steckt dahinter". If you would like to hear it, please have a look on this website. Beside this website there are a couple of further possibilities: e.g. search for "Was steckt dahinter" on spotify.

Beside Liv Presser and Gerhard Schmidt, also Corina Maetzler (from the geriatrics department of the university hospital here in Kiel (UKHS) and Heidi Nebel from the neurologic department (also UKSH) participated in the podcast recording. Together, we reported on our research results, the probelms that we faced when conduction our experiments and patient recordings, and on our future plans.

More details on Liv's podcast can be found here.

Sidebar Menu