Skip to content

Commit

Permalink
Einige kleine Änderungen für bessere Übersicht, Tabellen mit Rändern …
Browse files Browse the repository at this point in the history
…etc. in Tabellen mit mehreren Zeilen in einer Zelle ist \newline besser als \ und &
  • Loading branch information
dav1d-wright committed Mar 30, 2013
1 parent 83e6752 commit 811bf0c
Showing 1 changed file with 49 additions and 36 deletions.
85 changes: 49 additions & 36 deletions sections/07_RandomProcess.tex
Original file line number Diff line number Diff line change
Expand Up @@ -106,14 +106,18 @@ \subsubsection{Schwach Stationär (WSS - Wide Sense Stationary) - Stationarität
\mathbb{R})$$
%Zudem gilt:\\
\renewcommand{\arraystretch}{1.4}
\begin{tabular}[c]{ p{3.3cm} p{6.5cm} p{8cm} }
\begin{tabular}[c]{ |p{3.3cm}| p{6.5cm} |p{8cm}| }
\hline
\textbf{Mittelwert}: & $E[X(t)] = \mu_{X}(t) = \text{const.}$
& bleibt über die ganze Zeit konstant\\
\textbf{quad. Mittelwert}: & $E[X^{2}(t)] = R_{XX}(0)$ \\
\hline
\textbf{quad. Mittelwert}: & $E[X^{2}(t)] = R_{XX}(0)$ & \\
\hline
\textbf{Autokorrelation}: & $R_{XX}(t_{1},t_{2}) = R_{XX}(\tau)$
& \multirow{2}{8cm}{nur \textbf{abhängig} von der \textbf{Zeitdifferenz} $(\tau = t_2 - t_1)$ und \textbf{nicht direkt} von
der \textbf{Zeit} $t$} \\
\textbf{Autokovarianz}: & $ C_{XX}(t_{1},t_{2}) = R_{XX}(\tau) - \mu_{X}(t)^{2} = C_{XX}(\tau)$ \\
& nur \textbf{abhängig} von der \textbf{Zeitdifferenz} $(\tau = t_2 - t_1)$ und \textbf{nicht direkt} von der \textbf{Zeit} $t$ \\
\hline
\textbf{Autokovarianz}: & $ C_{XX}(t_{1},t_{2}) = R_{XX}(\tau) - \mu_{X}(t)^{2} = C_{XX}(\tau)$ & \\
\hline
\end{tabular}
\renewcommand{\arraystretch}{1}

Expand Down Expand Up @@ -197,26 +201,29 @@ \subsubsection{Thermisches Widerstandsrauschen}
Formeln in diesem Abschnitt gelten für \textbf{stationäre} Prozesse. \\

\renewcommand{\arraystretch}{1.6}
\begin{tabular}[c]{ p{4cm} p{6cm} p{7.5cm} }
\textbf{Autokorrelation}: &
$R_{XX}(\tau) = E[X(t)X(t+\tau)]$ \\
& $\mid \! R_{XX}(\tau) \! \mid \leq R_{XX}(0) = E[X^{2}(t)]$
& $R_{XX}(-\tau) = R_{XX}(\tau) \quad$ (gerade)\\
\textbf{Kreuzkorrelation}: &
$R_{XY}(\tau) = E[X(t)Y(t+\tau)]$
& $R_{XY}(-\tau) = R_{YX}(\tau) \quad$ (Reihenfolge Indizes!) \\
& $|R_{XY}(\tau)| \leq \frac{1}{2} \left[ R_{XX}(0)+R_{YY}(0)\right] $
& $|R_{XY}(\tau)| \leq \sqrt{R_{XX}(0)R_{YY}(0)}$ \\
\textbf{Autokovarianz}: &
\multicolumn{2}{l}{$C_{XX}(\tau) = E\!\left[ \left( X(t) - E[X(t)] \right) \cdot
\left( X(t+\tau) - E[X(t+\tau)] \right) \right] =
R_{XX}(\tau) - \mu^{2}_{X} $} \\
\textbf{Kreuzkovarianz}: &
\multicolumn{2}{l}{$C_{XY}(\tau) = E\!\left[ \left( X(t) - E[X(t)] \right) \cdot
\left( Y(t+\tau) - E[Y(t+\tau)] \right) \right] =
R_{XY}(\tau) - \mu_{X}\mu_{Y} $}\\
& \multicolumn{2}{l}{Zufallsprozesse bezeichnet man als zueinander
\textbf{unkorreliert}, wenn $C_{XY}(\tau) = 0$}
\begin{tabular}[c]{|p{4cm}|p{6cm}|p{7.5cm}|}
\hline
\textbf{Autokorrelation}: &
$R_{XX}(\tau) = E[X(t)X(t+\tau)]$\newline
$R_{XX}(\tau) \! \mid \leq R_{XX}(0) = E[X^{2}(t)]$
&$R_{XX}(-\tau) = R_{XX}(\tau) \quad$ (gerade)\\
\hline
\textbf{Kreuzkorrelation}: &
$R_{XY}(\tau) = E[X(t)Y(t+\tau)]$
& $R_{XY}(-\tau) = R_{YX}(\tau) \quad$ (Reihenfolge Indizes!) \\
& $|R_{XY}(\tau)| \leq \frac{1}{2} \left[ R_{XX}(0)+R_{YY}(0)\right] $
& $|R_{XY}(\tau)| \leq \sqrt{R_{XX}(0)R_{YY}(0)}$ \\
\hline
\textbf{Autokovarianz}: &
\multicolumn{2}{l|}{$C_{XX}(\tau) = E\!\left[ \left( X(t) - E[X(t)] \right) \cdot
\left( X(t+\tau) - E[X(t+\tau)] \right) \right] =
R_{XX}(\tau) - \mu^{2}_{X} $} \\
\hline
\textbf{Kreuzkovarianz}: &
\multicolumn{2}{l|}{$C_{XY}(\tau) = E\!\left[ \left( X(t) - E[X(t)] \right) \cdot
\left( Y(t+\tau) - E[Y(t+\tau)] \right) \right] = R_{XY}(\tau) - \mu_{X}\mu_{Y} $}\\
& \multicolumn{2}{l|}{Zufallsprozesse bezeichnet man als zueinander \textbf{unkorreliert}, wenn $C_{XY}(\tau)0$}\\
\hline
\end{tabular}
\renewcommand{\arraystretch}{1}

Expand All @@ -242,38 +249,44 @@ \subsubsection{Thermisches Widerstandsrauschen}
Ein Zufallsprozess wird durch ein LTI-System übertragen. \hspace{2cm} $Y(t) = L[X(t)] \Rightarrow
Y(t) = h(t) \ast X(t)$ \vspace{0.3cm}\\
\renewcommand{\arraystretch}{1.4}
\begin{tabular}[c]{ p{2cm} p{8.5cm} p{8cm} }
& \textbf{Allgemein} & \textbf{WSS-Prozess} \\
\begin{tabular}[c]{|p{2cm}| p{8.5cm} |p{8cm}| }
\hline
&\textbf{Allgemein}
& \textbf{WSS-Prozess} \\
\hline
\textbf{Mittelwert}
& $\mu_{Y}(t) = h(t) \ast \mu_{X}(t)$
& $\mu_{Y} = H(0) \mu_{X}$ \\
& $E(Y(t)) = h(t) \ast E(X(t))$
& $E(Y) = H(0) \cdot E(X)$ \\
\hline
\textbf{Auto-korrelation\textcolor{red}{*}}
& {$R_{YY}(t_{1},t_{2}) = \int\limits_{-\infty}^{+\infty}
\int\limits_{-\infty}^{+\infty} h(\alpha) h(\beta)
R_{XX}(t_{1}-\alpha, t_{2}-\beta) \; d\alpha \; d\beta$}
& {$R_{YY}(\tau) = \int\limits_{-\infty}^{+\infty}
\int\limits_{-\infty}^{+\infty} h(\alpha) h(\beta)
R_{XX}(\tau+\alpha-\beta) \; d\alpha \; d\beta$} \\
\hline
\textbf{Spektrale Leistung}
&
& $\boxed{S_{YY}(\omega)= H^{\ast}(\omega) H(\omega) S_{XX}(\omega)
= |H(\omega)|^{2} S_{XX}(\omega)}$ \\
& $S_{YY}(\omega)= H^{\ast}(\omega) H(\omega) S_{XX}(\omega)
= |H(\omega)|^{2} S_{XX}(\omega)$ \\
\hline
\end{tabular}
\renewcommand{\arraystretch}{1} \\
\textcolor{red}{*} = Es ist viel einfacher die Autokorrelation aus der Spektralen Leistung
(Transformationspaar) - anstatt aus diesem höllischen Integral - auszurechnen. \\
Ein WSS-Prozess am Eingang erzeugt auch einen WSS-Prozess am Ausgang.

\skriptsubsection{Spezielle Zufallsprozesse}{172-7.6}
\skriptsubsubsection{Gauss Zufallsprozess}{172-7.6.A}
\skriptsubsubsection{Gauss'scher Zufallsprozess}{172-7.6.A}
Bein diesem Prozess ist die \textbf{Zufallsvariable} $X(t_{i})$ zu \textbf{jedem Zeitpunkt} $t_{i}$
\textbf{gaussverteilt}. Bsp.: thermisches Rauschen. \\
Zweidimensionaler Fall (gauss'sche Verbunddichte):
$f_{XX}(x_{1},x_{2};t_{1},t_{2}) = \frac{1}{2\pi \sigma_{x_{1}}\sigma_{x_{2}}} \cdot
$$f_{XX}(x_{1},x_{2};t_{1},t_{2}) = \frac{1}{2\pi \sigma_{x_{1}}\sigma_{x_{2}}} \cdot
e^{-\frac{(x_{1}-\mu_{x_{1}})^{2}}{2 \sigma^{2}_{x_{1}}}} \cdot
e^{-\frac{(x_{2}-\mu_{x_{2}})^{2}}{2 \sigma^{2}_{x_{2}}}}$ \\
Für $X(t_{1})$ und $X(t_{2})$ unkorreliert ($C_{XX}(t_{1},t_{2})=0$) gilt:
$f_{XX}(x_{1},x_{2}; t_{1},t_{2}) = f_{X}(x_{1};t_{1}) \cdot f_{X}(x_{2};t_{2}) $ \\ \\
e^{-\frac{(x_{2}-\mu_{x_{2}})^{2}}{2 \sigma^{2}_{x_{2}}}}$$ \\
Für unkorrelierte $X(t_{1})$ und $X(t_{2})$ ($C_{XX}(t_{1},t_{2})=0$) gilt:
$$f_{XX}(x_{1},x_{2}; t_{1},t_{2}) = f_{X}(x_{1};t_{1}) \cdot f_{X}(x_{2};t_{2}) $$
$\mu_{X}(t)$ und $R_{XX}(t_{1}, t_{2})$ charakterisisieren einen gauss'schen Zufallsprozess
vollst\"andig. \\
Ist ein gauss'scher Prozess WSS ist er zugleich auch SSS. Zudem ist wird ein gauss'scher
Expand Down

0 comments on commit 811bf0c

Please sign in to comment.