path: root/usth/ICT2.10/homework/solutions.tex

\documentclass[a4paper,12pt]{article}
\usepackage[english,vietnamese]{babel}
\usepackage{amsmath}
\usepackage{booktabs}
\usepackage{circuitikz}
\usepackage{enumerate}
\usepackage{lmodern}
\usepackage{mathtools}
\usepackage{pgfplots}
\usepackage{siunitx}
\usepackage{textcomp}
\usepackage{tikz}
\usetikzlibrary{arrows,automata}

\newcommand{\N}{\mathcal N}
\newcommand{\ud}{\,\mathrm{d}}
\newcommand{\SIR}{\mathrm{SIR}}
\newcommand{\baud}{\mathrm{Bd}}
\newcommand{\bit}{\mathrm{b}}
\newcommand{\chip}{\mathrm{c}}
\newcommand{\problem}[1]{\noindent\textbf{#1.}}

\title{Mobile Wireless Communication}
\author{Nguyễn Gia Phong}
\date{Spring 2020}

\begin{document}
\maketitle
\setcounter{section}{1}
\section{Characteristics of Radio Environment}
\subsection{Propagation Models}
\problem 1  Consider radio waves propagating by two-slope model over the
distance under \SI{200}{\metre} in Orlando.  The average receive power is given by
\[\bar P_R = g(d) P_T G_T G_R\]

Assume the attenna gains are both 1 and apply the inverse variation of power
with distance for two-slope model we get
\begin{align*}
  &\bar P_R = d^{-n_1}\left(1 + \frac{d}{d_b}\right)^{-n_2} P_T\\
  \iff &P_T = d^{n_1}\left(1 + \frac{d}{d_b}\right)^{n_2}\bar P_R
\end{align*}

Substituting $d_b = \SI{90}{\metre}$, $n_1 = 1.3$ and $n_2 = 3.5$ gives us
\[P_T = d^{1.3}\left(1 + \frac{d}{90}\right)^{3.5}\bar P_R\]

With average power effect experienced, $P_{R,\si{\deci\bel}} = 10\lg P_R
= 10\lg \bar P_R$ and $P_{T,\si{\deci\bel}} = 10\lg P_T$ and thus
\begin{align*}
  &P_{T,\si{\deci\bel}} - P_{R,\si{\deci\bel}}
  = 13\lg d + 35\lg\frac{d + 90}{90}\\
  \iff &P_{R,\si{\deci\bel}} - P_{T,\si{\deci\bel}}
  = 35\lg\frac{90}{d + 90} - 13\lg d
\end{align*}
\pagebreak

This is plotted in the figure below
\begin{center}
  \begin{tikzpicture}
    \begin{axis}[
      xlabel={$d$ (\si{\metre})},
      ylabel={$P_{R,\si{\deci\bel}}-P_{T,\si{\deci\bel}}$ (\si{\deci\bel})}]
      \addplot[domain=0:200]{35*ln(90/(x+90))/ln(10) - 13*ln(x)/ln(10)};
    \end{axis}
  \end{tikzpicture}
\end{center}

\problem 2  Consider a log-normal shadow fading propagation, the receive power
is given by \[P_R = \sqrt[10]{10^X} g(d) P_T G_T G_R\] where $X$ is a zero-mean
normal random variable with STD $\sigma = \SI{6}{\deci\bel}$.
\begin{enumerate}[(a)]
  \item Given $\bar P_R = \SI{1}{\milli\watt}$ at $d = \SI{100}{\metre}$.
    \[P_R > \bar P_R \iff \sqrt[10]{10^X} > 1 \iff X > 0\]
    Since $X$ is zero-mean and normally distributed, the probability
    the received power at a mobile at that distance from the base station
    will exceed \SI{1}{\milli\watt} is \SI{50}{\percent}, and so is the
    probability it is less than \SI{1}{\milli\watt}.

  \item Let $Y = X/\sigma$, $Y \sim \N(0, 1)$
    and $F_X(x) = \Phi(X/\sigma) = \Phi(X/6)$.

    The probability a mobile has an acceptable received signal at
    \SI{10}{\milli\watt} or higher is
    \begin{align*}
      P\left(P_R \ge 10\bar P_R\right)
      &= P\left(\sqrt[10]{10^X} \ge 10\right) = P(X \ge 10)\\
      &= 1 - F_X(10) = 1 - \Phi\left(\frac{10}{6}\right) = \SI{4.78}{\percent}
    \end{align*}

  \item For $\sigma = \SI{10}{\deci\bel}$, $F_X(x) = \Phi(X/10)$.
    The probability a mobile has an acceptable received signal at
    \SI{10}{\milli\watt} or higher is
    \[P\left(P_R \ge 10\bar P_R\right) = 1 - F_X(10)
    = 1 - \Phi(1) = \SI{15.87}{\percent}\]

  \item If the lower limit for an acceptable received signal is
    \SI{6}{\milli\watt}, with $\sigma = 6$, the probability a received signal
    is acceptable is
    \begin{align*}
      P\left(P_R \ge 6\bar P_R\right)
      &= P\left(\sqrt[10]{10^X} \ge 6\right)
       = P\left(X \ge \lg{6^{10}}\right)\\
      &= 1 - F_X\left(\lg{6^{10}}\right)
       = 1 - \Phi\left(\frac{\lg{6^{10}}}{6}\right) = \SI{9.73}{\percent}
    \end{align*}
    With $\sigma = 10$, the probability a received signal is acceptable is
    \[P\left(P_R \ge 6\bar P_R\right) = 1 - F_X\left(\lg{6^{10}}\right)
      = 1 - \Phi\left(\frac{\lg{6^{10}}}{10}\right) = \SI{21.82}{\percent}\]
\end{enumerate}

\subsection{Random Channel Characterization}
Given $x(t) = e^t * (\Pi(t - 1) - \Pi(t - 3))$ and $h(t) = \delta(t - 1)$,
where $\Pi$ is the rectangular function:
\[\Pi(t) = \begin{dcases}
  0, &\text{if }|t| > \frac{1}{2}\\
  \frac{1}{2}, &\text{if }|t| = \frac{1}{2}\\
  1, &\text{if }|t| < \frac{1}{2}\\
\end{dcases}\]

The convolution sum of $x$ and $h$ is
\begin{align*}
  y(t) &= x(t - 1)\\
  &= \int_{-\infty}^\infty e^{t-z-1}(\Pi(z-2) - \Pi(z-4))\ud z\\
  &= \int_{-\infty}^\infty e^{t-z-1}\Pi(z-2)\ud z
   - \int_{-\infty}^\infty e^{t-z-1}\Pi(z-4)\ud z\\
  &= \int_{1.5}^{2.5} e^{t-z-1}\ud z
   - \int_{3.5}^{4.5} e^{t-z-1}\ud z\\
  &= e^{t-1}\left(e^{2.5} - e^{1.5} - e^{4.5} + e^{3.5}\right)
\end{align*}

\subsection{Fading}
\problem 1  Consider several delay spreeds $D$ of \SI{0.5}{\micro\second},
\SI{1}{\micro\second} and \SI{6}{\micro\second}.

\begin{itemize}
  \item For IS-95 and cdma2000 which uses the transmission bandwidth of
    \SI{1.25}{\mega\hertz}, their symbol interval is \SI{0.8}{\micro\second}.
    For the multipath rays to be resolvable, the delay spread must be
    greater than this (\SI{1}{\micro\second} and \SI{6}{\micro\second}).
  \item For WCDMA which uses the bandwidth of \SI{5}{\mega\hertz},
    the symbol interval is \SI{0.2}{\micro\second}, thus symbols
    are resolvable in all cases.
\end{itemize}

\problem 2  Indicate the condition for flat fading for each of the following
data rates with transmission in binary form: \SI{8}{kbps}, \SI{40}{kbps},
\SI{100}{kbps}, \SI{6}{Mbps}.

Assume information is transmitted in rectangular waves, the symbol interval
are \SI{125}{\micro\second}, \SI{25}{\micro\second}, \SI{10}{\micro\second}
and \SI{1/6}{\micro\second} respectively.  For flat fading to occur,
the delay spread must be significantly less than the symbol interval.
Since no data is provided or found, no conclusion is drawn on which
radio environments would result in flat fading for each of these data rates.

\section{Cellular Concept}
\subsection{Channel Allocation}
\problem 1  Assume the simplest path-loss model of $g(d) = d^{-3}$, calculate
down-link SIR at point P at the corner of a hexagonal cell in a 3-reuse case.

Using to path-loss model, the signal-to-interference ratio
can be approximated from the six first-tier interferers as follows
\[\SIR \approx \frac{1}{\left(\frac{R}{D-R}\right)^3
                        + \left(\frac{R}{D+R}\right)^3
                        + 4\left(\frac{R}{D}\right)^3}\]

In a 3-reuse case, $D = \sqrt{3C}R = 3R$, and thus
\[\SIR \approx \frac{1}{\left(\frac{R}{2R}\right)^3
                        + \left(\frac{R}{4R}\right)^3
                        + 4\left(\frac{R}{3R}\right)^3} = \frac{1728}{499}\]

\problem 2  Calculate the worst-case uplink SIR assuming the co-channel
interference is caused only by the closest interfering mobiles in radio cells
a distance $D = 3.46R$ away from the cell.  Assume the simplest path-loss model
of $g(d) = d^{-4}$, the signal-to-interference ratio is approximated by
\[\SIR \approx \frac{P_t/R^4}{6P_t/\left(\frac{3D}{4}\right)^4}
= \frac{(3D/4)^4}{6R^4}\]

With $D = 3.46R$ (4-reuse), this becomes
\[\SIR \approx \frac{(3\cdot3.46/4)^4}{6} = 7.56\]

\subsection{Erlang-B Formula and Sizing a Cell}
\problem 1  An user who makes a call attempt every 15 minutes, with each call
lasts an average of 2 minutes, generate the load of 2/15 erlangs.

\problem 2  Consider a mobile system supporting 832 frequency channels
and 7-reuse, there are over 118 channels per cell.  With the probility of
call blocking of $P_B \le \SI{1}{\percent}$, the traffic is around 101 erlangs.
Given the average call-holding time $h = \SI{200}{\second}$, the arrival rate
can be calculated to be $\lambda = \SI{0.505}{\mathrm{calls}\per\second}$.
Since an user makes a call every \SI{900}{\second} on average, there are
approximately 454.5 users.  As the density of mobile terminals is
\SI{2}{\mathrm{terminals}\per\square{\kilo\metre}}, the area is
\SI{227.25}{\square{\kilo\metre}}, which indicates a cell radius
of $R = \SI{9.35}{\kilo\metre}$, assuming a hexagonal topology.

\section{Modulation Techniques}
\problem 1  Consider communication system operating at the transmission
bandwidth of \SI{1}{\mega\hertz} with the rolloff factor of 0.25.
\begin{itemize}
  \item  Achievable data traffic rate is
    \[R_s = \frac{B}{1 + \beta} = \frac{10^6}{1 + 0.25}
    = \SI{800}{\kilo\baud\per\second}\]
  \item  Delay spread that no ISI occurs is much less than the symbol interval,
    which is $T = B^{-1} = \SI{1}{\micro\second}$.
  \item  Using OFDM with $N = 16$ equally spead carriers, for each subcarrier,
    $\Delta f = \SI{62.5}{\kilo\hertz}$, $R_s = \SI{50}{\kilo\baud\per\second}$
    and $T = \SI{16}{\micro\second}$.
  \item  Additionally use 16-QAM, the bit rate is
    $R_\bit = \SI{800}{\kilo\bit\per\second}$.
\end{itemize}

\problem 2  Given $B = \SI{1}{\mega\hertz}$, $\beta = 0.25$,
$R_\bit = \SI{4.8}{\mega\bit\per\second}$ and $T = \SI{25}{\micro\second}$.

$R_s = B/(1+\beta) = \SI{0.8}{\mega\baud\per\second}$, thus 64-QAM is used.

For OFDM, $N = R_s/\Delta f = R_\bit T \approx 128$.

\problem 3  Consider a transmission of bandwidth $B = \SI{2}{\mega\hertz}$,
where phase-shift keying and Nyquist rolloff shaping is used.

For rolloff factors of 0.2, 0.25, 0.5, the traffic rates are respectively
\SI{1.67}{\mega\baud\per\second}, \SI{1.6}{\mega\baud\per\second} and
\SI{1.33}{\mega\baud\per\second}.

In order to transmit at a rate of $R_\bit = \SI{6.4}{\mega\bit\per\second}$
when $\beta = 0.25$, 16-QAM should be used.

\problem 4  Given the input sequence 1001111010
and the following QPSK signal pairs
\begin{center}
  \begin{tabular}{c c c}
    \toprule
    Successive Signal & $a_i$ & $b_i$\\
    \midrule
    0 0 & $-1$ & $-1$\\
    0 1 & $-1$ & $+1$\\
    1 0 & $+1$ & $-1$\\
    1 1 & $+1$ & $+1$\\
    \bottomrule
  \end{tabular}
\end{center}

Let the carrier frequency be some multiple of 1/T

\begin{tikzpicture}
  \begin{axis}[scale only axis, width=0.8\textwidth, height=0.16\textwidth,
               xlabel=In-phase Carrier, xtick={0,1,2,3,4,5},
               xticklabels={0,T,2T,3T,4T,5T}, ymin=-2, ymax=2, samples=420]
    \addplot[domain=0:5]{cos(x*720)};
  \end{axis}
\end{tikzpicture}

\begin{tikzpicture}
  \begin{axis}[scale only axis, width=0.8\textwidth, height=0.16\textwidth,
               xlabel=Quadrature-phase Carrier, xtick={0,1,2,3,4,5},
               xticklabels={0,T,2T,3T,4T,5T}, ymin=-2, ymax=2, samples=420]
    \addplot[domain=0:5]{sin(x*720)};
  \end{axis}
\end{tikzpicture}

The output QPSK signal would then be

\begin{tikzpicture}
  \begin{axis}[scale only axis, width=0.8\textwidth, height=0.16\textwidth,
               xlabel=In-phase Component, xtick={0,1,2,3,4,5},
               xticklabels={0,T,2T,3T,4T,5T}, ymin=-2, ymax=2, samples=69]
    \addplot[domain=0:1]{+cos(x*720)};
    \addplot[domain=1:2]{-cos(x*720)};
    \addplot[domain=2:3]{+cos(x*720)};
    \addplot[domain=3:4]{+cos(x*720)};
    \addplot[domain=4:5]{+cos(x*720)};
  \end{axis}
\end{tikzpicture}

\begin{tikzpicture}
  \begin{axis}[scale only axis, width=0.8\textwidth, height=0.16\textwidth,
               xlabel=Quadrature-phase Component, xtick={0,1,2,3,4,5},
               xticklabels={0,T,2T,3T,4T,5T}, ymin=-2, ymax=2, samples=69]
    \addplot[domain=0:1]{-sin(x*720)};
    \addplot[domain=1:2]{+sin(x*720)};
    \addplot[domain=2:3]{+sin(x*720)};
    \addplot[domain=3:4]{-sin(x*720)};
    \addplot[domain=4:5]{-sin(x*720)};
  \end{axis}
\end{tikzpicture}

\begin{tikzpicture}
  \begin{axis}[scale only axis, width=0.8\textwidth, height=0.16\textwidth,
               xlabel=Output Signal, xtick={0,1,2,3,4,5},
               xticklabels={0,T,2T,3T,4T,5T}, ymin=-2, ymax=2, samples=69]
    \addplot[domain=0:1]{+cos(x*720)-sin(x*720)};
    \addplot[domain=1:2]{-cos(x*720)+sin(x*720)};
    \addplot[domain=2:3]{+cos(x*720)+sin(x*720)};
    \addplot[domain=3:4]{+cos(x*720)-sin(x*720)};
    \addplot[domain=4:5]{+cos(x*720)-sin(x*720)};
  \end{axis}
\end{tikzpicture}

\section{Multiple Access Techniques}
\subsection{Time-Division Multiple Access}
Transmission bit rate is the rate at which the bits are transmitted, while
the user information bit rate is the rate at which per data are transmitted.

In particular, GSM gives each time slot \SI{576.92}{\micro\second},
minus \SI{30.46}{\micro\second} guard time.  During this duration,
\SI{148}{\bit} are tramsmitted, thus the transmission bit rate is
$\SI{148}{\bit}/(\SI{576.92}{\micro\second}-\SI{30.46}{\micro\second})
= \SI{270.834}{\kilo\bit\per\second}$.  Of the \SI{148}{\bit},
\SI{114}{\bit} are data bits.  Furthermore, only one slot per GSM eight-slot
frame and 24 out of 26 frames are used to carry information.  Therefore,
the user bit rate is $\SI{114}{\bit}/\SI{4.615}{\milli\second}\cdot 24/26
= \SI{22.8}{\kilo\bit\per\second}$.

Similarly, IS-136 has the transmission bit rate of $\SI{1944}{\bit}
/ \SI{40}{\milli\second} = \SI{48.6}{\kilo\bit\per\second}$ and $\SI{520}{\bit}
/ \SI{40}{\milli\second} = \SI{13}{\kilo\bit\per\second}$.

\subsection{Code-Division Multiple Access}
Consider IS-95 with the bit rate of \SI{9.6}{\kilo\bit\per\second}
and the chip rate of \SI{1.2288}{\mega\chip\per\second}, the speading gain
is 128 chips per bit.

\section{Channel Coding Techniques}

\subsection{Block Coding}
Consider the generator matrix
\[\mathbf G = [\mathbf I_k \mathbf P] = \begin{pmatrix}
  1&0&0&0&1&0&1\\
  0&1&0&0&1&1&1\\
  0&0&1&0&1&1&0\\
  0&0&0&1&0&1&1
\end{pmatrix}\]
it is trivial that $n = 7$, $k = 4$ and
\[\mathbf P = \begin{pmatrix}
  1&0&1\\
  1&1&1\\
  1&1&0\\
  0&1&1
\end{pmatrix}\]

The parity check matrix is then given by
\[\mathbf H = [\mathbf P^T \mathbf I_{n-k}] = \begin{pmatrix}
  1&0&0&1&1&1&0\\
  0&1&0&0&1&1&1\\
  0&0&1&1&1&0&1
\end{pmatrix}\]

\subsection{Convolutional Coding}
Consider a $K = 3$, rate \textonehalf{} convolution encoder with generators
$g_1 = [101]$ and $g_2 = [011]$.
\begin{center}
  \begin{circuitikz}
    \draw (0,3) node (input) {input}
          (1,3) node[inputarrow] (in) {}
          (2,3) node[circ] (m1) {}
          (3.5,3) node[twoportshape] (port1) {}
          (5,3) node[circ] (m2) {}
          (6.5,3) node[twoportshape] (port2) {}
          (8,3) node[circ] (m3) {}
          (input) -- (in) -- (m1) -- (port1) -- (m2) -- (port2) -- (m3)

          (5,5.5) node[xor port, rotate=90] (xor1) {}
          (9,6) node[flowarrow] (out1) {}
          (10,6) node{$n_1$}
          (m1) |- (xor1.in 1)
          (m3) |- (xor1.in 2)
          (xor1.out) |- (out1)

          (6.5,0.5) node[xor port, rotate=270] (xor2) {}
          (9,0) node[flowarrow] (out2) {}
          (10,0) node{$n_2$}
          (m3) |- (xor2.in 1)
          (m2) |- (xor2.in 2)
          (xor2.out) |- (out2);
  \end{circuitikz}
\end{center}

Initialize the encoder with 01, we get the following state diagram
\begin{center}
  \begin{tikzpicture}[->,>=latex,shorten >=1pt,auto,node distance=42mm]
    \node[initial,state] (01) {01};
    \node[state] (00) [above right of=01] {00};
    \node[state] (11) [below right of=01] {11};
    \node[state] (10) [below right of=00] {10};

    \path (00) edge [dashed, loop above] node {00} (00)
               edge node {10} (10)
          (01) edge [dashed] node {11} (00)
               edge [bend left] node {01} (10)
          (10) edge [dashed, bend left] node {01} (01)
               edge node {11} (11)
          (11) edge [dashed] node {10} (01)
               edge [loop below] node {00} (11);

    \node [below of=10] {%
      \begin{tabular}{c c}
        \raisebox{2pt}{\tikz{\draw[dashed] (0,0) -- (10mm,0);}} & 0\\
        \raisebox{2pt}{\tikz{\draw (0,0) -- (10mm,0);}} & 1
      \end{tabular}};
  \end{tikzpicture}
\end{center}
Given the input bit sequence of 10011011, the output would be
0101111011100111.
\end{document}