On the total $(k,r)$-domination number of random graphs
Extracted source text ยท arXiv 1511.07249
\documentclass[
%%% one of
submission
%final
%proceedings
%%% if you compile a final version for the old OJS platform
% , ojs
%%% if all authors have the same affiliation
% , nomarks
]{dmtcs-episciences}
% DON'T LOAD ANY STYLES THAT CHANGE THE PAGE LAYOUT
% AND DON'T CHANGE THE PAGE LAYOUT BY HAND, EITHER.
\usepackage[utf8]{inputenc}
\usepackage{subfigure}
\usepackage{lineno,hyperref}
\usepackage{verbatim}
\usepackage{amssymb,amsmath,amsthm}
\usepackage{epsfig}
\usepackage{url}
% We strongly recommend to use natbib. Your colleagues deserve to be
% named in your text. PLEASE, ADAPT YOUR TEXT ACCORDINGLY, such that
% citations are grammatically correct.
\usepackage[square,numbers]{natbib}
\author{Louisa Harutyunyan\affiliationmark{1}
\thanks{Research supported by FQRNT (Le Fonds de Recherche du Qu\'{e}bec
- Nature et Technologies) Doctoral Scholarship.}}
\title{On the total $(k,r)$-domination number of random graphs}
% put your affiliation here, not your full address.
% If you like to give away your email or other parts of your address,
% THIS IS NOT THE RIGHT PLACE, your address will change, this paper
% will not.
% Just watch that your personal data that you want to communicate on
% the episcience server is always up to date.
\affiliation{
% one line per affiliation, no postal codes, grant numbers or similar
Sorbonne Universit\'{e}s, UPMC Univ Paris 06, CNRS, LIP6 UMR 7606,
4 place Jussieu 75005 Paris}
\keywords{random graphs, total $(k,r)$-domination}
% don't try to cheat here, we will check the dates!
%\received{2015-11-23}
%\revised{2002-07-19, 2014-02-05, 2015-09-09}
%\accepted{2015-09-09}
\begin{document}
\publicationdetails{VOL}{2015}{ISS}{NUM}{SUBM}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Author's definitions
%
\newcommand{\DEF}[1]{{\em #1\/}}
%\newcommand{\eopf}{\framebox[3mm]}
%\newenvironment{proof}%
%{\noindent{\bf Proof.}\ }%
%{\hfill\eopf\par\bigskip}%
\newcommand{\RR}{\ensuremath{\mathbb R}}
\newcommand{\C}{{\cal C}}
\newcommand{\PP}{\mathbb P}
\newcommand{\EE}{\mathbb E}
\newcommand{\tDelta}{\tilde{\Delta}}
\def\R {{\cal R}}
\def\NN {{\mathbb N}}
\newtheorem{theorem}{Theorem}
\newtheorem{lemma}{Lemma}
\newtheorem{definition}{Definition}
\newtheorem{corollary}{Corollary}
\newtheorem{conjecture}{Conjecture}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\maketitle
\begin{abstract}
A subset $S$ of a vertex set of a graph $G$ is a
\emph{total $(k,r)$-dominating set} if every vertex
$u \in V(G)$ is within distance $k$ of at least $r$ vertices in $S$.
The minimum cardinality among all total $(k,r)$-dominating sets of $G$ is
called the \emph{total $(k,r)$-domination number} of $G$, denoted by
$\gamma^{t}_{(k,r)}(G)$.
We previously gave an upper bound on $\gamma^{t}_{(2,r)}(G(n,p))$ in
random graphs with non-fixed $p \in (0,1)$.
In this paper we generalize this result to give an upper bound on
$\gamma^{t}_{(k,r)}(G(n,p))$ in random graphs with non-fixed $p \in (0,1)$
for $k\geq 3$ as well as present an upper bound on $\gamma^{t}_{(k,r)}(G)$
in graphs with large girth.
\end{abstract}
\input{intro}
\input{rw}
\input{general_graphs}
\input{random_graphs}
\input{rg_distk}
%\section*{References}
%\bibliography{biblo}
\begin{comment}
\begin{figure}[htbp]
\begin{center}
\subfigure[height of 1 cm\label{logo1cm}]{\includegraphics[height=1cm]{dmt}}
\hfil
\subfigure[width of 3 cm\label{logo3cm}]{\includegraphics[width=3cm]{dmt}}
\caption{The scaled logo of \protect\href{http://www.dmtcs.org/}{DMTCS}.}
\label{fig:logo2}
\end{center}
\end{figure}
\end{comment}
%\nocite{*}
\bibliographystyle{abbrvnat}
% use the following instead if you encounter problems
%\bibliographystyle{alpha}
\bibliography{biblo}
\end{document}
\section{Introduction}\label{intro}
\begin{comment}
Distance and multiple domination have been used to address
problems in networks, such as area monitoring, fault
tolerance in wireless sensor networks, where sensors
cooperatively collect data to monitor physical or
environmental conditions.
Such networks may contain \emph{sink} nodes, i.e. data collection
points.
Due to constraints of sensors, such as processing ability, storage
capacity and energy for communication, there may be node and/or
link failures in the network.
WSNs have no pre-configured network infrastructure
or centralized control and due to generally large number of
sensors or unreachable terrain, arranging sensors manually is
unrealistic.
Thus, for positive integers $k$ and $r$, a total $(k,r)$-dominating
set is a natural candidate for finding locations of
sinks in WSNs and to address node and/or link failures, where
robustness for dominators is achieved by choosing a value for
$r>1$ and the distance parameter $k$ allows increasing local
availability by reducing the distance to the dominators.
\end{comment}
\begin{comment}
A total $(k,r)$-dominating set achieves robustness for dominators by
choosing a value for $r>1$ and the distance parameter $k$ allows
increasing local availability by reducing the distance to the
dominators.
\end{comment}
\begin{comment}
Due to generally WSNs being constructed in unreachable terrain, sensors
may be arranged stochastically.
Because of this uncertainty and randomness in the network structure,
WSNs can be modelled as random graphs.
A random graph $G(n,p)$ consists of $n$ vertices with each of the
potential ${n \choose 2}$ edges being inserted independently with
probability $p$.
\end{comment}
In this paper we derive upper bounds on the total $(k,r)$-domination number
in graphs with large girth as well as in \emph{random graphs}.
A \emph{random graph} $G(n,p)$ consists of $n$ vertices with each of the
potential ${n\choose 2}$ edges being inserted independently with
probability $p$.
Random graphs can be used to model wireless sensor networks (WSNs), where
sensors cooperatively collect data to monitor physical or environmental
conditions.
Generally WSNs are constructed in unreachable terrain and sensors may be
arranged stochastically, which introduces uncertainty and randomness
in the network structure.
Distance and multiple domination have been used in the literature to address
problems in wireless networks, such as area monitoring, fault tolerance
in wireless sensor networks (WSNs).
Thus, for positive integers $k$ and $r$, a total $(k,r)$-dominating set
in random graphs is a natural candidate to address area monitoring and
fault tolerance in WSNs, where robustness for dominators is achieved
by choosing a value for $r>1$ and the distance parameter $k$ allows
increasing local availability by reducing the distance to the
dominators \cite{DW06,LWAB12,SG07,ZLL09}.
The rest of the paper is organized as follows.
In Section \ref{rw} we present a literature survey regarding upper bounds
on the domination number and its variants.
Section \ref{gg} derives an upper bound on $\gamma^{t}_{(k,r)}(G)$ in graphs of
large girth, and Section \ref{rg} derives an upper bound on
$\gamma^{t}_{(k,r)}(G(n,p))$ for $k\geq 3$ and non-fixed $p \in (0,1)$ in
random graphs.
\section{Related Work}\label{rw}
Distance and multiple domination have been studied extensively by several
authors \cite{CR90,CY00,CGS85,HMV07,L04,LTX09,MM75,SSE02,TX07,TX09}.
\begin{comment}
A set $D \subseteq V(G)$ is a \emph{distance $k$-dominating set} of $G$ if
for all vertices $u \in V \setminus D$ is within distance $k$ of a
vertex $v \in D$.
The minimum cardinality of a distance $k$-dominating set in $G$ is the
\emph{distance $k$-domination number} of $G$, denoted $\gamma_k(G)$.
\end{comment}
\begin{comment}
Meir and Moon gave an upper bound on $\gamma_k (G)$.
\begin{theorem}(\cite{MM75})
For any connected graph $G$ of order $n$ with $n\geq k+1$,
$\gamma_k (G) \leq \left\lfloor\frac{n}{k+1}\right\rfloor$.
\end{theorem}
%for any connected graph of order $n$ with $n \geq k+1$ \cite{MM75}.
\noindent Sridharan et al. considered the special case of $k=2$, independently,
and obtained $\displaystyle\gamma_2 (G) \leq \left\lfloor\frac{n}{3}\right\rfloor$
for $n \geq 3$ \cite{SSE02}.
Tian and Xu show that for a connected graph $G$
$\displaystyle\gamma_{k}(G) \leq n \frac{\ln[m(\delta + 1) + 2 - t]}
{m(\delta+1)+2-t}$, where $m = \left\lceil\frac{k}{3}\right\rceil$ and
$t = 3\left\lceil\frac{k}{3}\right\rceil - k$ \cite{TX09}.
\end{comment}
\begin{comment}
Liu et al. in \cite{LTX09}, give an upper bound on the $2$-connected distance-$k$
domination number, denoted $\gamma_{k}^2 (G)$.
%That is, $\displaystyle\gamma^{2}_{k}(G) \leq (1+o_{\delta}(1))n \frac{\ln[m(\delta + 1) + 1 - t]}
% {m(\delta+1)+1-t}$, where $m = \left\lceil\frac{k}{3}\right\rceil$ and
% $t = 3\left\lceil\frac{k}{3}\right\rceil - k$.
\begin{theorem}(\cite{LTX09})
Let $G$ be a $2$-connected graph with order $n$ and minimum degree $\delta$.
Then $\gamma^{2}_{k}(G) \leq (1+o_{\delta}(1))n \frac{\ln[m(\delta + 1) + 1 - t]}
{m(\delta+1)+1-t}$, where $m = \left\lceil\frac{k}{3}\right\rceil$,
$t = 3\left\lceil\frac{k}{3}\right\rceil - k$ and $o_{\delta}(1)$ denotes
a function in $\delta$ that tends to $0$ as $\delta$ tends to $\infty$.
\end{theorem}
\end{comment}
In \cite{GZ08,RV07} upper bounds are given on the $r$-tuple domination number.
A set $D \subseteq V(G)$ is a \emph{$r$-tuple dominating set} of $G$ if
for every vertex $v \in V(G)$, $|N[v] \cap D| \geq r$, where
$N[v] = \{u \in V(G)|(u,v) \in E(G)\} \cup \{v\}$.
The minimum cardinality of a $r$-tuple dominating set of $G$ is the
\emph{$r$-tuple domination number} of $G$, denoted $\gamma_{\times r}(G)$.
%\cite{CGS85,CR90,CY00,RV07}.
\begin{comment}
The first upper bound for $r$-domination number is due to Cockayne et al.
in \cite{CGS85}, where they prove that with $\delta \geq r$,
$\gamma_{(r)} (G) \leq \frac{r}{r+1}n$.
This result has been further improved in \cite{CR90,CY00}.
Rautenbach and Volkmann extended this result for smaller values of $\delta$
in \cite{RV07}.
\begin{theorem}(\cite{RV07})
In a graph $G$ of order $n$, where $\delta \geq 2r \ln (\delta + 1) -1$,
then $\gamma_{(r)}(G) \leq \frac{n}{\delta + 1}
\left(r \ln (\delta + 1) + \sum_{i=1}^{r-1} \frac{1}{i! (\delta + 1)^{r-1-i}}\right)$
and $\gamma_{\times r}(G) \leq \frac{n}{\delta + 1}
\left(r \ln (\delta + 1) + \sum_{i=1}^{r-1} \frac{r-i}{i! (\delta + 1)^{r-1-i}}\right)$.
\end{theorem}
\end{comment}
%\noindent
%In \cite{GZ08} Gagarin and Zverovich presented a generalized upper
% bound for the $r$-tuple domination number.
Chang \cite{C08} further improved these results for any positive integer $r$
and for any graph of $n$ vertices with minimum degree $\delta$, where
$\displaystyle\gamma_{\times r}(G) \leq
\frac{\ln(\delta - r + 2) + \ln \tilde{d}_{r-1} + 1}{\delta - r + 2}n$,
and $\displaystyle\tilde{d}_m = \frac{1}{n} \sum^{n}_{i=1}
{d_i + 1 \choose m}$ with $d_i$ being the degree of the $i$th vertex of $G$.
%In \cite{CDKRWZ12}, Chang et al. give an upper bound for $\gamma_{\times r}$
% for $r$-regular graphs.
%They also give a sharp bound for $\gamma_{\times 2}$ for claw-free $r$-regular
% graphs.
In \cite{CY00} Caro and Yuster give upper bounds on the
$r$-tuple and total $r$-domination numbers.
A set $D \subseteq V(G)$ is a \emph{total $r$-dominating set} of $G$ if
for every vertex $v \in V(G)$, $|N(v) \cap D| \geq r$, where
$N(v) = \{u \in V(G)|(u,v) \in E(G)\}$.
The minimum cardinality of a total $r$-dominating set of $G$ is the
\emph{total $r$-domination number} of $G$, denoted $\gamma^{t}_{\times r}(G)$.
\begin{comment}
\begin{theorem}(\cite{CY00})
In a graph $G$ of order $n$ and minimum degree $\delta$, if $\delta > e^{r^2}$
and $r \in \mathbb{N}$,
then $\gamma_{\times r}(G) \leq \gamma_{\times r}^t (G) <
\frac{\ln \delta + \sqrt{\ln \delta} + 2}{\delta} n$.
\end{theorem}
\end{comment}
%\noindent
In \cite{ZWX07} Zhao et al. study the total $r$-domination number in
graphs.%, where they give an upper bound for $\gamma^{t}_{\times r}$.
\begin{theorem}\cite{ZWX07}
In a graph $G$ of order $n$ and minimum degree $\delta \geq r$, where
$r \in \mathbb{N}$, if $\frac{\delta}{\ln\delta}\geq 2r$, then
$\gamma_{\times r}^t (G)\leq \frac{n}{\delta}
\left(r\ln\delta + \sum_{i=0}^{r-1}\frac{r-i}{i!\delta^{r-1-i}}\right)$.
\end{theorem}
%\vspace{5mm}
Some works in the literature study upper bounds on the
$(k,r)$-domination number.
A set $D \subseteq V(G)$ is a \emph{$(k,r)$-dominating set} of $G$ if for
every vertex $v \in V(G) \setminus D$ is within distance $k$ of $r$
vertices in $D$.
The minimum cardinality of a $(k,r)$-dominating set of $G$ is the
\emph{$(k,r)$-domination number} of $G$, denoted $\gamma_{k,r}(G)$.
In \cite{BHS94} Bean et al. posed the following conjecture.
\begin{conjecture}\cite{BHS94}
Let $G$ be a graph of order $n$ and let $\delta_k$ denote the smallest
cardinality among all $k$-neighbourhoods of $G$, where
$\delta_k \geq k+r-1$.
Then for positive integers $k$ and $r$
$\gamma_{(k,r)} (G) \leq \frac{r}{r+k} n$.
\end{conjecture}
\noindent Fischermann and Volkmann confirmed that the conjecture is valid
for all integers $k$ and $r$, where $r$ is a multiple of $k$ \cite{FV05}.
In \cite{KMV08} Korneffel et al. show that $\gamma_{2,2}(G) \leq \frac{n(G)+1}{2}$.
There are several works in the literature that study upper bounds on
the domination number and its variants in random graphs.
Recall that a random graph $G(n,p)$ consists of $n$ vertices with each
of the potential ${n \choose 2}$ edges being inserted independently
with probability $p$.
We say that an event holds \emph{asymptotically almost surely (a.a.s)} if
the probability that it holds tends to $1$ as $n$ tends to infinity.
Dreyer \cite{D00} in his dissertation studied the question of domination
in random graphs.
Wieland and Godbole proved that the domination number of a random graph,
denoted $\gamma (G(n,p))$, has a two point concentration \cite{WG01}.
\begin{theorem}\cite{WG01}\label{th1}
For $p \in (0,1)$ fixed, a.a.s $\,\,\gamma (G(n,p))$ equals
$\left\lfloor\mathbb{L}n - \mathbb{L}\left((\mathbb{L}n)(\log n)\right)\right\rfloor + 1$
or\\
$\left\lfloor\mathbb{L}n - \mathbb{L}\left((\mathbb{L}n)(\log n)\right)\right\rfloor + 2$,
where $\mathbb{L}n = \log_{1/(1-p)}n$.
\end{theorem}
\noindent Wang and Xiang \cite{WX09} extend this result for $2$-tuple domination number
of $G(n,p)$.
\begin{theorem}\cite{WX09}
For $p \in (0,1)$ fixed, a.a.s $\,\,\gamma_{\times 2} (G(n,p))$ equals
$\left\lfloor\mathbb{L}n - \mathbb{L}(\log n) +\mathbb{L}\left(\frac{p}{1-p}\right)\right\rfloor + 1$
or
$\left\lfloor\mathbb{L}n - \mathbb{L}(\log n) + \mathbb{L}\left(\frac{p}{1-p}\right)\right\rfloor + 2$,
where $\mathbb{L}n = \log_{1/(1-p)}n$.
\end{theorem}
Bonato and Wang \cite{BW08} study the total domination number and the \emph{independent
domination number} in random graphs.
For a graph $G$, a set $D\subseteq V(G)$ is an \emph{independent dominating
set} of $G$ if $D$ is both an independent set and a dominating set of $G$.
The \emph{independent domination number} of $G$, denoted $\gamma_i (G)$, is the
minimum order of an independent dominating set of $G$.
\begin{theorem}\cite{BW08}
For $p \in (0,1)$ fixed, a.a.s $\,\,\gamma_{t}(G(n,p))$ equals
$\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n))\right\rfloor+1$
or\\
$\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n)) \right\rfloor+2$,
where $\mathbb{L}n = \log_{1/(1-p)}n$.
\end{theorem}
\begin{theorem}\cite{BW08}
For $p \in (0,1)$ fixed, a.a.s
$\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n))\right\rfloor+1
\leq \gamma_{i}(G(n,p)) \leq \lfloor\mathbb{L}n\rfloor$,
where $\mathbb{L}n = \log_{1/(1-p)}n$.
\end{theorem}
\noindent Wang further studied the independent domination number of random graphs \cite{W10}.
\begin{theorem}\cite{W10}\label{th2}
Let $p \in (0,1)$ and $\epsilon \in \left(0, \frac{1}{2}\right)$ be two real
numbers.
Let $k = k(p,\epsilon)\geq 1$ be the smallest integer satisfying
$(1-p)^k < \frac{1}{2} - \epsilon$.
A.a.s. $\gamma(G(n,p))\leq \gamma_i (G(n,p)) \leq
\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n))\right\rfloor+k+1$,
where $\mathbb{L}n = \log_{1/(1-p)}n$.
\end{theorem}
\noindent If $p > \frac{1}{2}$, then for $\epsilon \in\left(0, p-\frac{1}{2}\right)
\subset \left(0, \frac{1}{2}\right)$, by Theorems \ref{th1} and \ref{th2}, the
following concentration result follows.
\begin{corollary}\cite{W10}
For $p \in \left(\frac{1}{2}, 1\right)$ fixed, a.a.s.
$\gamma(G(n,p))\leq \gamma_i (G(n,p)) \leq$
$\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n))\right\rfloor+2$,
where $\mathbb{L}n = \log_{1/(1-p)}n$.
\end{corollary}
\cite{H14} studies an upper bound on $\gamma_{(2,r)}^t (G(n,p))$ in
random graphs for non-fixed $p \in (0,1)$.
% $\gamma_{(2,r)}^t (G(n,p))$ and
% $\gamma_{(3,r)}^t (G(n,p))$ are given for $p$ non-fixed.
\begin{theorem}\cite{H14} \label{k2}
Let $c>1$ be a fixed constant.
Then for any positive integer $r$, in a random graph $G(n,p)$ with
$\displaystyle p \geq c\sqrt{\frac{\log n}{n}}$, \,\, a.a.s. \,
$\displaystyle \gamma^{t}_{(2, r)}\left(G(n, p)\right) = r+1$.
\end{theorem}
\begin{comment}
\begin{theorem}(\cite{H14}) \label{dist3}
Let $c>\sqrt[3]{2}$ be a fixed constant.
Then for any positive integer $r$, in a random graph $G(n,p)$ with
$\displaystyle p \geq c \,\sqrt[3]{\frac{\log n}{n^2}}$, \,\, a.a.s.
\, $\displaystyle \gamma^{t}_{(3, r)}\left(G(n, p)\right) = r+1$.
\end{theorem}
\end{comment}
To the best of our knowledge, there are no works in the literature that
study the upper bounds on the total $(k,r)$-domination number in general
graphs or in random graphs.
In this paper we give an upper bound on $\gamma_{(k,r)}^t (G)$ in graphs of large
girth and extend the results of \cite{H14} to derive an upper bound
on $\gamma_{(k,r)}^t (G(n,p))$ in random graphs.
\begin{comment}
To our knowledge there is no work in the literature that study upper bounds
of the distance-$k$ total $r$-domination number, $\gamma^{t}_{(k,r)}(G)$,
in random graphs.
Thus, in this thesis we derive upper bounds on $\gamma^{t}_{(2,r)}(G(n,p))$,
$\gamma^{t}_{(3,r)}(G(n,p))$ and $\gamma^{t}_{(k,r)}(G(n,p))$ in random graphs
as well as on $\gamma^{t}_{(k,r)}(G)$ in graphs with large girth.
\end{comment}
\begin{comment}
\cite{WG01} prove the domination number of $G(n,p)$
with $p \in (0,1)$ fixed, a.a.s. has a two point concentration, namely,
for $\mathbb{L}n = \log_{1/(1-p)}n$, a.a.s.
\begin{equation}
\left\lfloor\mathbb{L}n - \mathbb{L}\left((\mathbb{L}n)(\log n)\right)\right\rfloor + 1
\leq \gamma(G(n,p)) \leq
\left\lfloor\mathbb{L}n - \mathbb{L}\left((\mathbb{L}n)(\log n)\right)\right\rfloor + 2
\end{equation}
\end{comment}
\begin{comment}
Wang and Xiang \cite{WX09} extend this result for $2$-tuple domination number
of $G(n,p)$.
They prove that with $p \in (0,1)$ fixed and where
$\mathbb{L}n = \log_{1/(1-p)}n$, a.a.s.
\begin{equation}
\begin{split}
\gamma_{\times 2}(G(n,p)) \geq
\left\lfloor\mathbb{L}n - \mathbb{L}(\log n) +\mathbb{L}\left(\frac{p}{1-p}\right)\right\rfloor + 1 \\
\gamma_{\times 2}(G(n,p)) \leq
\left\lfloor\mathbb{L}n - \mathbb{L}(\log n) + \mathbb{L}\left(\frac{p}{1-p}\right)\right\rfloor + 2.
\end{split}
\end{equation}
\end{comment}
\begin{comment}
\begin{split}
\left\lfloor\mathbb{L}n - \mathbb{L}(\log n) +\mathbb{L}\left(\frac{p}{1-p}\right)\right\rfloor + 1
\leq \gamma_{\times 2}(G(n,p)) \leq \\
\left\lfloor\mathbb{L}n - \mathbb{L}(\log n) + \mathbb{L}\left(\frac{p}{1-p}\right)\right\rfloor + 2.
\end{split}
\end{comment}
\begin{comment}
Bonato and Wang \cite{BW08} study the total domination number in random graphs.
They prove that in $G(n,p)$ for $p \in (0,1)$ fixed and where
$\mathbb{L}n = \log_{1/(1-p)}n$, a.a.s.
\begin{equation}
\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n))\right\rfloor+1
\leq \gamma_t(G(n,p)) \leq
\left\lfloor\mathbb{L}n - \mathbb{L}((\mathbb{L}n)(\log n)) \right\rfloor+2.
\end{equation}
To our knowledge there is no work in the literature that study upper bounds
of the distance-$k$ total $r$-domination number, $\gamma^{t}_{(k,r)}(G)$,
in random graphs.
Thus, in this thesis we derive upper bounds on $\gamma^{t}_{(2,r)}(G(n,p))$,
$\gamma^{t}_{(3,r)}(G(n,p))$ and $\gamma^{t}_{(k,r)}(G(n,p))$ in random graphs
as well as on $\gamma^{t}_{(k,r)}(G)$ in graphs with large girth.
\end{comment}
\vspace{4mm}
\section{Total $(k,r)$-domination number in graphs of large girth}\label{gg}
In this section we derive an upper bound on the total $(k,r)$-domination
number in graphs with large girth.
We present our result in Theorem \ref{gen}.
Although our result is not tight, we do obtain a bound with relatively
simple expression.
\begin{theorem}\label{gen}
Consider a graph $G$, where $n=|V(G)|$.
Let $G$ be of minimum degree at least $d$, and of girth at least $2k+1$.
Then for any positive integers $k$ and $r$,
$\displaystyle\gamma^{t}_{(k,r)}(G) \leq \frac{2nr}{(d-1)^k} + nre^{-\frac{r}{4}}$.
\end{theorem}
\begin{proof}
Let us pick, randomly and independently, each vertex $v \in V(G)$
with probability $p$.
Let $S \subset V(G)$ be the set of vertices picked.
We will determine the value of $p$ by the end of the proof.
$S$ is a random set and is part of the total $(k,r)$-dominating set that
we would like to obtain.
Let the distance from a vertex $u$ to a vertex $v$ be denoted as $d(u,v)$,
which is the length of the shortest path between $u$ and $v$.
For every vertex $v \in V(G)$, let $X_v$ denote the number of vertices
in $N_{k}(v)$ that are also in $S$, where
$N_{k}(v) = \{u\in V|u \neq v \text{ and } d(u,v) \leq k\}$.
Let $Y$ be the set such that $Y = \{v \in V(G) | X_v \leq r-1 \}$.
Note that $S$ is a random set and $\EE\big[|S|\big]=np$.
We now estimate $\PP\big[X_v < r\big]$. %$\EE\left[|Y|\right]$.
\vspace{3mm}
For a given vertex $v \in Y$, let $m= |N_k(v)|$.
We will show by contradiction that $m \geq (d-1)^k$.
Assume that $m < (d-1)^k$.
Then there exist vertices $u_1, \, u_2 \in N_k(v)$ such that there is a
vertex $w \in N_k(v)$ and $w \in \left(N_k(u_1) \cap N_k(u_2)\right)$.
Vertex $w$ is at most distance $k$ from $v$.
Thus, the distance from $w$ to $v$ through the path containing $u_1$ is
at most $k$.
Similarly, the distance from $w$ to $v$ through the path containing $u_2$
is also at most $k$.
Thus, making a cycle of length at most $2k$, which is a contradiction.
Therefore, by the assumption that $G$ has girth at least $2k+1$, it follows
that $m \geq (d-1)^k$.
\vspace{3mm}
It can be seen that $X_v$ is a $B(m,p)$ random variable.
We use the well known Chernoff Bound \cite{Bern24,Cr33,Ch52,Ben62,Ho63,ASE} to bound
$\PP\big[X_v \leq r-1\big]= \PP\big[X_v < r\big]$.
% we use the Chernoff Bound \cite{ASE}.
\vspace{3mm}
\noindent The Chernoff Bound states: for any $a > 0$ and random variable $X$ that
has binomial distribution with probability $p$ and mean $pn$,
\begin{equation}\label{chernoff}
\PP\big[X - pn < -a\big] < e^{-a^2/2pn}.
\end{equation}
%\vspace{2mm}
\noindent We set $a = \epsilon pm$, where we let
$\displaystyle\epsilon = 1- \frac{r}{pm}$.
Hence, $a = pm - r$, which results in $r = pm - a$.
Then, by the Chernoff Bound given in Equation \ref{chernoff} we have,
\begin{align}\label{PXv}
\PP\big[X_v < r\big] &= \PP\big[X_v < pm-a\big]\nonumber\\
&< e^{-\frac{a^2}{2pm}}
= e^{-\frac{{\epsilon}^2 (pm)^2}{2 pm}}
= e^{-\frac{{\epsilon}^2 pm}{2}}\nonumber\\
&\leq e^{-\frac{{\epsilon}^2 p(d-1)^k}{2}}.
\end{align}
%\vspace{3mm}
\noindent Chernoff's bound holds whenever $\epsilon > 0$, that is when
$\displaystyle 1 - \frac{r}{pm} > 0$.
Thus, it holds when $\displaystyle p > \frac{r}{(d-1)^k}$.
By setting $\displaystyle p = \frac{2r}{(d-1)^k}$, from Equation \ref{PXv} we
obtain
\begin{align}
\PP\big[X_v < r \big] < e^{-{\epsilon}^2\left(\frac{2r}{(d-1)^k}\right)
\frac{(d-1)^k}{2}} = e^{-{\epsilon}^2 r}.\nonumber
\end{align}
%\vspace{3mm}
For each vertex $v \in Y$, where $\displaystyle X_v \leq r-1$, we pick
a set $A_v$ of $r$ vertices in $N_k(v)$ arbitrarily.
For vertices $v$ that satisfy $X_v \geq r$, $A_v = \emptyset$.
Let $\displaystyle A = \bigcup_{v=1}^{n} A_v$.
Clearly, $S \cup A$ is a total $(k,r)$-dominating set.
We now estimate $\EE\big[|A|\big]$.
%\vspace{3mm}
\noindent By linearity of expectation, we obtain
\begin{align}
\EE\big[|A|\big] &= \EE\left[\left|\bigcup_{v=1}^{n} A_v \right|\right]
\leq \EE\left[\sum_{v=1}^{n} \left|A_v\right|\right]\nonumber\\
&= \sum_{v=1}^{n} \EE\big[|A_v|\big]
\leq n r e^{-{\epsilon}^2 r}.\nonumber
\end{align}
%\vspace{2mm}
\noindent Again by the linearity of expectation, we now estimate
$\EE\big[|S \cup A|\big]$.
\begin{align}
\EE\big[|S \cup A|\big] &= \EE\big[|S|\big] + \EE\big[|A|\big]\nonumber\\
&\leq np + nre^{-{\epsilon}^2 r}\nonumber\\
&= \frac{2nr}{(d-1)^k} + nre^{- {\epsilon}^2 r}\nonumber.
\end{align}
\noindent Therefore, we have shown that there exists a total $(k,r)$-dominating
set in $G$, where %of size at most
\begin{align}
\gamma^{t}_{(k,r)}(G) &\leq \frac{2nr}{(d-1)^k} + nre^{-{\epsilon}^2 r}\nonumber\\
&\leq \frac{2nr}{(d-1)^k} + nre^{-\frac{r}{4}}\nonumber
\end{align}
since $\epsilon > 1/2$.
\end{proof}
%\vspace{5mm}
\vspace{4mm}
\section{Total $(k,r)$-domination number in random graphs}\label{rg}
In \cite{H14}, we presented an upper bound on the total
$(2,r)$-domination number of the random graphs.
In this section we generalize this result to derive an upper
bound on the total $(k,r)$-domination number in random graphs
for $k\geq 3$.
However, before doing so we briefly discuss the main difference
between the solutions of $\gamma_{(2,r)}^t (G(n,p))$ and
$\gamma_{(k,r)}^t (G(n,p))$, for $k\geq 3$.
In Theorem \ref{k2} it is proved that in a random graph $G(n,p)$
with $\displaystyle p \geq c \sqrt{\frac{\log n}{n}}$ and
a fixed constant $c > 1$, a.a.s. $\gamma_{(2,r)}^t (G(n,p)) = r+1$.
In the proof of Theorem \ref{k2} it is needed to calculate the
probability that a vertex $u$ is not within distance-$2$ from
a dominator vertex $v_i$, i.e. $\PP[v_i \notin N_2(u)]$.
To connect $u$ to $v_i$ via a path of length $2$, one connecting
vertex, denoted $w_i$, is needed.
To determine that $\PP[v_i \notin N_2(u)]$ uses the fact that the
edges between $u$, $w_i$ and $v$, $w_i$ that connect
$u$ to $v_i$ (in order to obtain a path of length $2$ and $u$ to
be dominated by $v_i$) cannot be chosen again to connect $u$ to
$v_i$ via a different path (since the two paths would be the
same).
%To determine $\PP[v_i \notin N_3(u)]$ uses the fact that the edge
% between two vertices $w_i$ and $w_{i+1}$ that connect $u$ to
% $v_i$ (in order to obtain a path of length $3$ and $u$ to be
% dominated by $v_i$) cannot be chosen again to $u$ and $v_i$
% via a different path (since the two paths would be the same).
Hence, the probability that there exists an edge between any two
connecting vertices among all different paths of length $2$ from
$u$ to $v_i$ are independent of each other.
So, the probability that there is a path from $u$ to $v_i$ via a
given connecting vertex $w_i$ is $p^2$.
There are $n-2$ such vertices and thus, the probability
that a vertex $u$ is not within distance-$2$ from a dominator
vertex $v_i$ is given by
$\displaystyle\PP[v_i \notin N_2(u)] \leq (1-p^2)^{(n-2)}$.
Similar calculation is needed in the proof of Theorem \ref{prop_distk}
that follows to determine the probability that a vertex $u$ is
not within distance-$k$ from a dominator vertex $v_i$ (Theorem
\ref{prop_distk}, Equation \ref{not_k-adj}).
However, once we generalize to give an upper bound on the total
$(k,r)$-domination number, we cannot easily obtain an independence
between the existing edges of any $k-1$ connecting vertices among
all the different paths of length $k$ from $u$ to $v_i$.
%However, once we generalize to give an upper bound for the total
% $(k,r)$-domination number, we cannot easily obtain this independence
% when considering paths of length greater than $3$ from $u$ to $v_i$.
When considering paths of length $k$ from $u$ to $v_i$ for the general
case of total $(k,r)$-domination number it becomes more difficult
to calculate the probability that there is a path of length $k$ from $u$ to
$v_i$ via $k-1$ vertices.
There may be two different paths $P_1$ and $P_2$ from $u$ to $v_i$ that
may share some edges between any of the connecting $k-1$ vertices
and hence, are not independent anymore as they were in the case of
total $(2,r)$-domination number (see Fig. \ref{fig_distk}).
\begin{figure}[h!]
\centering
\includegraphics[width=0.4\textwidth]{distk}
\caption{$P_1 = u\,a_1\,a_2\,a_3\cdots a_{k-2}\,a_{k-1}\,v_i$ and
$P_2 = u\,b_1\,a_2\,a_3\,b_2\cdots b_{k-2}\,b_{k-1}\,v_i$ are
two paths between $u$ and $v_i$ that share an edge, namely $(a_2, a_3)$. }
\label{fig_distk}
\end{figure}
%############################################################################################
%####################### Bollobas's theorem ##################################################
%#############################################################################################
Bollob\'{a}s has the following result on random graphs of diameter greater than two.
\begin{theorem}\cite{BB}\label{bb}
Let $c$ be a positive constant, $d = d(n) \geq 2$ a natural number, and
define $p = p(n,c,d)$, $0<p<1$, by
\[p^d n^{d-1} = \log(n^2/c).\]
Suppose that $pn/(\log n)^3 \to \infty$. Then in $G(n,p)$ we have
\[\lim_{n \to \infty}\mathbb{P}(diam\text{ }G = d) = e^{-c/2}
\quad\text{and}\quad
\lim_{n \to \infty}\mathbb{P}(diam\text{ }G = d+1) = 1 - e^{-c/2}.\]
\end{theorem}
%############################################################################################################
%############################################################################################################
%############################################################################################################
%############################################################################################################
%############################################################################################################
%############################################################################################################
Note that from Theorem \ref{bb}, the diameter of $G(n,p)$ is at most $d$
for $\displaystyle p=\sqrt[{d-1}]{\frac{\log (n^2/c)}{n^{d-2}}}$.
In Theorem \ref{prop_distk}, we weaken the value of $p$ to
$\displaystyle p' \geq d\sqrt[d]{\frac{\log n}{n^{d-1}}}$.
Easy calculation shows that $\displaystyle d\sqrt[d]{\frac{\log n}{n^{d-1}}} <
\sqrt[{d-1}]{\frac{\log (n^2/c)}{n^{d-2}}}$ for $d \leq (\log n)^\epsilon$
for a constant $\epsilon < \frac{1}{2}$.
In particular, $p' < p$ holds when $d$ is constant.
Our proof of Theorem \ref{prop_distk} uses Janson's Inequality, which
we present here first \cite{ASE}.
\begin{comment}
\begin{figure}[h!]
\centering
\includegraphics[width=0.4\textwidth]{Figures/distk}
\caption{$P_1 = u\,a_1\,a_2\,a_3\cdots a_{k-2}\,a_{k-1}\,v_i$ and
$P_2 = u\,b_1\,a_2\,a_3\,b_2\cdots b_{k-2}\,b_{k-1}\,v_i$ are
two paths between $u$ and $v_i$ that share an edge, namely $(a_2, a_3)$. }
\label{fig_distk}
\end{figure}
\end{comment}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\vspace{4mm}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%% setting for Janson's Inequality %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Let $\Omega$ be a finite universal set and let $R$ be a random subset
of $\Omega$ given by
\begin{equation}
\PP[r \in R] = p_r,
\end{equation}
these events are mutually independent over $r \in \Omega$.
Let $\{A_i\}_{i\in I}$ be subsets of $\Omega$, where $I$ is a finite index
set.
Let $B_i$ be the event that $A_i \subseteq R$.
Let $X_i$ be the indicator random variable for $B_i$ and $X = \sum \limits_{i \in I} X_i$
the number of $A_i \subseteq R$.
Hence, $\PP[X=0] = \PP\left[\bigcap \limits_{i\in I} \overline{B_i}\right]$.
For $i, j \in I$ we write $i \sim j$ if $i \neq j$ and $A_i \cap A_j \neq \emptyset$.
Thus, define $\Delta = \sum \limits_{i \sim j} \PP[B_i \cap B_j]$.
Set $\mu = \EE[X] = \sum \limits_{i \in I} \PP[B_i]$.
In Theorem \ref{Janson} we state Janson's Inequality. \cite{ASE}
\begin{theorem}\cite{ASE}\label{Janson}
Let $\{B_i\}_{i \in I}$, $\Delta$, $\mu$ be as above.
Then $\PP\left[\bigcap \limits_{i \in I} \overline{B_i}\right] \leq e^{-\mu + \Delta/2}$.
\end{theorem}
\vspace{3mm}
%############################################################################################################
%############################################################################################################
%############################################################################################################
%############################################################################################################
%############################################################################################################
%############################################################################################################
Now, we present our main result on the total $(k,r)$-domination number of the
random graphs.
\begin{theorem} \label{prop_distk}
For any positive integers $k\geq 3$ and $r$, in a random graph $G(n,p)$ with
$\displaystyle p \geq k \sqrt[k]{\frac{\log n}{n^{k-1}}}$, \,\, a.a.s.
\, $\displaystyle \gamma^{t}_{(k, r)}\left(G(n, p)\right) = r+1$.
\end{theorem}
\begin{proof}
Let $D \subseteq V(G(n,p))$ be a total $(k,r)$-dominating set and
let the vertices in $D$ be labelled as
$v_1,\, v_2,\,\cdots,\, v_i,\cdots,\, v_{r+1}$, where $1 \leq i \leq r+1$.
The probability that a vertex $u \in V(G(n,p))$ is not within
distance-$k$ from a vertex $v_i \in D$ is denoted by
$\mathbb{P}[v_i \notin N_k(u)]$.
Let $X$ be a random variable that denotes the number of vertices $u \in V(G(n,p))$,
where the number of $k$-adjacent vertices of $u$ in $D$ is less than $r$.
We would like to show that the number of vertices in $V(G(n,p))$ with less than
$r$ dominators tends to $0$.
That is, $\mathbb{P}[X>0] \to 0$ as $n \to \infty$.
We define a fixed vertex $u$ as \emph{bad}, if $u$ in its $k$-neighborhood
has less than $r$ dominators in $D$.
By linearity of expectation we have
\begin{equation}\label{EXk}
\EE[X] = n \cdot \mathbb{P}[\text{fixed $u$ is bad}].
\end{equation}
There are $n-2$ vertices aside from $u$ and $v_i$ to connect $u$ to $v_i$
via a path of length $k$.
To connect $u$ to $v_i$ such that $d(u, v_i) = k$, additional $k-1$
connecting vertices are necessary to create a path of length
$k$ from $u$ to $v_i$.
There are ${n-2 \choose k-1}$ possible ways to choose these $k-1$
vertices.
Hence, we have ${n-2 \choose k-1}$ such sets that consist of $k-1$
vertices.
We denote these sets by $S_1, \, S_2,\, \cdots , S_{n-2 \choose k-1}$.
We would like to show that $\PP[v_i \notin N_k(u)] \to 0$ as
$n \to \infty$.
This is equivalent to showing that the probability one of $S_i$ connects
$u$ to $v_i$ via a path of length $k$ tends to $1$ as $n \to \infty$.
Let $S_i = \{a_{i_1}, a_{i_2}, \cdots , a_{i_{k-1}}\}$.
For any pair $u$ and $v_i$ that are fixed, we number all other $n-2$
vertices and assume that all vertices in $S_i$ are connected in
ascending order of the vertex number.
Note that some edges in $S_i$ and $S_j$, where $i \neq j$ are the same.
To calculate the probability of the appearance of the $k-2$ edges in
each $S_i$ we must consider the dependencies between any two sets
$S_i$ and $S_j$ for $i \neq j$.
To do this, we use Janson's inequality from Theorem \ref{Janson}.
%#################################################################
%################## APPLYING JANSON ##############################
%#################################################################
Let $R$ be the set $E(G(n,p))$ and let $A_i$ be the set of edges such that
$A_i = \{ua_{i_1},a_{i_1}a_{i_2},\cdots, a_{i_{k-2}}a_{i_{k-1}},\\
a_{i_{k-1}}v_i\}$.
Let $B_i$ be the event that $A_i \subseteq R$.
So, $\PP[A_i \in R] = \PP[B_i]$.
Let $X_i$ be the indicator random variable for $B_i$ and
$\displaystyle X_B = \sum \limits_{i=1}^{n-2 \choose k-1} X_i$ be the number
of $A_i \subseteq R$.
Hence, $\PP[X_B = 0] = \PP\left[\bigcap \limits_{i=1}^{n-2 \choose k-1}
\overline{B_i}\right]$.
For $1 \leq i, j \leq {n-2 \choose k-1}$ we write $i \sim j$ if
$i \neq j$ and $A_i \cap A_j \neq \emptyset$.
$\Delta$ is defined as $\displaystyle\sum \limits_{i\sim j}\PP[B_i \cap B_j]$.
We would like to show that $\PP[X_B = 0] \to 0$ as $n \to \infty$.
%#################### DETERMINE \mu ################################
First we determine $\displaystyle\mu = \EE[X_B] = \sum
\limits_{i=1}^{n-2 \choose k-1}\PP[B_i]$.
\begin{align}\label{mu}
\EE[X_B] &= \EE\left[\sum \limits_{i=1}^{n-2 \choose k-1} X_i \right]
= \sum \limits_{i=1}^{n-2 \choose k-1} \EE[X_i]
= \sum \limits_{i=1}^{n-2 \choose k-1} \EE[B_i] \nonumber \\
&= {n-2 \choose k-1} p^k \geq \left(\frac{n-2}{k-1}\right)^{k-1} p^k
\qquad\qquad\qquad\qquad\left(\text{by }\, {n\choose k}
\geq \left(\frac{n}{k}\right)^k\right)\nonumber\\
&\geq \frac{(n-2)^{k-1}}{(k-1)^{k-1}}
\left(k \sqrt[k]{\frac{\log n}{n^{k-1}}}\right)^k \nonumber\\
&= \frac{(n-2)^{k-1}}{(k-1)^{k-1}}\, k^k \,\frac{\log n}{n^{k-1}} %\nonumber\\
= \left(\frac{k^k}{(k-1)^{k-1}}\right)\left(\frac{n-2}{n}\right)^{k-1}\log n \nonumber \\
&= k\left(\frac{k}{k-1}\right)^{k-1} \left(1 - \frac{2}{n}\right)^{k-1} \log n \nonumber\\
&\geq k \left(1 - \frac{2}{n}\right)^{k-1} \log n \nonumber\\
&\geq 0.9 k \log n
\end{align}
for $n$ large enough.
Thus, in Janson's Inequality let $\mu = 0.9 k \log n$.
%######################## determine DELTA #########################################
Now we determine $\Delta$.
Assume that the number of edges shared between any given $A_i$
and $A_j$ is given by $t$ and hence, $A_j$ shares at least $t$
vertices with $A_i$.
There are ${n-2 \choose k-1}$ such $A_i$ sets.
We fix one such set $A_i$ and determine the dependencies between
$A_i$ and all other sets $A_j$, where $j \neq i$.
Thus, we have
\begin{align}\label{delta1}
\Delta &= \sum \limits_{i=1}^{n-2 \choose k-1} \PP[B_i \cap B_j] \nonumber \\
&\leq {n-2 \choose k-1} \sum \limits_{\underset{j \sim i}{i \text{ fixed}}}
^{n-2 \choose k-1}\PP[B_j \cap B_i] \nonumber \\
&\leq {n-2 \choose k-1} \sum \limits_{t=1}^{k-1} {k \choose t}
{n-2 \choose k-1-t} p^{2k-t}.
\end{align}
In Equation \ref{delta1}, the probability that a fixed $A_i$ intersects
(i.e. shares) at $t$ edges with a set $A_j$ for $i \neq j$, is
$p^k p^{k-t} = p^{2k-t}$.
When calculating this probability we are interested in counting the
number of edges $t$ that are shared between $A_i$ and $A_j$.
That is, between which vertices $t$ edges are shared is not of interest.
Between any two vertices $u$ and $v_i$ there are $k$ edges and hence,
the number of ways to determine the $t$ shared edges is
${k \choose t}$.
For any $A_j$, the two vertices $u$ and $v_i$ are fixed.
From the $k-1$ other vertices on the path from $u$ to $v_i$, $t$ are shared
with $A_i$.
Hence, to complete $A_j$ that share $t$ edges with $A_i$, there are
${n-2 \choose k-1-t}$ possible ways to add the remaining vertices.
Thus, for a given value $t$, ${k \choose t} {n-2 \choose k-1-t}$ determine
how many sets $A_j$ share precisely $t$ edges with $A_i$.
Thus, from Equation \ref{delta1} we have
\begin{align}\label{delta2}
\Delta &\leq {n-2 \choose k-1}\sum \limits_{t=1}^{k-1} {k \choose t}
{n-2 \choose k-1-t} p^{2k-t} \nonumber\\
&\leq \frac{n^{k-1}}{(k-1)!} 2^k \sum \limits_{t=1}^{k-1}
{n-2 \choose k-1-t} p^{2k-t}
\qquad\qquad\left(\text{by }\, {n\choose k} \leq \frac{n^k}{k!}
\text{ and }\, {n\choose k}\leq 2^n\right)\nonumber\\
&\leq \frac{n^{k-1}}{(k-1)!} 2^k \sum \limits_{t=1}^{k-1}
\frac{(n-2)^{k-1-t}}{(k-1-t)!} p^{2k-t}
\qquad\qquad\left(\text{by }\, {n\choose k} \leq \frac{n^k}{k!}\right)\nonumber\\
&\leq \frac{n^{k-1}}{(k-1)!} 2^k \sum \limits_{t=1}^{k-1}
n^{k-1-t} p^{2k-t}.
\end{align}
%\vspace{4mm}
\noindent We now calculate $n^{k-1-t} p^{2k-t}$,
\begin{align}\label{in_sum}
n^{k-1-t} p^{2k-t} &= \frac{n^{k-t}}{n} p^k p^{k-t}
= \frac{n^{k-t} p^{k-t}}{n} p^k \nonumber\\
&= \frac{n^{k-t} p^{k-t}}{n}
\left(k \sqrt[k]{\frac{\log n}{n^{k-1}}} \right)^k
= \frac{n^{k-t} p^{k-t}}{n}\, k^k\, \frac{\log n}{n^{k-1}} \nonumber\\
&= \frac{n^{k-t} p^{k-t}}{n^k}\, k^k\, \log n
= n^{-t} p^{k-t} k^k \log n \nonumber\\
&= \left(np\right)^{-t} \left(p^k k^k \log n\right)
= \left(n k \sqrt[k]{\frac{\log n}{n^{k-1}}}\right)^{-t}
\left(p^k k^k \log n\right) \nonumber\\
&= \left(n^{1-(k-1)/k}\, k \,\sqrt[k]{\log n} \right)^{-t}
\left(p^k k^k \log n\right) \nonumber\\
&= \left(n^{1/k}\, k \,\sqrt[k]{\log n} \right)^{-t}
\left(p^k k^k \log n\right)
= \frac{\left(p^k k^k \log n\right)}
{\left(k n^{1/k}\, \sqrt[k]{\log n}\right)^t} \nonumber\\
&\leq \frac{\left(p^k k^k \log n\right)}{k n^{1/k}\, \sqrt[k]{\log n}}.
\end{align}
\noindent From Equations \ref{delta2} and \ref{in_sum} we have
\begin{align}
\Delta &\leq \frac{n^{k-1}}{(k-1)!} 2^k \sum \limits_{t=1}^{k-1}
n^{k-1-t} p^{2k-t} %\nonumber\\
\leq \frac{n^{k-1}}{(k-1)!} 2^k \sum \limits_{t=1}^{k-1}
\frac{p^k k^k \log n}{k n^{1/k} \sqrt[k]{\log n}} \nonumber\\
&\leq \frac{n^{k-1}}{(k-1)!} 2^k\, k\,\,
\frac{p^k k^k \log n}{k n^{1/k} \sqrt[k]{\log n}} %\nonumber\\
\leq 2^k \frac{n^{k-1}}{(k-1)!}
\left(k \sqrt[k]{\frac{\log n}{n^{k-1}}}\right)^k
\frac{k^k \log n}{n^{1/k} \sqrt[k]{\log n}} \nonumber\\
&\leq \frac{2^k}{(k-1)!} \, k^{2k}\,
\frac{n^{k-1} \log^2 n}{n^{k-1} n^{1/k} \sqrt[k]{\log n}} %\nonumber\\
\leq O(k) \, \frac{\log^2 n}{n^{1/k} \sqrt[k]{\log n}} %\nonumber\\
\leq O(k)\, \frac{\log^2 n}{n^{1/k}}. \nonumber
\end{align}
Thus, $\Delta \to 0$ as $n \to \infty$. Since $\Delta < \mu$ by Janson's
Inequality we have
\begin{align}
\PP[X_B = 0] &= \PP\left[\bigcap \limits_{i=1}^{n-2 \choose k-1}
\overline{B}_i\right] %\nonumber\\
\leq e^{-\mu/2} \leq e^{-\frac{0.9k \log n}{2}} %\nonumber\\
\leq e^{-\frac{9}{20} k \log n}. \nonumber
\end{align}
%########################################################################################
Thus, the probability that a vertex $u$ is not within distance-$k$ from a
dominator vertex $v_i$ is given by
\begin{align}\label{not_k-adj}
\PP\left[v_i \notin N_k(u)\right] &\leq \PP\left[\bigcap \limits_{i=1}^{n-2 \choose k-1}
\overline{B}_i\right] %\nonumber\\
\leq e^{-\frac{9}{20} k \log n}.
\end{align}
Let $X_u$ be the random variable that denotes the number of
non-dominators of $u$.
We note that $u$ may be a dominating vertex.
Then
\begin{eqnarray*}
\EE[X_u] &\leq& r\,e^{-\frac{9}{20} k \log n}.\nonumber
\end{eqnarray*}
By Markovs's Inequality we have
$\displaystyle\PP[X_u > 0] \leq \EE[X_u] \leq r\,e^{-\frac{9}{20} k \log n}$.
Thus,
\begin{align}\label{Pbadk}
\mathbb{P}[\text{fixed $u$ is bad}] &\leq \mathbb{P}[X_u > 0] %\nonumber
\leq r\,e^{-\frac{9}{20} k \log n}.
\end{align}
By Equation \ref{EXk} and Equation \ref{Pbadk} we have
$\displaystyle\EE[X] \leq n\,r\,e^{-\frac{9}{20} k \log n}$ and by
Markov's Inequality it follows,
\begin{equation}\label{distk}
\mathbb{P}[X>0] \leq \EE[X] \leq n\,r\,e^{-\frac{9}{20} k \log n}.
\end{equation}
\vspace{2mm}
From Equation \ref{distk}, we determine the value of $e^{-\frac{9}{20} k \log n}$
to be
\begin{equation}
e^{-\frac{9}{20} k \log n} \geq \left(e^{\log n}\right)^{-\frac{9}{20} k}
= n^{-\frac{9}{20}k}.\nonumber
\end{equation}
Thus, we have
\begin{equation}
nre^{-\frac{9}{20} k \log n} \leq \frac{nr}{n^{\frac{9}{20}k}}
\leq \frac{r}{n^{\frac{9}{20}k-1}}.\nonumber
\end{equation}
For $k \geq 3$, $\,\displaystyle{\frac{r}{n^{\frac{9}{20}k-1}} \to 0}$ as $n \to \infty$.
Therefore, $\PP[X>0] \to 0$ as $n \to \infty$.
\end{proof}
\vspace{2mm}
\begin{thebibliography}{34}
\providecommand{\natexlab}[1]{#1}
\providecommand{\url}[1]{\texttt{#1}}
\expandafter\ifx\csname urlstyle\endcsname\relax
\providecommand{\doi}[1]{doi: #1}\else
\providecommand{\doi}{doi: \begingroup \urlstyle{rm}\Url}\fi
\bibitem[Alon and Spencer(1992)]{ASE}
N.~Alon and J.~Spencer.
\newblock \emph{The Probabilistic Method}.
\newblock John Wiley \& Sons, Inc., 1992.
\bibitem[Bean et~al.(1994)Bean, Henning, and Swart]{BHS94}
T.~Bean, M.~Henning, and H.~Swart.
\newblock On the integrity of distance domination in graphs.
\newblock \emph{Australasian Journal of Combinatorics}, 10:\penalty0 29--43,
1994.
\bibitem[Bennett(1962)]{Ben62}
G.~Bennett.
\newblock Probability inequalities for the sum of independent random variables.
\newblock \emph{Journal of the American Statistical Association}, 57:\penalty0
33--45, 1962.
\bibitem[Bernstein(1924)]{Bern24}
S.~Bernstein.
\newblock On modification of {Chebyshev's} inequality and of the error formula
of {Laplace} (russian).
\newblock \emph{Section Math\'{e}matique des Annales Scientifique des
Institutions Savantes de l'Ukraine}, 1:\penalty0 38--49, 1924.
\bibitem[Bollob\'{a}s(2001)]{BB}
B.~Bollob\'{a}s.
\newblock \emph{Random Graphs}.
\newblock Cambridge University Press, 2001.
\bibitem[Bonato and Wang(2008)]{BW08}
A.~Bonato and C.~Wang.
\newblock A note on domination parameters in random graphs.
\newblock \emph{Discussiones Mathematicae Graph Theory}, 28:\penalty0 335--343,
2008.
\bibitem[Caro and Roditty(1990)]{CR90}
Y.~Caro and Y.~Roditty.
\newblock A note on the $k$-domination number of a graph.
\newblock \emph{Internation Journal of Mathematics and Mathematical Sciences},
13, 1990.
\bibitem[Caro and Yuster(2000)]{CY00}
Y.~Caro and R.~Yuster.
\newblock Dominating a family of graphs with small connected subgraphs.
\newblock \emph{Combinatorics, Probability and Computing}, 9\penalty0
(4):\penalty0 309--313, 2000.
\bibitem[Chang(2008)]{C08}
G.~Chang.
\newblock The upper bound on $k$-tuple domination numbers of graphs.
\newblock \emph{European Journal of Combinatorics}, 29:\penalty0 1333--1336,
2008.
\bibitem[Chernoff(1952)]{Ch52}
H.~Chernoff.
\newblock A measure of asymptotic efficiency of tests of a hypothesis based on
the sum of ovservations.
\newblock \emph{Annals of Mathematical Statistics}, 23:\penalty0 493--507,
1952.
\bibitem[Cockayne et~al.(1985)Cockayne, Gamble, and Shepherd]{CGS85}
E.~J. Cockayne, B.~Gamble, and B.~Shepherd.
\newblock An upper bound for the k-domination number of a graph.
\newblock \emph{Journal of Graph Theory}, 9\penalty0 (4):\penalty0 533--534,
1985.
\bibitem[Craig(1933)]{Cr33}
C.~Craig.
\newblock On the {Tchebychef} inequality of {Bernstein}.
\newblock \emph{Annals of Mathematical Statistics}, 4:\penalty0 94--102, 1933.
\bibitem[Dai and Wu(2006)]{DW06}
F.~Dai and J.~Wu.
\newblock On constructing $k$-connected $k$-dominating set in wireless
networks.
\newblock \emph{Parallel and Distributed Computing}, 66\penalty0 (7):\penalty0
947--958, 2006.
\bibitem[Dreyer(2000)]{D00}
P.~Dreyer.
\newblock \emph{Applications and variations of domination in graphs}.
\newblock PhD thesis, Department of Mathematics (Rutgers University), 2000.
\bibitem[Fischermann and Volkmann(2005)]{FV05}
M.~Fischermann and L.~Volkmann.
\newblock A remark on a conjecture for the $(k,p)$-domination number.
\newblock \emph{Utilitas Mathematica}, 67:\penalty0 223--227, 2005.
\bibitem[Gagarin and Zverovich(2008)]{GZ08}
A.~Gagarin and V.~Zverovich.
\newblock A gerneralised upper bound for the $k$-tuple domination number.
\newblock \emph{Discrete Mathematics}, 308:\penalty0 880--885, 2008.
\bibitem[Hansberg et~al.(2007)Hansberg, Meierling, and Volkmann]{HMV07}
A.~Hansberg, D.~Meierling, and L.~Volkmann.
\newblock Distance domination and distance irredundance in graphs.
\newblock \emph{The Electronic Journal of Combinatorics}, 14:\penalty0 \#R35,
2007.
\bibitem[Harutyunyan(2014)]{H14}
L.~Harutyunyan.
\newblock On placement of multiple sinks in wireless sensor networks.
\newblock In \emph{To appear in Proceedings of the 15th IEEE International
Symposium on A World of Wireless, Mobile and Multimedia Networks
(WoWMoM'14)}, 2014.
\bibitem[Hoeffding(1963)]{Ho63}
W.~Hoeffding.
\newblock Probability inequalities for sums of bounded random variables.
\newblock \emph{Journal of the American Statistical Association}, 58:\penalty0
13--30, 1963.
\bibitem[Korneffel et~al.(2008)Korneffel, Meierling, and Volkmann]{KMV08}
T.~Korneffel, D.~Meierling, and L.~Volkmann.
\newblock A remark on the $(2,2)$-domination number.
\newblock \emph{Discussiones Mathematicae Graph Theory}, 28\penalty0
(2):\penalty0 361--366, 2008.
\bibitem[Li(2004)]{L04}
S.~Li.
\newblock On connected $k$-domination numbers of graphs.
\newblock \emph{Discrete Mathematics}, 274:\penalty0 303--310, 2004.
\bibitem[Li et~al.(2012)Li, Wu, Ai, and Beyah]{LWAB12}
Y.~Li, Y.~Wu, C.~Ai, and R.~Beyah.
\newblock On the construction of $k$-connected $m$-dominating set in wireless
networks.
\newblock \emph{Journal of Combinatorial Optimization}, 23\penalty0
(1):\penalty0 118--139, 2012.
\bibitem[Liu et~al.(2009)Liu, Tian, and Xu]{LTX09}
Z.~Liu, F.~Tian, and J.~Xu.
\newblock Probabilistic analysis of upper bound for $2$-connected distance
$k$-dominating set in graphs.
\newblock \emph{Theoretical Computer Science}, 410:\penalty0 3804--3813, 2009.
\bibitem[Meir and Moon(1975)]{MM75}
A.~Meir and J.~W. Moon.
\newblock Relations between packing and covering numbers of a tree.
\newblock \emph{Pacific Journal of Mathematics}, 61\penalty0 (1):\penalty0
225--233, 1975.
\bibitem[Rautenbach and Volkmann(2007)]{RV07}
D.~Rautenbach and L.~Volkmann.
\newblock New bounds on the $k$-domination number and the $k$-tuple domination
number.
\newblock \emph{Applied Mathematics Letters}, 20:\penalty0 98--102, 2007.
\bibitem[Spohn and Garcia-Luna-Aceves(2007)]{SG07}
M.~Spohn and J.~Garcia-Luna-Aceves.
\newblock Bounded distance multi-clusterhead formation in wireless ad hoc
networks.
\newblock \emph{Ad Hoc Networks}, 5\penalty0 (4):\penalty0 504--530, 2007.
\bibitem[Sridharan et~al.(2002)Sridharan, Subramanian, and Elias]{SSE02}
N.~Sridharan, V.~Subramanian, and M.~Elias.
\newblock Bounds on the distance two-domination number of a graph.
\newblock \emph{Graphs and Combinatorics}, 18\penalty0 (3):\penalty0 667--675,
2002.
\bibitem[Tian and Xu(2007)]{TX07}
F.~Tian and J.~Xu.
\newblock On distance connected domination numbers of graphs.
\newblock \emph{Ars Combinatoria}, 84:\penalty0 357--367, 2007.
\bibitem[Tian and Xu(2009)]{TX09}
F.~Tian and J.~Xu.
\newblock A note on distance domination numbers of graphs.
\newblock \emph{Australasian Journal of Combinatorics}, 43:\penalty0 181--190,
2009.
\bibitem[Wang and Xiang(2009)]{WX09}
B.~Wang and K.-N. Xiang.
\newblock On $k$-tuple domination of random graphs.
\newblock \emph{Applied Mathematics Letters}, 22:\penalty0 1513--1517, 2009.
\bibitem[Wang(2010)]{W10}
C.~Wang.
\newblock The independent domination number of random graph.
\newblock \emph{Utilitas Mathematica}, 82:\penalty0 161--166, 2010.
\bibitem[Wieland and Godbole(2001)]{WG01}
B.~Wieland and A.~Godbole.
\newblock On the domination number of a random graph.
\newblock \emph{The Electronic Journal of Combinatorics}, 8:\penalty0 \#R37,
2001.
\bibitem[Zhang et~al.(2009)Zhang, Liu, and Li]{ZLL09}
Z.~Zhang, Q.~Liu, and D.~Li.
\newblock Two algorithms for connected $r$-hop $k$-dominating set.
\newblock \emph{Discrete Mathematics, Algorithms and Applications}, 1\penalty0
(4):\penalty0 485--498, 2009.
\bibitem[Zhao et~al.(2007)Zhao, Wang, and Xu]{ZWX07}
W.~Zhao, H.~Wang, and G.~Xu.
\newblock Total $k$-domination in graphs.
\newblock \emph{International Journal of Pure and Applied Mathematics},
35:\penalty0 235--242, 2007.
\end{thebibliography}