beamer.tex

   1 \documentclass{beamer}
   2 \usetheme{i4}
   3 \usepackage[utf8]{inputenc}
   4 \usepackage{tikz}
   5 \usepackage{multicol}
   6 \usepackage{listings}
   7 \lstloadlanguages{lisp}
   8
   9 \usetikzlibrary{svg.path,positioning,intersections}
  10 \usepgflibrary{shapes.geometric}
  11 \usepgflibrary{shapes.misc}
  12 \usepgflibrary{shapes.symbols}
  13
  14 \usepgflibrary{shapes}
  15 \usetikzlibrary{shapes,decorations,shadows}
  16 \usetikzlibrary{decorations.pathmorphing}
  17 \usetikzlibrary{decorations.shapes}
  18 \usetikzlibrary{fadings}
  19 \usetikzlibrary{patterns}
  20 \usetikzlibrary{calc}
  21 \tikzstyle{netdb}=[anchor=center,color=black,rectangle,draw,minimum
  22  size=.6em,minimum height=.2em]
  23 \tikzstyle{client}=[fill=i4gray,rectangle,draw]
  24 \tikzstyle{chain}=[rectangle,draw,minimum size=1em,minimum height=.5em]
  25 \tikzstyle{arrow}=[->,thick,draw,shorten <=2pt,shorten >=2pt,]
  26 \tikzstyle{tunnel}=[fill=gray,shape=ellipse,minimum size=4em,minimum height=1.1em]
  27
  28 \newcommand{\iip}[1]{\textcolor{i4red}{#1}}
  29
  30 \author[Christoph Egger]{{\bf Christoph Egger}, Johannes Schlumberger, Christopher
  31   Kruegel, Giovanni Vigna}
  32 \title{Practical Attacks Against The I2P Network}
  33 \institute{Friedrich-Alexander University Erlangen-Nuremberg\\
  34 University of California, Santa Barbara}
  35 \date{October 25, 2013}
  36 \begin{document}
  37
  38 \begin{frame}[plain]
  39   \vspace{1.5em}
  40   \titlepage
  41   \begin{center}
  42   \includegraphics[width=0.23\paperwidth]{fau_siegel}
  43   \hspace{1.5em}
  44   \includegraphics[width=0.25\paperwidth]{ucsbseal}
  45   \end{center}
  46 \end{frame}
  47
  48 \begin{frame}
  49   \frametitle{Outline}
  50   \begin{multicols}{2}
  51     \begin{block}{What is I2P?}
  52       \begin{itemize}
  53       \item Tunnels
  54       \item Network Database
  55       \item \textcolor{gray}{Floodfill Participation}
  56       \item Threat model
  57       \end{itemize}
  58     \end{block}
  59     \begin{block}{\textcolor{gray}{Floodfill Takeover Attack}}
  60
  61     \end{block}
  62     \begin{block}{Sibyl Attack}
  63       \begin{itemize}
  64       \item Attack Description
  65       \item Evaluation
  66       \end{itemize}
  67     \end{block}
  68     \begin{block}{\textcolor{gray}{Eclipse Attack}}
  69
  70     \end{block}
  71     \begin{block}{Deanonymization Attack}
  72       \begin{itemize}
  73       \item Attack Description
  74       \item Evaluation
  75       \end{itemize}
  76     \end{block}
  77     \begin{block}{Conclusions}
  78       \begin{itemize}
  79       \item Limitations
  80       \item I2P improvements
  81       \item \textcolor{gray}{Related Work}
  82       \end{itemize}
  83     \end{block}
  84   \end{multicols}
  85 \end{frame}
  86
  87 \begin{frame}
  88   \frametitle{Anonymity}
  89   \begin{block}{Who needs anonymity}
  90     \begin{itemize}
  91     \item Criminals
  92     \item Civil rights activists
  93     \item Everyone else
  94     \end{itemize}
  95   \end{block}\pause
  96   \begin{block}{I2P and tor}
  97     \begin{itemize}
  98     \item Tor: directory authorities $\leftrightarrow$ I2P:
  99       decentralized DHT
 100     \item Tor: proxy to the outside world $\leftrightarrow$ I2P:
 101       separated \emph{Darknet}
 102     \end{itemize}
 103   \end{block}
 104 \end{frame}
 105
 106 \begin{frame}
 107   \frametitle{Introduction to I2P}
 108   \begin{itemize}\addtolength{\itemsep}{1\baselineskip}
 109   \item Solution for anonymous communication
 110   \item Separated from the ``Internet'' -- \emph{Darknet}
 111   \item Fully distributed design
 112   \item Based on onion-routing
 113   \item Between 18,000 and 28,000 active users
 114   \end{itemize}
 115 \end{frame}
 116
 117 \section{I2P}
 118 \begin{frame}
 119   \frametitle{I2P}
 120   \begin{multicols}{2}
 121     \begin{block}{Router}
 122       \begin{itemize}
 123       \item Handle connections
 124       \item Provide name services
 125       \end{itemize}
 126     \end{block}
 127 %    \pause
 128     \begin{block}{Applications}
 129       \begin{itemize}
 130       \item Server, client or P2P software
 131       \item Sockets interface with TCP-like or UDP-like semantics
 132       \end{itemize}
 133     \end{block}
 134 %    \pause
 135     \begin{figure}
 136       \centering
 137       \begin{tikzpicture}[scale=1.2]
 138         \tikzstyle{every node}=[font=\small]
 139         \node[minimum width=9em,minimum height=9em,draw=gray](clientpc) at (25mm,9mm) {};
 140         \node[above=0mm of clientpc.south] {User's Computer};
 141         \node[cloud,drop shadow,fill=white,draw,minimum
 142         width=4.5em,minimum height=2.5em](ip) at (22mm,30mm) {I2P};
 143         \node[client](client) at (22mm, 18mm) {I2P Router};
 144         \node[rectangle,draw,below=0mm of client.south east](app1) {Application};
 145         \node[rectangle,draw,below=0mm of app1.south](app2) {Application};
 146         \node[rectangle,draw,below=0mm of app2.south](app3) {Application};
 147
 148         \path[arrow] (app1.west) -| ([xshift=3.5mm]client.south west);
 149         \path[arrow] (app2.west) -| ([xshift=2.5mm]client.south west);
 150         \path[arrow] (app3.west) -| ([xshift=1.5mm]client.south west);
 151         \path[arrow] ([xshift=.5mm]client.north) -- ([xshift=.5mm]ip.south);
 152         \path[arrow] ([xshift=-.5mm]ip.south) -- ([xshift=-.5mm]client.north);
 153       \end{tikzpicture}
 154     \end{figure}
 155
 156   \end{multicols}
 157 \end{frame}
 158
 159 \begin{frame}
 160   \frametitle{Tunnels}
 161   \begin{itemize}
 162   \item Using onion-routing for anonymity
 163   \item Unidirectional
 164   \item Paired for bi-directional communication
 165   \end{itemize}\pause
 166   \begin{block}{Client Tunnels}
 167     \begin{itemize}
 168     \item Used for data interactions
 169     \item Several per application
 170     \end{itemize}
 171   \end{block}
 172 %  \pause
 173   \begin{block}{Exploratory Tunnels}
 174     \begin{itemize}
 175     \item Used for database interaction
 176     \item 2 to 3 per node
 177     \end{itemize}
 178   \end{block}
 179 \end{frame}
 180
 181 \begin{frame}
 182   \frametitle{Network Database}
 183   \begin{itemize}
 184   \item<1-> Kademlia-like DHT based on \texttt{XOR}-distance run on
 185     320 super-nodes
 186   \item<1-> Layout of the database changes completely every day
 187   \item<2-> \iip{databaseRecord}\\
 188     Information named using a hash over their cryptographic Keys
 189   \item<2-> \iip{storageLocation}\\
 190     Hash over name and today's date
 191   \item<3-> \iip{routerInfo}\\
 192     Peer information: IP address, port, protocol, keys
 193   \item<3-> \iip{leaseSet}\\
 194     Service information: Entry tunnels, keys
 195   \end{itemize}
 196 \end{frame}
 197
 198 \begin{frame}
 199   \frametitle{Sample Interaction}
 200   Accessing a hidden website -- ``http://civilrights.i2p''
 201   \begin{figure}
 202     \centering
 203     \input{sample-interaction}
 204   \end{figure}
 205 \end{frame}
 206
 207 \begin{frame}
 208   \frametitle{Threat Model}
 209   \begin{itemize}\addtolength{\itemsep}{1\baselineskip}
 210   \item Implicitly specified in terms of attacks considered
 211   \item Only allows local adversaries: No global view about traffic
 212     passing through the network
 213   \item Only allows limited number of malicious nodes -- around 20\,\%
 214     of \iip{netDB} super-nodes and 20\,\% of total nodes
 215   \item Secure cryptographic primitives
 216   \end{itemize}
 217 \end{frame}
 218
 219 \section{Sibyl Attack}
 220 \begin{frame}
 221   \frametitle{Sybil Attack}
 222   \begin{block}{Definition}
 223     In a sybil attack, the adversary utilizes multiple identities to
 224     break assumptions about the system
 225   \end{block}\pause
 226   \begin{block}{Goal}
 227     Gaining control over parts of the keyspace in the \iip{netDB} with
 228     limited resources. As a result be the only source considered for
 229     certain pieces of data and therefore able to monitor every access
 230     to it
 231   \end{block}\pause
 232   \begin{block}{Challenge}
 233     Active identities require considerable resources to be useful
 234   \end{block}\pause
 235   \begin{block}{Solution}
 236     Compute a large pool of identities and only activate the relevant ones
 237   \end{block}
 238 \end{frame}
 239
 240 \begin{frame}
 241   \frametitle{Sybil Attack}
 242   \begin{block}{Generating identities}
 243     \begin{itemize}
 244     \item Building a database of 50,000 identities takes around 30
 245       minutes on 12-core Xeon server
 246     \item 156 nodes on average between two adjacent database nodes
 247     \item All identities available to all malicious nodes
 248     \end{itemize}
 249   \end{block}\pause
 250   \begin{block}{Using identities}
 251     \begin{itemize}
 252     \item Malicious nodes can calculate the correct identities and
 253       change identity at any time
 254     \item Nodes coordinate to avoid duplicate identities
 255     \end{itemize}
 256   \end{block}
 257 \end{frame}
 258
 259 \section{Deanonymizing Users}
 260 \begin{frame}
 261   \frametitle{Deanonymizing Users}
 262   \begin{block}{Goal}
 263     Given an user Alice and a resource R, we want to confirm or refute
 264     Alice using R with high probability.
 265   \end{block}\pause
 266   \begin{block}{Vulnerability}
 267     \begin{itemize}
 268     \item<2-> Nodes store their \iip{routerInfo} directly in the \iip{netDB}
 269     \item<3-> Nodes verify the storage 20 seconds later using one of their
 270       \iip{exploratory tunnels}
 271     \item<4-> Nodes use the same \iip{exploratory tunnel} again for
 272       resource lookups
 273     \end{itemize}
 274   \end{block}
 275 \end{frame}
 276
 277 \begin{frame}
 278   \frametitle{Deanonymizing Users}
 279   \begin{figure}
 280     \centering
 281     \input{deanonymization}
 282   \end{figure}
 283 \end{frame}
 284
 285 \begin{frame}
 286   \frametitle{Deanonyizing Attack}
 287   \begin{block}{Setup}
 288     \begin{itemize}
 289     \item 20 attacking nodes in a single network
 290       \begin{itemize}
 291       \item 10 nodes capturing resource lookups
 292       \item 10 nodes performing timing attack on \iip{routerInfo} storage
 293       \end{itemize}
 294     \item 6 monitoring nodes: split between two continents
 295     \end{itemize}
 296   \end{block}\pause
 297   \begin{block}{Results}
 298     \begin{itemize}
 299     \item 60\,\% of potentially observable links detected
 300     \item 52\,\% of attributed hits correct
 301     \item Working equally well for geographically remote hosts
 302     \end{itemize}
 303   \end{block}
 304 \end{frame}
 305
 306 \section{Conclusions}
 307 \begin{frame}
 308   \frametitle{Limitations}
 309   \begin{itemize}\addtolength{\itemsep}{1\baselineskip}
 310   \item Only works reliably for longer/repeated resource access
 311   \item Less reliable for popular resources
 312   \item Needs extra resources per tracked user and per resource
 313   \end{itemize}
 314 \end{frame}
 315
 316 \begin{frame}
 317   \frametitle{I2P Improvements}
 318   \begin{itemize}\addtolength{\itemsep}{1\baselineskip}
 319   \item Working with I2P developers to make it secure again
 320   \item<2-> Implemented improvements
 321     \begin{itemize}
 322     \item Limiting \iip{netDB} nodes per IPv4 network
 323     \item Randomizing the time delta
 324     \item Ongoing discussion about deeper modifications to the \iip{netDB}
 325     \end{itemize}
 326   \item<3-> Further improvements
 327     \begin{itemize}
 328     \item Ignoring new \iip{netDB} nodes
 329     \item Removing storage verification
 330     \item Expiring tunnels after storage verification
 331     \end{itemize}
 332   \end{itemize}
 333 \end{frame}
 334
 335 \begin{frame}{Questions?}
 336     \vspace*{\fill}
 337     \begin{center}
 338         \includegraphics[width=7cm]{42.pdf}
 339     \end{center}
 340     \vspace*{\fill}
 341 \end{frame}
 342
 343 \begin{frame}
 344   \frametitle{Bibliography}
 345   \nocite{Mittal:2012}
 346   \nocite{Timpanaro:2011}
 347   \nocite{Herrmann:2011}
 348   \nocite{Douceur:2002}
 349   \nocite{Dingledine:2004}
 350   \nocite{Evans:2011}
 351   \bibliographystyle{plain}
 352   \tiny{\bibliography{../i2p}}
 353 \end{frame}
 354
 355 \end{document}
 356
 357 %%% Local Variables:
 358 %%% mode: latex
 359 %%% TeX-master: t
 360 %%% TeX-PDF-mode: t
 361 %%% End: