Conference PaperPDF Available

Visualization assisted detection of sybil attacks in wireless networks


Abstract and Figures

In wireless networks, the authenticity and uniqueness of node identities are essential to the fundamental operations such as routing, resource allocation, and intrusion detec- tion. In this paper, we investigate Sybil attack, an attack in which a malicious node illegitimately acquires multiple identities and performs as these nodes simultaneously. We propose an effective approach to monitoring and detecting such attacks by integrating network security and visualiza- tion methods. The security component explores the time- varying network topology and its statistical and geometry information to detect the existence of Sybil attacks. The visualization component incorporates the detection results and provides an effective mechanism to illustrate abnormal topology patterns and locate fake identities. These two com- ponents are integrated into a practical system that takes advantage of both interactive visualization and intelligent security methods. Experimental studies are conducted to investigate the impacts of the network parameters such as node connectivity on the detection capability of the pro- posed mechanism.
Content may be subject to copyright.
Visualization Assisted Detection of Sybil Attacks in
Wireless Networks
Weichao Wang
ITTC and Department of EECS
University of Kansas
Aidong Lu
CS Department
University of North Carolina at Charlotte
In wireless networks, the authenticity and uniqueness of
node identities are essential to the fundamental operations
such as routing, resource allocation, and intrusion detec-
tion. In this paper, we investigate Sybil attack, an attack
in which a malicious node illegitimately acquires multiple
identities and performs as these nodes simultaneously. We
propose an effective approach to monitoring and detecting
such attacks by integrating network security and visualiza-
tion methods. The security component explores the time-
varying network topology and its statistical and geometry
information to detect the existence of Sybil attacks. The
visualization component incorporates the detection results
and provides an effective mechanism to illustrate abnormal
topology patterns and locate fake identities. These two com-
ponents are integrated into a practical system that takes
advantage of both interactive visualization and intelligent
security methods. Experimental studies are conducted to
investigate the impacts of the network parameters such as
node connectivity on the detection capability of the pro-
posed mechanism.
Categories and Subject Descriptors
C.2.0 [Computer-Communication Networks]: General—
Security and protection; H.5.2 [Information Systems]: In-
formation Interfaces and Presentation—User interfaces
General Terms
Algorithms, Security
Interactive Detection, Sybil Attacks, Visualization on Net-
work Security, Wireless Networks, Topology Visualization
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
VizSEC’06, November 3, 2006, Alexandria, Virginia, USA.
Copyright 2006 ACM 1-59593-549-5/06/0011 ...$5.00.
As wireless networks are widely adopted in various envi-
ronments and applications, security has become one of their
top priorities. Among many features that need to be pro-
tected, the uniqueness and authenticity of the node identity
must be enforced to enable the fundamental operations such
as routing, resource allocation, and misbehavior detection.
For example, Intrusion Detection System (IDS) in wireless
networks [36] detects the attacks and isolates the malicious
nodes by matching the patterns of the known intrusions [3]
or discovering the anomalies [6, 14, 20] in the network activ-
ities. If the attackers can easily generate a fake identity and
participate into the network operations, the effectiveness of
IDS will be drastically weakened. Therefore, mechanisms
must be properly designed to prevent or detect such attacks.
In this paper, we focus on the Sybil attack [11], an at-
tack that specifically targets the node identities in wireless
networks. In a Sybil attack, a single malicious node plays
the roles of multiple legitimate members of the network by
impersonating their identities or claiming fake IDs. If there
are a group of collusive attackers, each of them can pretend
to be the whole group simultaneously at different places in
the network, thus manipulating the results of localized vot-
ing or data aggregation [30]. Sybil attacks will also enable
the malicious nodes to take over the control of the whole
network by compromising a limited number of physical de-
vices, and defeat the replication mechanisms in distributed
systems [11].
Noticing the serious harm that Sybil attacks can make, re-
searchers have proposed several approaches to defend against
such attacks and encouraging results have been collected [5,
10, 11, 30]. Existing approaches usually detect Sybil attacks
by verifying whether a pair of nodes has distinct resources,
distinct knowledge, or distinct positions. Since the verifica-
tions are conducted in a localized manner, the approaches
are especially effective in environments with a relatively sta-
ble topology such as sensor networks or when the nodes
move slowly. However, under the scenarios that a fake iden-
tity dynamically switches among multiple collusive attack-
ers, global network topology must be monitored. Since the
ever-increasing network size and lengthened network lifetime
will drastically increase the amount of information, we need
more powerful techniques such as scientific visualization to
assist the representation of the data and the discovery of the
hidden connections among them.
In this paper, we present an approach to detecting Sybil
attacks in wireless networks that integrates security and vi-
sualization methods. The mechanism monitors the neigh-
bor relations among wireless nodes and the network topol-
ogy changes, and identifies the suspicious Sybil nodes by
visualizing the anomalies introduced by fake identities. A
comprehensive visualization interface is designed to provide
a global view of the network topology evolution so that the
attackers can be located even when they dynamically switch
among multiple compromised physical devices.
The contributions of the proposed approach include: (1)
Since the proposed approach is based on the evolution of
global network topology, it provides an effective method
to detect the Sybil attacks that cannot be identified in a
localized manner. (2) The integration of the visualization
techniques and security methods provides an intuitive and
scalable vehicle to information representation and attack de-
tection. (3) Since the proposed approach detects Sybil nodes
solely based on neighbor relationships among wireless nodes,
it can be applied to more dynamic environments such as mo-
bile ad hoc networks.
As we will demonstrate, the proposed mechanism can ef-
fectively identify the Sybil nodes. Moreover, since many
attacks on wireless networks will lead to anomalies in the
network topology, the proposed analysis methods and visu-
alization tools can be adapted to detect other attacks.
The remainder of the paper is organized as follows: Sec-
tion 2 provides the background of wireless networking, how
Sybil attacks are conducted, and the research challenges.
Section 3 reviews the previous research efforts that con-
tribute to our approach. Section 4 provides an overview
of the approach. In section 5, we describe the design and
development of the visualization tools. The details of the
mechanism to detect Sybil attacks are presented in section
6. Section 7 presents the experimental results. Finally, sec-
tion 8 concludes the paper and discusses future extensions.
While the Sybil attacks can be summarized as a malicious
device presenting multiple identities to the network, a more
detailed classification of the attacks will help improve our
understanding and illustrate the detection capability of the
proposed approach. Here we borrow the taxonomy defined
in [30] and describe two dimensions of the attacks.
The first dimension focuses on the connections among the
Sybil nodes and the legitimate nodes. If they can communi-
cate directly, this is a direct Sybil attack. On the contrary,
in an indirect Sybil attack, a malicious device will claim to
have the paths to reach the Sybil nodes and all messages
have to go through it. In the second dimension, the attacks
are divided into two groups based on whether multiple fake
identities participate into the network activities simultane-
ously. For example, in a non-simultaneous Sybil attack, only
after a fake identity “leaves” the network, the next one will
While it is more difficult to link together multiple fake
identities that appear in different periods of network lifetime
and detect non-simultaneous Sybil attacks, their impacts on
network security are also limited. For example, a Sybil node
that is not a member of the network cannot cast a vote dur-
ing the leader election procedure. Therefore, in this paper,
we focus on the simultaneous Sybil attacks. To evaluate the
proposed mechanism in a more realistic environment, we as-
sume that both direct and indirect Sybil attacks exist in
the network and a malicious node can dynamically switch
between the two kinds. We also assume that multiple ma-
licious physical devices co-exist in the network and a Sybil
node can switch among them. This assumption leads to
the need to monitor the global network topology. Consider-
ing the large amount of information to be processed that is
caused by the ever-increasing network size and lengthened
network lifetime, we adopt visualization techniques to as-
sist the representation of the data and the discovery of the
implicit connections among them.
3.1 Sybil Attack Detection in Networks
Sybil attack is one particularly harmful attack on distrib-
uted systems and wireless networks [11]. This attack has
been demonstrated to be detrimental to many important
network functions. For example, Sybil attack is discussed in
an architecture for secure resource peering in an Internet-
scale computing infrastructure [16]. The problem of Sybil
attack has been formalized [10] to show that there is no
symmetric sybilproof reputation function and a collection
of flow-based asymmetric reputation functions can be given
under some conditions. Newsome et al. have systematically
classified these attacks into several types and analyzed their
threats to wireless sensor networks [30].
Based on the detection mechanisms, we roughly divide
previous approaches into two categories: identity-based or
location-based methods. Identity-based approaches usually
mitigate the Sybil attacks by limiting the generation of valid
node information, such as the pre-distributed secret keys
[30]. To secure routing for structured peer-to-peer over-
lay networks, Sybil attacks are decreased through charg-
ing expensive fees to each newly created nodeID or bind-
ing nodeIDs to real-world identities [8]. In evaluating the
admission control framework, a special purpose certificate
asserting group membership is issued to each peer upon ad-
mission [33]. Another detecting approach is proposed for
vehicular ad hoc networks through possible explanations for
collected data of each node [18]. The method of radio re-
source testing [30] is based on the assumption that a node
cannot send out two signals at different frequencies at the
same time.
Location-based approaches utilize the fact that each node
can only be at one position at a specific moment. Local-
ization algorithms, such as SeRLoc [24], are proposed to
allow sensors to passively determine their locations under
known attacks including Sybil attack. The geometric prop-
erties of message transmission delay are also explored to
reduce the impacts of Sybil attacks [5]. This technique is
based on several assumptions which may limit the attack
complexities. Our approach utilizes the network topology
information. With the integration of visualization and se-
curity techniques, our approach can be used to detect Sybil
attacks under more sophisticated scenarios.
3.2 Visualization for Network Security
With the fast development of computer security mecha-
nisms, the scale and complexity of the security data put ever-
increasing challenges to the representation and understand-
ing of the information. Visualization techniques have been
adopted by the researchers to bridge the gap. For exam-
ple, researchers have designed mechanisms that can provide
an overview of the traffic patterns of thousands of hosts [4].
Mechanisms have been developed to provide a scalable rep-
resentation of the intrusion alarms in multiple class-B IP ad-
dress ranges [2, 22, 23]. Researchers have also developed a
visualization methodology to characterize the most common
and versatile intrusions, network scans, based on their pat-
terns and wavelet scalograms [29]. Another approach uses
IP address and port number histographs to detect and an-
alyze the scan attacks [32]. VisFlowConnect-IP [35] allows
anomalous traffic detection through a link-based network
flow visualization tool.
Under many conditions, the security data acquired from
multiple methods must be investigated jointly to improve
the detection accuracy and efficiency. The research efforts in
[15] provide a visual correlation between the host processes
and network traffic. In both [29] and [32], the approaches
can identify the similarity among different scan attacks or
NetFlow signatures.
While many visualization approaches to network security
require a large amount of high-dimensional data, several
mechanisms focus on the big picture. For example, the
mechanism in [28] takes very coarsely detailed data to help
uncover interesting security events. The mechanism in [31]
overcomes the scalability issues inherent in visualizing mas-
sive networks through sampling. In [19], low level textual
data are provided in the context of a high-level, aggregated
graphical display. Disparate logs are also visualized to show
the correlation of network alerts based on what, when, and
where [26].
4.1 Network Assumptions
We assume that the links among wireless nodes in the
network are bidirectional and two neighboring nodes can
always send packets to each other. This assumption will
hold under most conditions when the power of the nodes
has not been exhausted. We assume that two nodes are
neighbors when the distance between them is shorter than
r,whereris defined as the radio range.
We assume that a special node exists in the network,
which is called the “controller”. It will integrate, process,
and visualize the topology information that is collected by
the wireless nodes and detect the Sybils. We assume that
the controller has the storage and computation resources
that are required by the proposed mechanism. For example,
if the network topology is viewed as a graph, the controller
can locate the cut points of it in a short period of time. In
our experimental studies, we use a PC with 1.8GHz CPU
as the controller and it can process the information of a
network containing several hundreds of nodes in real time.
For the wireless networks that have an infrastructure, the
controller can be chosen from the special nodes. For ex-
ample, in a multi-hop cellular network [27], a base station
can play the role of controller. In those pure ad hoc net-
works, leader election mechanisms [34] can be adopted to
determine the controller based on the trustworthiness of the
mobile nodes and the available resources.
For each wireless node in the network, we assume that it
has established a pair-wise key with the controller. This task
can be accomplished during the network initiation procedure
or based on some pre-distributed information [9, 12, 13, 25].
We also assume that every node moves independently in
the network [21]. The impacts of group movement on the
proposed approach will be investigated in future extensions.
4.2 System Overview
To monitor and detect Sybil attacks, we have integrated
security and visualization methods into a convenient de-
tection tool. Our system concentrates on visualizing and
observing significant patterns from time-dependent network
topology information. Intelligent algorithms are embedded
in the system to identify potential abnormal events and pro-
vide additional validation methods.
As shown in Figure 1, the information of network topol-
ogy is first collected (section 5.1), processed, and visualized
at the controller (section 5.2). To simplify user interaction
in the visualization process, we use statistical topology in-
formation to identify a suspicious node list (section 6.1.1).
The users can adjust this suspicious node list with their ex-
pertise and change the visualization easily to better reveal
the event correlations (section 5.2). To provide additional
validation of the user decisions, two algorithm components
are designed for the detection of direct Sybil attacks (section
6.1.2) and indirect Sybil attacks (section 6.1.3) respectively.
Details of the system design will be discussed in section 6.2.
Figure 1: The system architecture.
5.1 Collection of Topology Information
Since the proposed mechanism detects the Sybil attacks
by monitoring the changes and anomalies in the network
topology, in this section, we describe how the information
can be correctly collected and integrated by the controller.
We know that the neighbor relations among wireless nodes
may change because of various reasons such as node move-
ment, device malfunction, battery exhaustion, and unreli-
able transmission medium. Therefore, a node must be able
to detect its active neighbors dynamically. A widely adopted
approach is to let the mobile node periodically broadcast a
short message containing its identity (called ‘beacon’ packet)
and the neighbors receiving this packet will add the node
into their neighbor lists. In the proposed approach, every
node will periodically send its neighbor list to the controller.
To prevent the list from being altered during transmission,
it is protected by the pair-wise key between the controller
and the node.
While a legitimate node will faithfully report its neigh-
bors, an attacker will manipulate the list to avoid being de-
tected. For example, a malicious physical device may claim
to have a route to an indirect Sybil node so that more traf-
fic will be attracted to it. However, it will not report the
Sybil node as its neighbor to the controller. To prevent the
manipulation and its impacts on Sybil detection, we require
every neighbor list that is transmitted to the controller to be
authenticated by the Message Authentication Code (MAC)
of the nodes in the list. The nodes that are not in the list
will not be adopted by the neighbors in routing or other net-
work activities. Therefore, a Sybil node cannot be hidden
from the controller.
Because of the topology changes, the routes from the mo-
bile nodes to the controller need to be updated. In the
proposed mechanism, the controller will periodically broad-
cast a route discovery packet to the nodes within the radio
range and mark the path length to itself as 0. The nodes
receiving the packet will increase the path length by one and
re-broadcast it. With every node remembers the previous
hop, increases the path length by one, and re-broadcasts
the packet, the routes to the controller will be established.
The frequency to broadcast the route discovery packets can
be determined by the radio range and the node movement
patterns [17].
Using the received neighbor lists, the controller can re-
generate the network topology. For example, a matrix rep-
resenting the connectivity and shortest paths among the
wireless nodes can be calculated. When multiple neighbor
matrices are sorted by their sampling time, the changes in
network topology can be illustrated as a volumetric data.
Since the amount of information will increase fast with the
size of the network and the number of topology snapshots,
we adopt visualization techniques to represent the data and
assist the Sybil detection.
5.2 Visualization of Network Topology
In a wireless network, the node mobility has created severe
challenges to defending against malicious attacks. Since the
network topology is common information in many network-
ing applications, we concentrate on building a visualization
tool for this data that can be easily extended to detect mul-
tiple attacks.
The network topology data often contains enough infor-
mation to monitor and detect intrusions. However, it is very
difficult to visualize this information in a manner that can
be easily understood by users. Since the topology informa-
tion acquired from multiple time steps composes a regular
volumetric data, we first try to use general 2D and 3D visu-
alization approaches to look at this data, including various
2D cut views, statistical views, and 3D direct volume ren-
dering techniques. As shown in Figure 2 (a)(b) and Figure 3
(a), it is difficult to obtain much useful information directly
from these visualizations.
An interesting feature we find is that when we sort the
node sequence according to certain criteria, we may see some
obvious patterns in both 2D and 3D visualizations. For
example, the 2D statistical view shows a grid pattern in
Figure 2 (c). Therefore, our basic idea is to develop an
approach to reveal the significant patterns in the topology
information by grouping the nodes based on the similarity
among their topology features, thus enabling the detection
of malicious attacks.
We have explored the typical patterns under Sybil at-
tacks, including both direct and indirect simultaneous at-
tacks. Figure 2 (c) and Figure 3 (b) show a grid structure
from the indirect attacks which is caused by the lack of di-
rect communication among the fake identities and legitimate
nodes. The bright square in the bottom left corner in Fig-
ure 4 (c)(d) indicates direct Sybil attacks. On the contrary,
there does not exist any obvious pattern under normal net-
work operations.
Scalability is a practical issue for the topology visualiza-
tion, since the rendering resolution is limited by the screen
size and human perception capability. To preserve the sig-
nificant features in the topology information across different
scales, we need to enlarge the range-of-interest in the pre-
vious patterns, such as the highlighted regions in Figure 5
(a). One simple solution is to assign the range-of-interest
according to the total neighbor numbers of each node in the
specified time period. As shown in Figure 5, this scaling
method preserves more significant information than general
zoom out function.
It will be time consuming if a user is asked to manually
adjust the node sequence. Therefore, we integrate an au-
tomatic computation process to assist in determining the
suspicious nodes and to accelerate the Sybil attack detec-
tion. This computation mechanism will be discussed in the
next section and the details about the interaction will be
described in the system design section.
6.1 Detection Algorithms
6.1.1 Determining the Suspicious Node List
As we will demonstrate in the later parts, the detection
operations of the Sybil attacks will focus on the neighbor
relations among the wireless nodes. Although the opera-
tions on every single node are not computationally intensive
(e.g. determine whether removing a node will disconnect
the network), when the network consists of hundreds or even
thousands of nodes, the controller will be overwhelmed by
the processing overhead. Therefore, an efficient mechanism
must be designed to filter out a suspicious node list.
Studying the scenarios of direct and indirect Sybil attacks,
we find that the connectivity among the Sybil nodes attach-
ing to the same physical device presents a “locality”. For
example, for two fake identities under direct Sybil attacks,
although they can pretend not to be neighbors, a two-hop
path must exist through a legitimate neighbor. Similarly, for
indirect Sybil attacks, the path length among the fake iden-
tities is solely determined by the number of Sybil nodes and
their claimed organizations. On the contrary, for the legiti-
mate nodes, since we assume that they move independently,
the average path length depends on the network scale and
node density.
Therefore, we can efficiently calculate the distribution of
the path length between every pair of nodes based on the
collected topology information and identify the group of sus-
picious nodes. Figure 6 illustrates the cumulative distribu-
tion function (CDF) of the path length between a pair of
legitimate nodes and Sybil nodes. The identities under di-
rect Sybil attack pretend not to be neighbors, and a two-hop
path exists between them. The malicious device claims that
a three-hop path exists between the two identities under in-
direct Sybil attack. The anomalies can be easily identified.
6.1.2 Locating Direct Sybil Node Pairs
After the group of suspicious nodes are determined, more
computationally expensive detection operations can be con-
ducted. In this part we introduce the detection of direct
Figure 2: Three statistical neighbor relation views of the nodes, with bright colors indicating high connection
values. (a) It is very difficult for users to directly obtain useful information from the topology, although it
contains enough information to detect anomalies. (b) Changing the node sequence does not necessarily show
additional information. (c) Only when arranged appropriately, the topology information can demonstrate
significant features for users to detect intrusions. The left colormap beside each 2D view shows the node
group information and the bottom red point suggests the main role in the attack.
Figure 3: The 3D visualization of the original node sequence (a) and sorted sequence according to a suspicious
node list (b). The three axes are two node sequences and one time series. A blue to red colormap is used in
the rendering according to the statistical neighbor relationships between two nodes, with red indicating high
connections. A label function is used to inquire the node sequence numbers.
Figure 4: (a)(b) 2D and 3D views from the original node sequence. (c)(d) An obvious pattern, located at
the left bottom corner, is revealed and may indicate a direct Sybil attack. Combined statistical neighbor
relationship (left top) and similarity information (right bottom) are used in both 2D and 3D views.
Figure 5: A 2D pattern is continuously zoomed out from (a) to (c), noticing that (c) preserves more significant
patterns than general zoom out result (d).
02468101214 and up
Shortest path between two nodes (in hop)
CDF value
legitimate nodes
under direct Sybil attack
under indirect Sybil attack
Figure 6: Abnormal patterns of Sybil attacks: dis-
tribution of the shortest path length between node
Sybil nodes based on the similarity of their neighbor rela-
tions. The detection of indirect Sybil attacks will be dis-
cussed in section 6.1.3.
Let us consider a pair of fake identities under direct Sybil
attacks. Since their neighbor discovery packets are trans-
mitted by the same physical device, the group of legitimate
nodes that can receive the packets and list them as neighbors
are almost the same. On the contrary, a pair of legitimate
nodes will not have this similarity when each of them moves
Based on the collected network topology information, we
can calculate the similarity of the neighbor relations between
two suspicious nodes s1and s2. If the neighbor set of a node
s1in a regenerated network topology Gis represented as
s1, a normalized value to describe the similarity between
the neighbor sets of two nodes can be calculated as:
The controller can estimate the similarity of the neighbor
relations between two nodes in each regenerated network
topology. Since a Sybil node can dynamically switch its at-
tached physical device, different pairs of nodes may demon-
strate the similarity in different periods of the network life-
time. For example, Figure 7 illustrates the scenario when
s2switches from node s1to s3at round 11. As a compar-
ison, we also present the similarity value between a pair of
legitimate nodes.
0 2 4 6 8 10 12 14 16 18 20
Index of regenerated network topology
Similarity of neighbors
similarity b/w s1 s2
similarity b/w s3 s2
legitimate nodes
Figure 7: Abnormal patterns of direct Sybil attacks:
similarity of neighbor relations.
6.1.3 Locating Anchor Nodes for Indirect Sybils
In an indirect Sybil attack, only through the malicious
node that claims to have paths to the fake identities can
other legitimate members reach them. Therefore, if the net-
work topology is viewed as a graph, this “anchor” node is a
cut point in it: removing this node and the associated links
will disconnect the graph.
Based on this observation, we examine the frequency that
a node is a cut point in the topology. Since multiple fake
identities can attach to the same physical device in an in-
direct Sybil attack, a special method must be adopted to
mitigate the impacts on detection accuracy when the attack-
ers rotate the identities of the “anchor” node. Therefore, we
will also count the frequency that a node is isolated from the
majority of the network when the anchor node is removed.
Figure 8 illustrates an example of indirect Sybil attacks and
the accumulated count values for each node. The differences
between the four fake identities and the legitimate nodes can
be easily identified.
6.2 System Design
To provide a robust Sybil attack monitoring and detection
tool, we have developed a detection algorithm and a topol-
ogy visualization approach as two essential components of
our security system. The integration of these two compo-
nents can benefit each other from multiple aspects. Our ba-
sic idea is to use the visualization component to intuitively
understand the network topology and security results. This
also allows users to adjust and interact with the network
information according to their expertise. The security com-
Figure 9: Our system interface is composed of a 3D view (a), a 2D view (b), and a pattern organization
window (c). Both 2D and 3D views are provided to visualize data correlations from multiple aspects. The
pattern organization window helps users to visualize similar typed patterns and locate attackers quickly.
0 50 100 150 200 250
Identity of node
Numbr of rounds that is a cut point
Figure 8: Abnormal patterns of indirect Sybil at-
tacks: frequency to be cut points.
ponent is used as an identification and validation tool to
assist users to draw final decisions and reduce their inter-
action overhead. The following will discuss our interface
design and a case study respectively.
As shown in Figure 9, we have arranged three parallel win-
dows for visualizing the event correlations: a 3D view, a 2D
view, and a pattern organization window. The 3D view is
mainly designed for displaying neighbor relationships within
a time sequence. The 2D view alternatively displays cut
views from the volumetric topology data (neighbor relation-
ships at a time step or neighbors of a node through the time
sequence) and statistical topology matrices (node connec-
tions or similarities). We also add a pattern organization
window to store typical abnormal patterns and undecided
patterns for users to compare with. These three windows al-
low us to observe the network topology information through
multiple aspects and reveal the data correlations.
The interaction of node sequence in the 2D view is achieved
through picking and dragging into the corresponding node
group. The system automatically adjusts the node sequence
according to their topology features, such as total neigh-
bor numbers. The interaction of the 3D view is managed
from the 2D view, which is used as a transfer function. The
users can interactively adjust the 3D view through common
1D transfer functions of the connection values to reveal the
focus-of-interest and reduce the overlapping in space. As
shown in Figure 3(a), we use an easy pick function by the
movement of the mouse to display the node IDs.
We use one case study to demonstrate this interaction
process. As shown in Figure 10, the network topology data
is first collected and the 2D view shows the statistical neigh-
bor relationship in Figure 10 (a). The security component
calculates the suspicious node list and guides users to adjust
the 2D view to reveal an abnormal pattern (Figure 10 (b)
and (c)). These patterns are compared with typical direct
and indirect Sybil attack patterns and suggest that this is a
hybrid attack. The suspicious node list is sent to the secu-
rity validation component to justify node behaviors, which
can be run through the data of a longer period. The users
can make final decisions based on both visualization and se-
curity results. The located fake identities will be removed
from the network.
The proposed mechanism is examined through simulation.
The experiments are conducted in two phases. In the first
phase, we use ns2 [1] to simulate the neighbor discovery
procedures and the report of the topology to the controller.
In the second phase, the proposed mechanism tries to detect
Sybil attacks and locate the fake identities. The mobile
nodes are deployed in a square area with the edge length of
1500m. The radio range rof the nodes is 170m, and any
two nodes that have a distance shorter than rcan directly
communicate to each other.
Within the simulated area, 300 nodes are randomly and
uniformly distributed and the average degree of connectivity
is 12.0. We adopt the random trip movement model that is
Figure 10: A case study: a hybrid direct and indirect Sybil attack is discovered through abnormal pat-
terns when arranged according to the identified suspicious node list. (a) The original statistical neighbor
relationship pattern; (b) The adjusted neighbor pattern; (c) The adjusted 3D topology pattern.
proposed in [7] and the highest moving speed of the nodes is
17m/s. The controller collects the network topology every
10 seconds, which is a rough estimation of the lifetime of a
link based on the radio range rand the highest node moving
speed. In every simulation, 200 rounds of topology will be
We will investigate both direct and indirect Sybil attacks.
The number of compromised physical devices and fake iden-
tities will be described in detail in each group of experiments.
When a Sybil node switches its attached physical device, it
will temporarily leave the network and rejoin later so that
the moving speed restriction will not be violated.
Since the proposed mechanism uses the network topology
information to detect Sybil attacks, we will investigate the
impacts of the parameters such as the degree of connectivity
on the detection capability. Every data point in the follow-
ing figures represents the average value over 15 trials under
different network setups.
7.1 Direct Sybil Attacks
The proposed mechanism detects the direct Sybil attacks
in two steps: it first determines the suspicious node list
based on the distribution of the path length between two
nodes, then uses the similarity of the neighbors to locate
the Sybil pairs. This mechanism will be impacted by the
length of the duration that the neighbor relations are moni-
tored. For example, under the extreme condition, when only
one snapshot of the network topology is available, the dis-
tribution of the path length between two nodes cannot be
derived. Therefore, in this group of experiments, we exam-
ine the impacts of the length of the monitored duration on
the detection capability.
Since the frequency of network topology changes heavily
depends on the radio range rand the movement model of
the mobile nodes, we use the ratio between the radio range
and the highest moving speed as a time unit to measure the
length of the monitored duration. The experiment results on
the two steps of the mechanism are presented and discussed
The cumulative distribution function (CDF) of the path
length between two nodes is illustrated in Figure 11. For the
pair of fake identities under direct Sybil attacks, we assume
that they pretend not to be neighbors. Therefore, their dis-
direct Sybil
highest speed
1 unit time = radio range r
0 0 2 4 6 8 10 12
path length b/w two nodes (in hop)
CDF value
1 unit time
5 unit time
10 unit time
20 unit time
50 unit time
100 unit time
legitimate nodes with
different monitored durations:
Figure 11: Relationship between the distribution of
path length and the monitored duration.
tance is always two hops during the monitored duration. As
the comparison, we randomly choose ten pairs of legitimate
nodes that are two hops away at the beginning of the mon-
itored duration and the average values are illustrated in the
figure. From the figure, we find that when more than 10
time units are monitored, the differences between legitimate
nodes and fake identities can be easily identified.
The CDF values of the similarity between the neighbors
of two nodes are illustrated in Figure 12. For the pair of
Sybil nodes, their neighbors will always be the same. As
the comparison, we randomly choose ten pairs of legitimate
nodes that are neighbors at the beginning of the monitored
duration and the average values are illustrated in the figure.
When more than 5 time units are monitored, the differences
can be easily identified.
Combining the results in Figure 11 and 12, we find that
when the length of monitored duration is longer than 10 time
units, the proposed approach can locate the fake identities.
In real deployment of the mechanism, the length can be
determined based on previous simulation results or field test
results. A threshold of the differences in the CDF values can
be adopted to determine the list of suspicious nodes.
7.2 Indirect Sybil Attacks
The proposed mechanism detects the indirect Sybil at-
tacks based on the frequency that a node is a “cut point” in
direct Sybil
CDF value
10% 20% 30% 40% 50% 60% 70% 80% 90% 100
similarity of neighbors b/w a pair of nodes
1 unit
5 unit
10 unit
20 unit
50 unit
100 unit
1 unit time = radio range / highest speed
legitimate nodes with different monitored durations:
Figure 12: Relationship between similarity of neigh-
bors and the monitored duration.
Indirect Sybil
legitimate nodes
0 2 4 6 8 10 12 14 16
Average degree of connectivity
# of rounds that is a cut point
Figure 13: Relationship between the frequency to
be a cut point and the degree of connectivity.
the network. This feature, however, is also impacted by sev-
eral network parameters. In this group of experiments, we
investigate the impacts of the average degree of connectivity
on the detection capability of the proposed mechanism. We
adjust the degree of connectivity by altering the radio range.
The simulation results are presented in Figure 13.
For an indirect Sybil node, it will always be a cut point or
in the isolated subnetwork when the cut point is removed.
On the contrary, the frequency of a legitimate node will
change with the degree of connectivity. As shown in Figure
13, when the average number of neighbors is very low, most
of the nodes are isolated and they are not the cut points
of the graph. As the node density increases, the nodes be-
come connected but few pairs of nodes have disjoint paths.
Therefore, more of them become the cut points. When the
network becomes denser, most of the nodes have multiple
disjoint paths among them and removing a single node will
not disconnect the graph. Therefore, the frequency will de-
crease again.
From the results in Figure 13, we find that the proposed
mechanism will have a better detection accuracy when the
mobile nodes have a relatively large degree of connectivity.
The detection of Sybil attacks in sparse networks remains
an open problem and will be investigated in the future work.
In this paper, we propose an approach for detecting Sybil
attack in wireless networks, which is a particular harmful
attack for many network functions. Our approach concen-
trates on visualizing, organizing, and detecting significant
abnormal patterns from network topology information. We
have designed security methods to locate suspicious nodes
and validate their behaviors using topology geometry infor-
mation. By integrating these intelligent algorithms, a user-
friendly visualization method is designed to reveal mean-
ingful event correlations from the network topology. This
approach allows users to monitor and detect simultaneous
direct and indirect Sybil attacks effectively.
Because of the popularity of network topology informa-
tion, our approach can be expanded to a common intrusion
detection tool for many applications and attack types. In
this paper, we integrate security and visualization methods
to provide a robust Sybil attack monitor and detection tool.
The intelligent detection component significantly accelerates
and simplifies the user interaction in the system; while vi-
sualization component increases the accuracy and tolerance
of the proposed security algorithm. This allows users to
detect more complex attack scenarios than single security
In the future, we will design and perform a systematic user
study for the proposed approach and incorporate the eval-
uation results to further simplify the system interface and
user interaction. We will investigate methods to effectively
distinguish topology patterns from group movements and
malicious attacks. We are also interested in exploring de-
tection methods for non-simultaneous Sybil attacks, which
may create more complex scenarios when conducted by mul-
tiple attacker groups in a longer time duration. Finally, since
topology visualization and organization is a common prob-
lem for network security, we will extend the proposed ap-
proach to detect more kinds of intrusions for administration
and application purposes.
The authors would like to thank the anonymous reviewers
for their valuable comments. This research is supported
in part by KU New Faculty General Research Fund and
Department of Energy under Award DE-FG02-06ER25733.
[1] Proceedings of IEC Wor ks hop on Internet Simulatio ns
with the NS simulator, 2000.
[2] K. Abdullah, C. Lee, G. Conti, J. Copeland, and
J. Stasko. IDS RainStorm: Visualizing IDS Alarms. In
Proc. of VizSEC, 2005.
[3] F. Anjum, D. Subhadrabandhu, and S. Sarkar.
Signature based Intrusion Detection for Wireless
Ad-Hoc Networks: A Comparative study of various
routing protocols. In Proc. of VTC, 2003.
[4] R.Ball,G.Fink,A.Rathi,S.Shah,andC.North.
Home-Centric Visualization of Network Traffic for
Security Administration. In Proc.ofACM
VizSEC/DMSEC, 2004.
[5] R. Bazzi and G. Konjevod. On the establishment of
distinct identities in overlay networks. In PODC ’05:
Proceedings of the twent y- fourth annual ACM
SIGACT-SIGOPS symposium on Principles of
distributed computing, pages 312–320, 2005.
[6] V. Bhuse and A. Gupta. Anomaly intrusion detection
in wireless sensor networks. Journal of High Speed
Networks, 1(15), 2006.
[7] J.-Y. Le Boudec and M. Vojnovic. Perfect Simulation
and Stationarity of a Class of Mobility Models. In
Proc. of IEEE Infocom, 2005.
[8] M. Castro, P. Druschel, A. Ganesh, A. Rowstron, and
D. S. Wallach. Secure routing for structured
peer-to-peer overlay networks. In OSDI ’02:
Proceedings of the 5th sym pos ium on Operating
systems design and implementation, pages 299–314,
[9] H. Chan, A. Perrig, and D. Song. Random key
predistribution schemes for sensor networks. In Proc.
of IEEE Symposium on Security and Privacy, 2003.
[10] A. Cheng and E. Friedman. Sybilproof reputation
mechanisms. In P2PECON ’05: Proceedin g of t he
2005 ACM SIGCOMM workshop on Economics of
peer-to-peer systems, pages 128–132, 2005.
[11] J. Douceur. The Sybil Attack. In IPTPS ’01: Revised
Papers from the First International Workshop on
Peer-to-Peer Systems, pages 251–260, 2002.
[12] W. Du, J. Deng, Y. S. Han, and P. K. Varshney. A
pairwise key pre-distribution scheme for wireless
sensor networks. In Proc. of ACM CCS, pages 42–51,
[13] L. Eschenauer and V. D. Gligor. A key-management
scheme for distributed sensor networks. In Proceedings
of ACM CCS, pages 41–47, 2002.
[14] W. Fan, M. Miller, S. Stolfo, W. Lee, and P. Chan.
Using Artificial Anomalies to Detect Unknown and
Known Network Intrusions. Know ledge and
Information Systems, 6(5), 2004.
[15] G. Fink, P. Muessig, and C. North. Visual Correlation
of Host Processes and Traffic. In Proc.ofVizSEC,
[16] Y. Fu, J. Chase, B. Chun, S. Schwab, and A. Vahdat.
SHARP: an architecture for secure resource peering.
In SOSP ’03: P roceedi ngs of the nineteenth ACM
symposium on Operating systems principles, pages
133–148, 2003.
[17] M. Gerharz, C. de Waal, M. Frank, and P. Martini.
Link Stability in Mobile Wireless Ad Hoc Networks.
In Proceedings of the IEE E Co nf erence on Local
Computer Networks (LCN), pages 30–39, 2002.
[18] P. Golle, D. Greene, and J. Staddon. Detecting and
correcting malicious data in VANETs. In VANET ’04:
Proceedings of the 1st ACM internationa l wo rkshop on
Vehicular ad hoc networks, pages 29–37, 2004.
[19] J. Goodall, P. Rheingans, W. Lutters, and
A. Komlodi. Preserving the Big Picture: Visual
Network Traffic Analysis with TNV. In Proc. of
VizSEC, 2005.
[20] J. Hall, M. Barbeau, and E. Kranakis. Using mobility
profiles for anomaly based intrusion detection in
mobile networks. In Proc. of Wireless and Mobile
Security Workshop, 2005.
[21] W. J. Hsu, K. Merchant, H. Shu, C. Hsu, and
A. Helmy. Weighted waypoint mobility model and its
impact on ad hoc networks. SIGMOBILE Mob.
Comput. Commun. Rev., 9(1):59–63, 2005.
[22] A. Komlodi, P. Rheingans, U. Ayachit, J. Goodall,
and A. Joshi. A User-Centered Look at Glyph-Based
Security Visualization. In Proc.ofVizSEC, 2005.
[23] K. Lakkaraju, W. Yurcik, A. Lee, R. Bearavolu, Y. Li,
and X. Yin. NVisionIP: NetFlow Visualizations of
System State for Security Situational Awareness. In
Proceedings of ACM VizSEC/DMSEC, 2004.
[24] L. Lazos and R. Poovendran. SeRLoc: Robust
localization for wireless sensor networks. ACM Trans.
Sen. Netw., 1(1):73–100, 2005.
[25] D. Liu, P. Ning, and R. Li. Establishing Pairwise Keys
in Distributed Sensor Networks. ACM Transactions
on Information and System Security, 8(1):41–77, 2005.
[26] Y. Livnat, J. Agutter, S. Moon, R. Erbacher, and
S. Foresti. A Visualization Paradigm for Network
Intrusion Detection. In Proceedi ng s of the IEEE
Information Asssurance Workshop, pages 92-99, 2005.
[27] B. S. Mano j, R. Ananthapadmanabha, and
C. Siva Ram Murthy. Multi-hop cellular networks:
architecture and protocols for best-effort and real-time
communication. J. Parallel Distrib. Comput.,
65(6):767–791, 2005.
[28] J. McPherson, K. Ma, P. Krystosk, T. Bartoletti, and
M. Christensen. PortVis: A Tool for Port-Based
Detection of Security Events. In Proc. of ACM
VizSEC/DMSEC, 2004.
[29] C. Muelder, K. Ma, and T. Bartoletti. A Visualization
Methodology for Characterization of Network Scans.
In Proc.ofVizSEC, 2005.
[30] J. Newsome, E. Shi, D. Song, and A. Perrig. The sybil
attack in sensor networks: analysis & defenses. In
IPSN ’04: P roceedi ngs of the third internati on al
symposium on Information processing in sensor
networks, pages 259–268, 2004.
[31] D. Rafiei and S. Curial. Effectively Visualizing Large
Networks Through Sampling. In Proc. of IEEE
Visualization, 2005.
[32] P. Ren, Y. Gao, Z. Li, Y. Chen, and B. Watson.
IDGraphs: Intrusion Detection and Analysis Using
Histographs. In Proc.ofVizSEC, 2005.
[33] N. Saxena, G. Tsudik, and J. Yi. Admission control in
Peer-to-Peer: design and performance evaluation. In
SASN ’03: P roceedi ngs of the 1st AC M wo rkshop on
Security of ad hoc and sensor networks, pages
104–113, 2003.
[34] S. Vasudevan, B. DeCleene, N. Immerman, J. Kurose,
and D. Towsley. Secure Leader Election Algorithms
for Wireless Ad Hoc Networks. In Proc. of IEEE
DARPA Information Survivability Conference and
Exposition (DISCEX), 2003.
[35] W. Yurcik. VisFlowConnect-IP: A Link-Based
Visualization of NetFlows for Security Monitoring. In
18th Annual FIRST Conference on Computer Security
Incident Handling, 2006.
[36] Y. Zhang and W. Lee. Intrusion detection in wireless
ad-hoc networks. In Proc. of ACM MobiCom, pages
275–283, 2000.
... Further approaches that tried to include visualizations only made use of static charts or graphs such as the trust network depicted by O'Donovan et al. (2007). Another proposal closely related to this work was presented by Wang and Lu (2006). Although their application scenario was the detection of Sybil attacks in wireless networks and thus different from ours, they had similar goals and ideas. ...
... Moreover, they argued that the information on the network topology have to be arranged appropriately in order to demonstrate significant features of intrusions. In contrast to our approach, Wang and Lu (2006) assumed that the links among the nodes in the network are bidirectional and thus not directed. They furthermore do not allow multiple edges and edge weights, which makes our visualization technique strongly different. ...
Conference Paper
Full-text available
Reputation systems provide a valuable method to measure the trustworthiness of sellers or the quality of products in an e-commerce environment. Due to their economic importance, reputation systems are subject to many attacks. A common problem are unfair ratings which are used to unfairly increase or decrease the reputation of an entity. Although being of high practical relevance, unfair rating attacks have only rarely been considered in literature. The few approaches that have been proposed are furthermore quite non-transparent to the user. In this work, we employ visual analytics to identify colluding digital identities. The ultimate benefit of our approach is the transparent revelation of the true reputation of an entity by interactively using both endogenous and exogenous discounting methods. We thereto introduce a generic conceptual design of a visual analytics component that is independent of the underlying reputation system. We then describe how this concept was implemented in a software prototype. Subsequently, we demonstrate its proper functioning by means of an empirical study based on two real-world datasets from eBay and Epinions. Overall, we show that our approach notably enhances transparency, bares an enormous potential and might thus lead to substantially more robust reputation systems and enhanced user experience.
... To ensure privacy-preserving node authentication, several anonymous authentication protocols [61], [62] have been proposed for VANETs. To detect the nodes which assume the identity of other nodes by launching the Sybil attack, several existing schemes [63], [64], [65] can be adopted. ...
With wireless vehicular communications, Vehicular Ad Hoc Networks (VANETs) enable numerous applications to enhance traffic safety, traffic efficiency, and driving experience. However, VANETs also impose severe security and privacy challenges which need to be thoroughly investigated. In this dissertation, we enhance the security, privacy, and applications of VANETs, by 1) designing application-driven security and privacy solutions for VANETs, and 2) designing appealing VANET applications with proper security and privacy assurance. First, the security and privacy challenges of VANETs with most application significance are identified and thoroughly investigated. With both theoretical novelty and realistic considerations, these security and privacy schemes are especially appealing to VANETs. Specifically, multi-hop communications in VANETs suffer from packet dropping, packet tampering, and communication failures which have not been satisfyingly tackled in literature. Thus, a lightweight reliable and faithful data packet relaying framework (LEAPER) is proposed to ensure reliable and trustworthy multi-hop communications by enhancing the cooperation of neighboring nodes. Message verification, including both content and signature verification, generally is computation-extensive and incurs severe scalability issues to each node. The resource-aware message verification (RAMV) scheme is proposed to ensure resource-aware, secure, and application-friendly message verification in VANETs. On the other hand, to make VANETs acceptable to the privacy-sensitive users, the identity and location privacy of each node should be properly protected. To this end, a joint privacy and reputation assurance (JPRA) scheme is proposed to synergistically support privacy protection and reputation management by reconciling their inherent conflicting requirements. Besides, the privacy implications of short-time certificates are thoroughly investigated in a short-time certificates-based privacy protection (STCP2) scheme, to make privacy protection in VANETs feasible with short-time certificates. Secondly, three novel solutions, namely VANET-based ambient ad dissemination (VAAD), general-purpose automatic survey (GPAS), and VehicleView, are proposed to support the appealing value-added applications based on VANETs. These solutions all follow practical application models, and an incentive-centered architecture is proposed for each solution to balance the conflicting requirements of the involved entities. Besides, the critical security and privacy challenges of these applications are investigated and addressed with novel solutions. Thus, with proper security and privacy assurance, these solutions show great application significance and economic potentials to VANETs. Thus, by enhancing the security, privacy, and applications of VANETs, this dissertation fills the gap between the existing theoretic research and the realistic implementation of VANETs, facilitating the realistic deployment of VANETs.
... A controller node is a node that collects network data from each sensor node and then it proceeds to data integration, process, and visualisation [17]. In the context of this paper it is considered that the sink node plays the role of the controller node. ...
Conference Paper
Internet of Things (IoT) have emerged as a valuable, flexible, and interoperable network of devices, objects, items, and electronics. Fuelled by recent advances in networking, communications, computation, software, and hardware technologies, IoT has stepped out of its infancy and is considered as the next breakthrough technology in transforming the Internet into a fully integrated Future Internet.Wireless Sensor Networks (WSNs) are utilized by IoT to collect, exchange, and deliver data remotely leveraging the potential of IoT in practical applications and services. However, delivering data remotely might be threatened by various and serious security attacks. This work focuses on developing a visual-assisted tool for exposing security threats in IP-enabled WSNs. The proposed tool, called EyeSim, is a human interactive visual-based anomaly detection system that is capable of monitoring and promptly alerting for the presence of wormhole links. In addition, it is capable of indicating the malicious nodes that form the wormhole link. EyeSim may expose adversaries by conducting cognitive network data analysis based on dynamic routing information. The efficacy of EyeSim is assessed in terms of detection accuracy. The simulation results show that EyeSim has the capabilities to accurately detect multiple wormhole attacks in real-time.
... Though an area in the infancy stage, two visualization approaches have already been proposed. In Wang and Lu (2006), the authors use multiple 2D and 3D views enabling the user to observe the network topology information through multiple aspects and reveal data correlations relevant to Sybil attacks. Simulation studies showed that the proposed mechanism can effectively identify both direct and indirect Sybil attacks. ...
Security is becoming a major concern for many mission-critical applications wireless sensor networks (WSNs) are envisaged to support. The inherently vulnerable characteristics of WSNs appoint them susceptible to various types of attacks. This work restrains its focus on how to defend against a particularly harmful form of attack, the Sybil attack. Sybil attacks can severely deteriorate the network performance and compromise the security by disrupting many networking protocols. This paper presents a rule-based anomaly detection system, called RADS, which monitors and timely detects Sybil attacks in large-scale WSNs. At its core, the proposed expert system relies on an ultra-wideband (UWB) ranging-based detection algorithm that operates in a distributed manner requiring no cooperation or information sharing between the sensor nodes in order to perform the anomaly detection tasks. The feasibility of the proposed approach is proven analytically, while the performance of RADS in exposing Sybil attacks is extensively assessed both mathematically and numerically. The obtained results demonstrate that RADS achieves high detection accuracy and low false alarm rate appointing it a promising ADS candidate for this class of wireless networks.
... With no wormhole present, the network topology is flat, while a wormhole would be seen as a string pulling different ends of the network together. In a subsequent research, the same authors proposed an effective approach for monitoring and detecting Sybil attacks by integrating network security and visualization methods [45]. ...
Conference Paper
Full-text available
Security is becoming a major concern for many mission-critical applications wireless sensor networks (WSNs) are envisaged to support. This is because WSNs are susceptible to various types of attacks or to node compromises that exploit known and unknown vulnerabilities of protocols, software and hardware, and threaten the security, integrity, authenticity, and availability of data that resides in these networked systems. While various security mechanisms have been proposed for these networks dealing with either MAC layer or network layer security issues, or key management problems, the security benefits that can be obtained from an upper visualization layer have not been adequately considered in their design. In this paper, we explore the issues and concerns surrounding the application of visual analysis for wireless sensor network security purposes. This paper focuses on several distinct advantages information visualization and visual analytics can offer in the security domain. In addition, this paper reviews security visualization tools that are available to network security analysts. Finally, it concludes by identifying challenges for this new area of research.
... With the ever increasing data size and complexity, many visualization approaches have been developed to improve the processing of a large amount of network data including traffic patterns, network flows and logs [4, 23, 30]. Because of the importance of the network topology, it has been used to help enforce Internet and wireless network security in multiple network visualization mechanisms [3, 15, 29] . For example , topologies have been visualized using graph drawing or parallel coordinates [1, 2] to show interesting patterns of malicious attacks. ...
We present a robust intrusion detection approach for wireless networks based on a new multi-matrix visualization method with a set of pattern generation, evaluation, organization and interaction functions. Our approach concentrates on assisting users to analyze statistical network topology patterns that could expose significant attack features. Specifically, we investigate Sybil attacks that have severe impacts on the fundamental operations of wireless networks. We have analyzed the features of network topologies under various Sybil attacks and, consequently, designed several matrix reordering algorithms to generate statistical patterns. These topology patterns are automatically evaluated and classified through the measured structural similarities to the signature attack patterns. We have also designed a new time-series analysis method to identify attack durations with a time histogram generation and an automatic segmentation method. To handle complex Sybil attacks, we have integrated our pattern generation, evaluation and organization methods to construct a prototype detection system, in which specialized interaction functions are provided to assist the analysis and comparison of network data. Simulation results show that this approach can effectively locate Sybil attacks under different combinations of network parameters. Our multi-matrix visualization method provides a flexible framework to handle the intricacies and implications from building a complex visual analytics system, which can be extended to defend against a wide range of attacks.
This chapter treats an interactive visual analysis tool called PCTT, Parallel Coordinates Version of Time-tunnel, for multidimensional data and multi-attributes data. Especially, in this chapter, the author introduces the combinatorial use of PCTT and 2Dto2D visualization functionality for visual analytics of network data. 2Dto2D visualization functionality displays multiple lines those represent four-dimensional (four attributes) data drawn from one (2D, two attributes) plane to the other (2D, two attributes) plane in a 3D space. Network attacks like the intrusion have a certain access pattern strongly related to the four attributes of IP packet data, i.e., source IP, destination IP, source Port, and destination Port. So, 2Dto2D visualization is useful for detecting such access patterns. Although it is possible to investigate access patterns of network attacks at the attributes level of IP packets using 2Dto2D visualization functionality, statistical analysis is also necessary to find out suspicious periods of time that seem to be attacked. This is regarded as the macro level visual analytics and the former is regarded as the micro level visual analytics. In this chapter, the author also introduces such combinatorial use of PCTT for macro level to micro level visual analytics of network data as an example of multidimensional data. Furthermore, the author introduces other visual analytics example about sensor data to clarify the usefulness of PCTT.
Network and service management encompasses a set of activities, methods, procedures, and tools, whose ultimate goal is to guarantee the proper functioning of a networked system. Computational tools are essential to help network administrators in their daily tasks, and information visualization techniques are of great value in such context. In essence, information visualization techniques associated to visual analytics aim at facilitating the tasks of network administrators in the process of monitoring and maintaining the network health. This paper surveys the use of information visualization techniques as a tool to support the network and service management process. Through a systematic literature review (SLR), we provide a historical overview and discuss the current state of the art in the field. We present a classification of 285 articles and papers from 1985 to 2013, according to an information visualization taxonomy, as well as a network and service management taxonomy. Finally, we point out future research directions and opportunities regarding the use of information visualization in network and service management.
Conference Paper
Full-text available
Security concerns are a major deterrent in many applications wireless sensor networks are envisaged to support. To date, various security mechanisms have been proposed for these networks dealing with either Medium Access Control (MAC) layer or network layer security issues, or key management problems. Security visualization is the latest weapon that has been added in the arsenal of a security officer who is tasked with detecting network anomalies by analyzing large amounts of audit data. This paper proposes a novel security visualization system for analyzing and detecting complex patterns of sensor network attacks, called SRNET. Both selective forwarding and jamming attacks are identified through visualizing and analyzing network traffic data on multiple coordinated views, namely the multidimensional crossed view, the crossed view perspective, and the track area view. Through simulations, we demonstrate that SRNET is able to help detect and further identify the root cause of the aforementioned sensor network attacks.
Conference Paper
This paper treats network data visualization using Parallel Coordinates version of Time-tunnel (PCTT) for intrusion detection. Originally, Time-tunnel is a multidimensional data visualization tool and its Parallel Coordinates version provides the functionality of Parallel Coordinates visualization. It can be used for the visualization of network data because IP packet data have many attributes and such multiple attribute data can be visualized using Parallel Coordinates. In this paper, the authors propose the combinatorial use of PCTT and 2Dto2D visualization functionality for the intrusion detection. 2Dto2D visualization functionality, whose concept is originally derived from nicter Cube, displays multiple lines those represent four dimensional (four attributes) data drawn from one (2D of two attributes) plane to the other (2D of the other two attributes) plane in a 3D space. This 2Dto2D visualization functionality was introduced to PCTT. Network attacks have a certain access pattern strongly related to the four attributes of IP packet data, i.e., source IP, destination IP, source Port, and destination Port. So, 2Dto2D visualization is useful for detecting such access patterns. In this paper, the authors show several network attack patterns visualized using PCTT with 2Dto2D visualization as examples for the intrusion detection.
Full-text available
To realistically evaluate performance of ad hoc networks we propose a generic framework called the Weighted Way Point (WWP) mobility model. WWP model captures preferences in choosing destinations of pedestrian mobility patterns in a campus environment. We estimate the parameters of this model using mobility survey data for the USC campus. We further compare WWP model with widely used Random Waypoint (RWP) model and demonstrate that in the WWP model mobile nodes display uneven (clustering), time-varying spatial distribution. WWP model is also less mobile than RWP model with typical parameter settings. The clustering effect can cause lower success rate of route discovery in ad hoc networks.
Many distributed monitoring applications of Wireless Sensor Networks (WSNs) require the location information of a sensor node. In this article, we address the problem of enabling nodes of Wireless Sensor Networks to determine their location in an untrusted environment, known as the secure localization problem. We propose a novel range-independent localization algorithm called SeRLoc that is well suited to a resource constrained environment such as a WSN. SeRLoc is a distributed algorithm based on a two-tier network architecture that allows sensors to passively determine their location without interacting with other sensors. We show that SeRLoc is robust against known attacks on a WSNs such as the wormhole attack , the Sybil attack , and compromise of network entities and analytically compute the probability of success for each attack. We also compare the performance of SeRLoc with state-of-the-art range-independent localization schemes and show that SeRLoc has better performance.
Conference Paper
Security is important for many sensor network applications. A particularly harmful attack against sensor and ad hoc networks is known as the Sybil attack [6], where a node illegitimately claims multiple identities. This paper systematically analyzes the threat posed by the Sybil attack to wireless sensor networks. We demonstrate that the attack can be exceedingly detrimental to many important functions of the sensor network such as routing, resource allocation, misbehavior detection, etc. We establish a classification of different types of the Sybil attack, which enables us to better understand the threats posed by each type, and better design countermeasures against each type. We then propose several novel techniques to defend against the Sybil attack, and analyze their effectiveness quantitatively.