HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks
Ikbel Daly
1
, Faouzi Zarai
1
, M. S. Obaidat
2
, K.F. Hsiao
3
and Lotfi Kamoun
1
1
LETI Laboratory, University of Sfax, Sfax, Tunisia
2
Department of Computer and Information Science, Fordham University, NY, U.S.A.
3
Dep. of Information Management, Ming-Chuan University, Taoyuan County 333, Taiwan
Keywords: Heterogeneous Wireless Network, Security, Hybrid IP Traceback, Marking Packet, Logging Packet, Denial
of Service Attack.
Abstract: The Denial of Service attack becomes increasingly vulnerable with heterogeneous wireless networks. Thus,
it is fundamental to identify the source of attack by the execution of an IP traceback technique. There are
two major categories: packet marking and packet logging. The first approach moderates the problem of
overhead, but requires a large amount of packets to reconstruct the attack path. In packet logging, saving
packets in digest tables enables the identification of attack source through a single packet but necessitates a
huge storage space. In this paper, we propose a novel Hybrid IP Traceback for Heterogeneous wireless
networks, which is called HITH (Hybrid IP Traceback for Heterogeneous wireless network). Our solution
presents a precise IP traceback method with low overhead storage and improved accuracy. Indeed, the
mathematical analysis and the comparison with existing solutions prove the capacity to trace a single IP
packet while reducing storage overhead and data access time.
1 INTRODUCTION
The infrastructure of communication becomes
increasingly heterogeneous following the integration
of several technologies in order to meet the growing
needs of the users’ community. This heterogeneity
has several mechanisms and techniques that are
characterized by a specific composition and precise
services and features.
To ensure this connectivity, we use several
methods and strategies. Indeed, interoperability
depends on the network topology, traffic pattern,
interference, etc. According to Ren et al., 2011, the
connectivity of low-priority network component
depends on the characteristics of high-priority
component, in order to ensure the diversity of
techniques and to reduce infrastructure complexity.
In order to ensure the networks’ heterogeneity,
we must first procure the interworking between
existing technologies from the third generation (3G)
to the fifth generation (5G). Figure 1 illustrates an
example of Heterogeneous Wireless Networks
(HWN) which integrates two technologies; LTE
(Long Term Evolution) and Wireless Mesh Network
(WMN).
In heterogeneous wireless network, characterized
by the integration of multiple network types and
wireless technologies, the problem of security is
becoming more critical. Consequently, this issue
may deserve urgent and effective solutions to ensure
secure communications and maintain the confidence
between network equipment
.
DoS (Denial of Service) (Houle and Weaver,
2001) and DDoS (Distributed DoS) (Phatak et al.,
2013) attacks present the most spread vulnerability
in the Internet, which can affect multiple targets,
even critical, in a very short duration.
To prevent networks against these attacks, we
can apply some traceability techniques, known by IP
Traceback. This mechanism allows tracking packets
and ensures the identification of the real source of
attack.
The remaining part of this paper is organized as
follows. In section II, we introduce some related
works, which treat the existing IP traceback
techniques. Section III describes our proposed
hybrid IP traceback approach for Heterogeneous
wireless Networks in details. Section IV evaluates
the performance of our approach. Finally, Section V
gives the conclusion.
84
Daly, I., Zarai, F., Obaidat, M., Hsiao, K. and Kamoun, L.
HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks.
DOI: 10.5220/0006042500840093
In Proceedings of the 13th International Joint Conference on e-Business and Telecommunications (ICETE 2016) - Volume 6: WINSYS, pages 84-93
ISBN: 978-989-758-196-0
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
Figure 1: Example of Heterogeneous Wireless Networks
topology (LTE-Mesh).
2 RELATED WORKS
The knowledge of the origin of vulnerability serves
to protect the network from different types of attacks
and even those which may occur in the future.
For this reason, researchers do not stop treating
this topic by proposing new solutions and
techniques.
A. IP Traceback Techniques
Therefore, there is a variety of IP traceback methods
including primarily:
Probabilistic Packet Marking (PPM) (Savage et
al., 2001)
ICMP traceback (ITrace) (Bellovin, 2000)
Hash-based IP traceback (Packet logging)
(Sager, 1998)
Hybrid IP traceback (Choi and Dai, 2004)
1) Probabilistic Packet Marking (PPM)
This solution is based on the idea of marking the IP
packets. This operation can be performed either by
using some bits in the IP packet header, or by
generating a new packet based on router’s address or
a part of its address. This approach has been
improved in several works such as Sattari et al.,
2010 and Yaar et al., 2005. The choice of marking
packets depends on a probability value in the order
of 0.04 (Savage et al., 2001).
The major constraint to this method is the need
to collect a large amount of packets in order to
identify the source of attack. But thanks to works
done in Song and Perrig, 2001 as the number was
reduced to 1000 packets. In addition, this method
uses two additional functions; marking and path
reconstruction. This addition, which requires the
handling of packets, brings additional work with
processor and therefore causes a processing
overhead.
Furthermore, the packet can be transformed or
modified in order to falsify the traceback results. For
this reason, PPM supports different packets
transformations within an environment without
reflectors to ensure the sureness of IP traceback
procedure results. Thanks to its efficiency, PPM can
identify the source of majority of DoS and DDoS.
A variety of packets marking techniques have
been proposed; we metion as examples the technical
Probabilistic Packet Marking (PPM) (Song and
Perrig, 2001) and Proactive Signaling Architecture
for IP Traceback (PSAT) (Fadlallah and
Serhrouchni, 2006).
2) ICMP traceback (ITrace)
This approach is based on the concept of adding a
message named "ICMP traceback message" or
"ITrace". This idea appeared with Bellovin, 2000.
First, the router selects a packet from 20000
forwarding packets. Then, it generates a packet
containing the ITrace message. Finally, this
information will be forwarded to the same
destination.
In order to handle this method, routers must be
improved with the aim of designing new services.
The scalability is assured because this addition does
not affect the operation of other network equipment.
To execute this approach, we need a huge
number of packets for the construction of path and
the implementation of some additional functions
such as the generation of ITrace message.
Concerning memory storage, ICMP traceback
method does not require a large quantity of
additional memory. As with PPM, this method
supports the transformation and modification of
packets except with reflectors.
3) Hash-based IP Traceback (Packet Logging)
This technique is known as SPIE (Source Path
Isolation Engine) (Sager, 1998). The idea is based
on saving parts of packets as a digest or a packet
signature. The main drawback for this IP traceback
solution is the huge amount of resources reserved for
storing packet digests. In order to reduce the
HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks
85
required storage space, routers operate a space-
efficient data structure technique called Bloom filter
(Bloom, 1970).
Concerning the network infrastructure, the
packet Logging approach requires the addition of
new equipment to ensure the storage packets
procedure and to upgrade routers software.
Consequently, packet logging approach does not
provide the scalability needed in heterogeneous
wireless network.
On the other hand, the primary advantage of this
technique is the ability to establish the attack path
through a single packet. Moreover, even with packet
modification or transformation, packet logging is
able to provide reliable and effective results for most
DoS and DDoS attacks.
4) Hybrid IP Traceback (HIPT)
This method is based on two techniques: Packet
Marking and Packet Logging. This alliance benefits
from the advantages provided by each approach.
Indeed, hybrid IP traceback is characterized by the
reduction of reserved resources for packet marking
and the utilization of a small number of packets to
identify the attack source by using saved digests
(Yang and Yang, 2012; Sai Priyanka and Srihari
Rao, 2013; Gong and Sarac, 2005). Despite the
multiplicity of advantages, this combination does not
eliminate the drawbacks brought by the two used
techniques.
B. Evaluation of IP Traceback Techniques
This subsection involves a comparative study
between the different IP traceback techniques
mentioned in the previous subsection (A). Based on
Murugesan et al., 2014, a representative method in
each category was evaluated. Indeed, the proposed
scheme in Goodrich, 2008 was selected to represent
the Probabilistic Packet Marking technique. The
work in Bellovin, 2000 is chosen as a representative
ICMP based traceback technique, SPIE (Snoeren et
al., 2002) represents the Packet Logging method and
RIHT (Yang and Yang, 2012) is chosen under
Hybrid Traceback scheme.
The comparative study of IP traceback
techniques is based on the following evaluation
metrics:
ISP Involvement: In order to trace the attack
route, in some IP traceback schemes, we call for
ISP (Internet Service Provider) intervention to
provide some additional information aimed to
identify the source(s) of attack.
Number of attack packets: The number of
packets required to determine the source of
attack.
Processing overhead: It represents the additional
processing related to the traceback scheme. It
can take place in two levels, either in ISP’s
devices or in the part of victim.
Protection: To address this matter, we should
consider the non-belonging of equipment to its
network if this device becomes subverted.
Scalability: In some traceback schemes, we need
to add some new devices in the network. Such
complementary equipment may require an
independent configuration or with others
devices. Thus, minimizing the dependency
improves the scalability of the scheme.
Memory requirement: This represents the
quantity of additional storage needed at the
network equipment. This metric can be
computed in two levels: at the network
components (routers and servers) and at the
victim levels.
Accuracy : This metric evaluates the precision of
IP traceback method by defining the false
positive and the false negative parameters.
Knowledge of Network : This decides if the IP
traceback scheme requires a prior knowledge
about the topology of studied network.
Ability to handle major DDoS attacks: This
metric proves the capacity of the scheme to
perform the traceback of DDoS attacks under
rigorous circumstances such as IP spoofing, and
manipulation of reflectors (Mölsä, 2005).
Based on these selected evaluation metrics, Table I
illustrates a comparison between different IP
traceback techniques.
3 PROPOSED SOLUTION
After studying the different IP traceback approaches,
we can conclude that hybrid method is the most
suitable approach for the studied network, which is
characterized by the variety and the heterogeneity of
technologies. On the other hand, this IP traceback
technique has proved its efficiency and feasibility.
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
86
Table 1: Comparison of IP traceback techniques.
Evaluation
Metrics
PPM ITrace
Packet
logging
Hybrid
Scheme
ISP
Involvement
Fair Good Poor
(Huge
memory
requireme
nt)
Fair
Number of
attack
packets
Large
number of
packet
Number of
ICMP
messages
and huge
number of
attack
packets
One
packet
One
packet
Processing
overhead
(Router)
Medium Low High Low
Protection
Good Good and
practically
feasible
Poor Poor
Scalability
Poor Good Fair Fair
Memory
requirement
(Network)
Not
required
Not
required
Very High
Low
Memory
requirement
(Victim)
Very High
Medium Not
required
Not
required
Accuracy
Medium
(Huge
false
positive
rate in
case of
DDoS
attack)
Good for
less
numbers
of
attackers
Medium
with high
false
positive
and false
negative
High (less
false
positive
and false
negative
rate)
Knowledge
of Network
Not
needed
(Faster
traceback
and low
false
positive if
known)
Not
needed
Not
needed
Not
needed
Ability to
handle
major DoS
attacks
DoS/DDo
S flooding
attacks
DoS/DDo
S network
layer
attacks
DoS/DDo
S flooding
attacks
DoS/DDo
S flooding
attacks
A. Main Idea
In this subsection, we describe the principle of our
proposed solution named HITH (Hybrid IP
Traceback for Heterogeneous wireless network).
This approach needs to meet several specifications
related to the studied environment and the conditions
of traceback process implementation.
HWN is characterized by a variety of wireless
networks in which each category has its own
properties. On the other hand, the proposed model
allows encompassing the notion of mobility by
studying the interworking between different
networks and the handover procedure in layer 3.
In order to ensure reliability and robustness of
our proposed IP traceback solution and in particular
sureness of the established attack paths, the
traceback model is based on a set of assumptions:
Network routers are trusted,
The attacker does not know in advance the
traceback mechanism,
The IP header can be changeable.
HITH is a hybrid IP traceback method, which
owns the packet marking capabilities and the storage
of some network information. This combination of
two traceback techniques ensures the effectiveness
of the proposed approach and the reduction of
storage overhead problem.
The principle of HITH method is based on the
identification of the first packet for each new
connection crossing a new router. Indeed, following
the receipt of a packet, the router checks a field
called “LogMark”. If it is set to 0, the router
proceeds with the logging operation. Otherwise (i.e.
the value of “LogMark” is different to 0), the router
proceeds by packet marking method.
1) Marking Procedure
In this part, we describe in details the packets
marking mechanism at routers for the studied
network. First of all, we must point out that our
proposed solution is valid for IPv4 packets. And for
IPv6 packet, more research and improvements
should be procured to HITH since this new version
includes new features that should be considered by
our method in future work.
The majority of marking approaches use the 3
fields of IP header; “Identification”, “Reserved flag”
and “Fragment offset”. Similarly, HITH uses these
three fields to perform the marking operation and
logging digests at network routers.
¾ Identification
This field is carried on 16 bits. It is useful in the case
of packets’ fragmentation, since it designates the
fragment number. A study in Stoica and Zhang,
1999 performed in the Internet environment, shows
that less than 0.25% of packets crossing the network
is fragmented. Therefore, the research community in
the field of IP traceback processes the ability to
reuse this field for other functionalities.
Indeed, a study of feasibility of this proposal and
also of incompatibility problems of that field
introduces a topic of discussion in Savage et al.,
2001. Ultimately, the obtained results confirm the
hypothesis of “Identification” field reusing.
As illustrated in Figure 2, HITH replaces the
“Identification” field by two new fields; “LogMark”
HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks
87
and “Router ID”. For “LogMark”, this first element
is carried on 2 bits and is used to define the
operation to be performed by the router either
marking or logging.
A logical combination of two bits can compose four
possible values; 00, 01, 10 and 11. This field is
initialized to (00) by the first router. Then, it
undergoes an incrementation until reaching the last
value (11). Afterwards, it returns to the first value
(00) and increments once again till the attainment of
destination. This value reveals the operation to
perform at router level.
Ver Hlen TOS Total Length
Identification
R
F
Flag
Fragment
offset
TTL Protocol Header Checksum
Source IP Address
Destination IP Address
Figure 2: Structure of IP packet Header in marking
procedure with HITH.
The second part of “Identification” field
indicates the router identifier (Router ID). This
information may be fixed to a length of 14 bits based
on Muthuprasanna et al., 2006, which illustrates a
study performed on the Internet and also the concept
of routers neighborhood. Reference Muthuprasanna
et al., 2006 asserts that the allocation of 14bits
allows identifying routers uniquely.
¾ Reserved Flag
This field is carried on one bit, which is not yet
assigned. In the domain of IP traceback, Dean et al.,
2002 proposes the use of this field to execute the
marking operation.
¾ Fragment Offset
In the IP header and following the packet
fragmentation, this field is used to specify the offset
of the fragment from the original datagram (64 bit
words). But because of the elimination of the
“Identification” field, “Fragment offset” becomes
useless. This facilitates its exploitation in favor of IP
traceback approaches Gao and Ansari, 2005 and
Gong and Sarac, 2009.
Indeed, HITH operates these two fields
(“Reserved flag” and “Fragment offset”) to denote
the identifier of the previous marking router (Last
Marking Router ID). This identifier is generated by
the network administrator for each traceback-
enabled router.
2) Logging Procedure
To pursue and rebuild the attack path and minimize
the storage space at routers, various traceback
approaches apply hash functions on the data set from
the IP packet to extract the digests.
With hash-based approach, digests are derived
from the integration of different IP header fields
(except TTL “Time To Live”, TOS “Type Of
Service” and checksum) with the first 8 bytes of
payload. These fields are still operated in the hybrid
traceback approach with the addition of a new field
named "Logging flag" which introduces the
innovation brought by this type of traceback
mechanism.
In our solution, HITH calculates the packet
digest in the same way as PPIT (Precise and
Practical IP Traceback) approach presented in Yan
et al., 2012 by integrating diverse parts of the IP
header stated in the hybrid approach excluding the
TTL field.
¾ TTL Integration in Digest
In this part, we show through an illustrative example
the utility behind the addition of TTL field in the
input of digests and subsequently in the
reconstruction of attack path. In Figure 3, we present
a sample network topology exploiting HIPT (Hybrid
IP Traceback) approach, composed of 14 routers.
The attack path is exposed through continuous red
arrows and dotted arrows show the returned path to
identify the source of attack.
Since this network adopts our hybrid IP
traceback approach, routers (R1, R7, R10 and R14)
executes packet digests storage procedure, known by
logging and the other routers (R2, R3, R4, R5, R6,
R8 R9, R11, R12 and R13) carried out the marking
procedure. Following the detection of an attack, the
traceback approach begins with the reconstruction of
the attack path by identifying the concerned routers.
Firstly, the hybrid approach recognizes the first
router R14 since it is directly related to the victim.
LogMark Router
Last Marking Router ID
2bit 14 14
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
88
Figure 3: Example of false attack path reconstruction with HIPT.
Thanks to the implementation of the logging
operation, we can identify the next router by
referring to the router identity recorded in the packet
digest. Then, the packet marking provides
information on the previous router in the field "Last
Marking Router ID" filled in place of “Fragment
offset”. On arrival at R11 router, this latter sends
queries to its neighbors (R7 and R10), which are
both included in the attack path. And since the
responses to requests will be received randomly, it
can be considered the case where the path is
continued with the router R10. Consequently, that
may falsify the construction of the attack path.
To remedy this problem, we must differentiate
between the two responses received from R7 and
R10 by a time reference which may be presented in
the IP header by the TTL field. In this case, by
comparing the TTL fields from digests of R7 and
R10 packets, we find that the TTL value of R7 is
lower than that of R10. Thus, this comparison
proves that the attack packet has traversed R10
before reaching R7. This scenario can be executed
only if we include the TTL field in the logging
process as a part of packet digest.
¾ Digest Table
The HITH approach saves packet digests in a digest
table implemented with Bloom Filter method (Song
and Perrig, 2001), which reduces the storage
overhead and makes the storage procedure more
convenient. As illustrated in Figure 4, Bloom Filter
uses (k) hash functions to calculate (k) distinct
packet digests, each one is composed of (n) bits.
These results are indexed in a list of (2
n
) bits,
initialized to 0.
To ensure the rapidity of the reconstruction
process of the attack path, we have adopted the idea
of multiplying digest tables between neighbors,
which is exposed in HIT (Hybrid single-packet IP
Traceback) approach (Gong and Sarac, 2008). The
routers in this approach are characterized by the
management of different digest tables at the same
time. Each table is associated with one or more
routers identities (Router ID) used in the packet
marking procedure.
After extracting the packet digest, this latter will
be recorded in digest tables, which are necessarily
associated with the router identity supported by the
concerned IP packet. Indeed, when a router decides
to execute the logging operation, it looks first at the
router identity on the contemplated packet (Router
ID). Then, it stores the resulting digest in the
corresponding table.
Figure 4: Bloom Filter procedure.
HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks
89
In this context, we note that the existence of a
given table depends on the router vicinity, which can
find a table with its identity in each of its neighbors.
Therefore, packets from different routers can
undergo the logging operation and be stored in
different tables simultaneously. This may cause the
reduction of access time to digest tables because this
parameter is no longer proportional to the packet
arrival rate, but also to the maximum rate of arrival
packets from the whole neighborhood router.
Despite this reduction, the results of the adopted
traceback mechanism remain depending on the
capacity of each router essentially the memory
access time. Indeed, if a router is characterized by a
bad memory access time, it cannot take advantage of
traceback mechanism benefits. To remedy this
problem, this category of low-speed router profits
from access to digest table associated with all
neighboring routers, carrying all identities instead of
giving each router its own table.
In case of table saturation, packet digests will be
archived for a period of time that depends on the
configuration and the requirements of the studied
network. Thanks to Bloom Filter procedure, the
storage overhead of these tables for each router is
still negligible and does not affect the network
quality of service or the performance of the adopted
traceback mechanism.
B. Traceback Process
In order to clarify the steps of identifying attack path
by applying our IP traceback approach HITH, we
adopt the same network topology mentioned in
Figure 3. The studied network has four logging
routers (R1, R7, R10 and R14) and the other entities
constitute marking routers. First, the HITH
procedure identifies R14 since it is directly related to
the victim.
Then, it joined to the attack packet the value of
R14 router identity, its own TTL value and the
different values of the neighboring routers identities.
After calculating the digest by exploiting the same
hash function as in the logging process, we proceed
by comparing the obtained results with the entries in
the digest table. In case of correspondence, the
concerned router presents the next hop in attack path
reconstruction.
After identifying the router R13, the next step is
to broadcast queries to all routers’ neighbors. Upon
receiving this request, each router examines all
digest tables referring to the time interval of packet
reception to carry on with correspondence. Thanks
to the exploitation of TTL value, we avoid the risk
of false paths. Therefore, R12 then R11 routers are
identified for the attack path reconstruction. Then,
we proceed in the same manner to achieve the router
R1 in order to rebuild the path leading to the source
of attack.
4 PERFORMANCE
EVALUATION
In this section, we evaluate our proposed approach
HITH with analytical methods by comparing it to
some existing solutions: HIT (Hybrid single-packet
IP Traceback) (Gong and Sarac, 2008) and SPIE
(Source Path Isolation Engine) (Snoeren et al.,
2002).
A. Traceback Accuracy
Traceback accuracy presents a very important
criterion in the evaluation of IP traceback
mechanisms. Indeed, it defines the success rate of
attack path reconstruction and subsequently the
sureness of the obtained sources. However, a
traceback mechanism, which does not ensure this
specification, may give false results, incorrect paths,
and finally the failure of the traceback procedure.
On the other hand and owing to the adoption of
existing techniques in such a traceback mechanism,
some imperfections can be inherited from these
procedures. For example, we quote the Bloom Filter
technique for the management of packet digests in
the logging phase. Therefore, we cannot completely
eliminate all its problems, but we try to control the
selected parameters in order to reduce the resulting
anomalies.
In our study, we focus on two parameters: false-
positive rate and IP traceback precision.
1) False-positive Rate
Bloom Filter introduces a space-efficient data
structure that is used to organize elements and then
to check membership in this set. During the phase of
membership test, Bloom Filter can produce false-
positive results and never false-negative results. We
suppose the set of Bloom Filter parameters which is
listed in Table II.
The false-positive rate (P) depends on the
memory size of Bloom Filter as well as the size of
the stored digests. Therefore, it is exponentially
related to the value of memory efficiency factor (a /
b) (Broder and Mitzenmacher, 2005).
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
90
Table 2: Bloom Filter parameters.
Parameter Symbol Description
A Number of elements
(packets)
B Number of bits
a/b Memory efficiency factor
P False-positive rate
K Number of hash functions
=
1−1−
1

≈1
/
(1)
From equation (1), the false-positive rate (P) can be
managed and controlled by the right choice of the
factor (a / b) (Snoeren et al., 2002) and the (k) hash
functions. In our approach, HITH uses Bloom Filter
procedure with the addition of the TTL field. This
addition does not affect the rate (P) because we use
(k) hash functions with the same management of
digest table.
2) IP Traceback Precision
This parameter must be taken into account since the
design phase of such an IP traceback mechanism to
achieve precise and accurate results. Therefore, this
precision presents an essential factor in ensuring the
success of traceback by tracing the attack path
perfectly and identifying the true source of intrusion.
According to the example mentioned in Figure 3,
we can conclude that HIT approach still suffers from
some problems and precision vulnerabilities that
cause the uncertainty of the obtained results. In
HITH, this problem is solved by the introduction of
TTL field in the packet digest in order to reduce the
risk of erroneous and incorrect paths.
B. Storage Overhead
In this part, we borrow the evaluation methods from
HIT. Indeed, we evaluate the storage overhead
criterion following two parameters:
¾ Digest Table Storage (DTS): It reveals the
quantity of memory required for the registration
of packet digests in a router.
¾ Digest Table Access time (DTA): It designates
the number of packet digests stored in a table per
time unit.
Similar to SPIE, hybrid traceback approaches can
resort to a single packet or transformed packet in
order to identify the source of attack. During the
design phase of traceback approaches, we must not
neglect the case of packets transformation. Indeed,
IP packets may undergo various transformations,
such as fragmentation and tunneling, crossing the
network.
Table 3: Percentages of different types of IP packets.
Type de paquet IP Pourcentage
1) IP fragments
α
2) Non-fragmented
packets to be logged
at the router (includes
2.a, 2.b et 2.c)
(
1−α
)
2.a) Non-fragmented
packet not logged in the
two upstream routers.
(
1−α
)(
1−
)(
1−
)
2.b) Non-fragmented
packet logged at the
upstream router but
transformed at the current
router.
(
1−α
)

2.c) Non-fragmented
packet logged in the
upstream router two-hop
away and transformed in
the current router.
(
1−α
)(
1−
)

In this context and for our HITH approach, stored
packets in routers can be:
1) IP fragments.
2) Non-fragmented packets to be logged at the
router, comprising the following sub-cases:
a. Non-fragmented packet not logged in the two
upstream routers.
b. Non-fragmented packet logged at the
upstream router but transformed at the
current router.
c. Non-fragmented packet logged in the
upstream router two-hop away and
transformed in the current router.
Similar to HIT approach, we consider (
)the
percentage of packets to be logged at a router. We
assume (α) to be the percentage of fragmented IP
pack.ets and (β) the percentage of transformed
packets in the router. In addition, we set
(
)
to the
percentage of packets to be logged at router without
fragmentation. A consolidated list of these
percentages is shown in Table III.
According to the parameters listed in Table III, the
percentage of all of IP packets to be logged at the
router is expressed by:
=+
(
1−
)

(2)
The percentage of packets to be logged without
fragmentation is:
=
−
1−
(3)
And
1−=
1−
1−
(4)
HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks
91
Since the second case of fragmentation includes
three possible scenarios, () can be expressed by the
following equation:
=
(
1−
)
++
(
1−
)
(5)
We replace the value of () by (3) and
(
1−
)
by
(4) in (5), we obtain:
=1
(
1−
)
1+4
(
1−
)
−1
2
(
1−
)
(6)
Some measurement studies have proved that ≤
0.25% (McCreary and Claffy, 2000) and ≤3%
(Broder and Mitzenmacher, 2005), (Stoica and
Zhang, 1999). Therefore, we observe that:
0.382
0.392
(7)
The obtained result demonstrates that 39% of
packets crossing a router require the execution of
logging operation. On the other side, with HIT
approach and according to (Gong and Sarac, 2008),
the result is expressed by:
0,50
≤0,51
(8)
This means that 50% of IP packets must be logged
in the current router. In SPIE approach, all packets
crossing the router require the execution of logging
operation.
If we consider
, 
and 
the values
of  in HITH, HIT and SPIE approaches can be
given by:

=
×
≈
2
3

2
5

(9)
In addition, for our approach, logging packets can be
performed in multiple neighboring routers
simultaneously as detailed in digest table part.
Therefore, the rate of access to a digest table may be
reduced with the number of existing neighbors in
network. We suppose that a router has (n)
neighbors. In the ideal case where the traffic arrives
to router in a balanced way from each neighbor, and:

=
×
1

2
5
×
1

(10)
DTA
and DTA
represent the access time to digest
table with HITH and SPIE approaches, respectively.
In the worst case, where all the traffic is derived
from a single neighbor(=1), we note that:

=
×
2
5
×
(11)
5 CONCLUSION
To ensure a more effective and precise IP traceback
technique, the research community resort to combine
the two existing approaches; packet marking and
packet logging to establish a hybrid approach.
Although this method inherits the benefits provided
by each category, it still suffers from some
vulnerability. Indeed, the hybrid IP traceback
approach can cause incorrect paths due to accuracy
problems. In addition, overhead storage remains
high because of the inefficient use of the marking
space.
In this paper, we have proposed a novel Hybrid IP
Traceback approach for Heterogeneous wireless
networks, which is called HITH. Our solution is
based on some supplementary information added in
the reconstruction of the attack path to avoid
incorrect results. Moreover, HITH defines an
efficient mechanism to reduce storage overhead by
distributing the marking and logging roles between
routers. Besides, in order to decrease the digest table
access time, we have gathered the log information in
multiple routers taking into account the notion of
neighborhood and the limitation of some network
equipment capacities. The effectiveness of the
proposed IP traceback approach is proved by
mathematical analysis. Indeed, HITH incurs little
overhead at routers, improves accuracy and reduces
overhead storage and data access time. As a future
work, we will evaluate HITH by simulation analysis.
REFERENCES
Ren, W., Zhao, Q. and Swami, A., “Connectivity of
Heterogeneous Wireless Networks”. IEEE
Transactions on Information Theory, vol. 57, no. 7,
pp. 4315-4332, July 2011.
Houle, K. J. and Weaver, G. M., “Trends in Denial of
Service Attack Technology”. Computer Emergency
Response Team (CERT) Coordination Center,
technical report v1.0, October 2001.
Phatak, D., Sherman, A. T., Joshi, N., Sonawane, B.,
Relan, V. G., Dawalbhakta, A., “Spread identity: A
new dynamic address remapping mechanism for
anonymity and DDoS defense”.
Journal of Computer Security, vol. 21, no. 2, pp. 233-
281, 2013.
Savage, S., Wetherall, D., Karlin, A. and Anderson, T.,
“Network Support for IP Traceback”. IEEE/ACM
Transactions on Networking (TON), vol. 9, no. 3, pp.
226-237, June 2001.
Bellovin, S. M., “ICMP Traceback Messages”, IETF draft,
2000,
WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems
92
http://www.research.att.com/smb/papers/draftbel
lovin-itrace-00.txt.
Sager, G., “Security Fun with OCxmon and cflowd”.
Internet2 Working Group Meeting, November 1998,
http://www.caida.org/funding/ngi/content/security1198.
Choi, K. H. and Dai, H. K., “A marking scheme using
Huffman codes for IP traceback”. 7th International
Symposium Parallel Architectures, Algorithms
Networks (I-SPAN’04), pp. 421–428, Hong Kong,
China, May 2004.
Sattari, P., Gjoka, M. and Markopoulou, A., “A network
coding approach to IP traceback”. IEEE International
Symposium on Network Coding (NetCod’10), pp. 1-6,
Toronto, Canada, June 2010.
Yaar, A., Perrig, A. and Song, D., “FIT: Fast internet
traceback”. IEEE Conference on Computer
Communications (INFOCOM’05), vol. 2, pp. 1395-
1406, March 2005.
Song, D. X. and Perrig, A., “Advanced and Authenticated
Marking Schemes for IP Traceback”. IEEE
Conference on Computer Communications
(INFOCOM’01), vol. 2, pp. 878-886, Arkansas, USA,
April 2001.
Fadlallah, A. and Serhrouchni, A., “PSAT: Proactive
signaling architecture for IP traceback”. IEEE 4
th
Annual Communication Networks and Services
Research Conference (CNSR’06), pp. 293-299,
Washington, DC, USA, May 2006.
Bloom, B. H., “Space/time trade-offs in hash coding with
allowable errors”. Communications of ACM, vol. 13,
no. 7, pp. 422-426, July 1970.
Yang, M. H. and Yang, M. C., “RIHT: A Novel Hybrid IP
Traceback Scheme”. IEEE Transactions on
Information Forensics and Security, vol. 7, no. 2, pp.
789-797, April 2012.
Sai Priyanka, B. and Srihari Rao, N., “IP Traceback
Techniques – A Selective Survey”. International
Journal of Computer Science and Mobile
Applications, vol. 1, no. 3, pp. 40-44, September 2013.
Gong, Ch. and Sarac, K., “IP Traceback based on Packet
Marking and Logging”, IEEE International
Conference on Communications (ICC’05), vol. 2, pp.
1043-1047,16-20 May 2005.
Murugesan, V., Shalinie, M., Neethimani, N., “A Brief
Survey of IP Traceback Methodologies”. Acta
Polytechnica Hungarica, vol. 11, no. 9, pp. 197-216,
2014.
Goodrich, M. T., “Probabilistic Packet Marking for Large
Scale IP Traceback”. IEEE/ACM Trans. Networking,
vol. 16, no. 1, pp. 15-24, February 2008.
Snoeren, A. C., Partridge, C., Sanchez, L., Jones, C.,
Tchakountio, F., Schwartz, B., Kent, S. and Strayer,
W., “Single-packet IP traceback”. IEEE/ACM
Transactions on Networking, vol. 10, no. 6, pp. 721-
734, 2002.
Mölsä, J., “Mitigating denial of service attacks: A
tutorial”. Journal of Computer Security, vol. 13, no. 6,
pp. 807-837, 2005.
Stoica, I. and Zhang, H., “Providing guaranteed services
without per flow management”. ACM conference on
Applications, technologies, architectures, and
protocols for computer communication
(SIGCOMM’99), vol. 29, no. 4, pp. 81-94, Cambridge,
MA, USA , October 1999.
Muthuprasanna, M., Manimaran, G., Manzor, M., and
Kumar, V., “Coloring the internet: IP Traceback”. 12
th
International Conference on Parallel and Distributed
Systems (ICPADS’06), pp. 589-598, Minneapolis,
USA, August 2006.
Dean, D., Franklin, M. and Stubblefield, A., “An
algebragic appproach to IP traceback”. ACM
Transactions on Information and System Security, vol.
5, no. 2, pp. 119-137, 2002.
Gao, Z. and Ansari, N., “Enhanced probabilistic packet
marking for IP traceback”. IEEE Global
Telecommunications Conference (GLOBECOM’05),
vol. 3, pp. 1676-1680, 28 November - 2 December
2005.
Gong, C. and Sarac, K., “Toward a practical packet
marking approach for IP traceback”. International
Journal of Network Security, vol. 8, pp. 271-281,
2009.
Yan, D., Wang, Y., Su, S. and Yang, F., “A Precise and
Practical IP Traceback Technique Based on Packet
Marking and Logging”. Journal of Information
Science and Engineering, vol. 28, pp. 453-470, 2012.
Gong, Ch. et Sarac, K., “A More Practical Approach for
Single-Packet IP Traceback using Packet Logging and
Marking”. IEEE Transactions on Parallel and
Distributed System, vol. 19, no. 10, pp. 1310-1324,
October 2008.
Broder, A. and Mitzenmacher, M., “Network applications
of Bloom filters: A survey”. Internet Mathematics,
vol. 1, no. 4, pp. 485-509, 2005.
McCreary, S. and Claffy, K., “Trends in wide area IP
traffic patterns: A view from Ames internet
exchange”. 13
th
ITC Specialist Seminar on IP Traffic
Modeling, Measurement and Management, pp. 1-25,
Monterey, CA, USA, 2000.
HITH: Hybrid IP Traceback for Heterogeneous Wireless Networks
93