Attack countermeasure tree (act) meets with

International Journal of Computer Networks & Communications (IJCNC) Vol.7, No.4, July 2015
DOI : 10.5121/ijcnc.2015.7407 99
Attack Countermeasure Tree (ACT) meets with
the Split-protocol
Bharat S Rawal
Department of Information Sciences and Technology, Penn State Abington, Abington,
PA 19001, USA.
ABSTRACT
In this paper, we present a novel attack tree paradigm called attack countermeasure tree (ACT) comprising
an additional attack resistant element known as the Split-protocol. ACT which circumvent the fabrication
and way out of a state-space representation and takes keen on account attack, as well as countermesures
(in the form of detection and mitigation events). Split-protocol as an attack resistant element enhances the
availability of the system during or after a security attack on the system. We compare ACT with or without
Split-protocol implantation. The split-protocol concept stemmed from splitting the HTTP/TCP protocol in
webserver application. An HTTP/TCP protocol is standard on a webserver can be split into two segments,
and each part can be separately run on a different Web server, thus constituting dual servers. These servers
communicate across a network by using inter-server messages or delegate messages.In ACT, recognition
and alleviation are allowed not just at the leaf node but also at the intermediate
nodes,andsimultaneouslythe state-space explosion problem is avoided in its analysis. We study the
consequences of incorporating countermeasures in the ACT and Split-protocol using various case studies.
KEYWORDS
attack trees, non-state-space model, mincuts, split-protocol, and reliability.
1.INTRODUCTION
The concept of attack trees is introduced from fault trees in software safety. Fault trees are used to
describe how errors disseminate in software systems, and analysis of this could be used to exam
software [10] and [11]. Although fault trees are most commonly used to model how problems
occur in critical systems, given the built in focus on error propagation, attack trees have a slightly
dissimilar perspective. The starter of an attacker, or group of attackers, makes it believable to
model extortion to an institute as well as the aspect of targeted attacks. Deliberations of
likelihoods based on existing money, tools or incentive for the attacker provides a more carefully
grounded duplicate of the risk level. Bruce Schneier offered the concept of attack trees as a way
to model threats against computer systems[12] The basic tree model describes the Attack Tree
with two different types of nodes, AND-nodes, and OR-nodes. At OR-nodes, a smallest of one
sub-goal should be fulfilled to achieve the goal of the node. At AND-nodes, every sub-goal
should be achieved to realize the objective of the node. The Boolean calculation could be done on
the values of sub goal nodes based on a Boolean expression (AND-/OR) in the node, giving the
following account in the node [9].

100
To evaluate the safety of the system security, modeling is used.Regular step towards security
assessment is to plan and build a scalable model [12], [13] that helps to compute the security [14]
in terms of important characteristics such as the damage produced by an outbreak or the gain
achieved by implementing a certain set of countermeasures [15]. The simplest model issued in
this context is attack tree (AT) [1]. AT utilize the genetic algorithms to find optimal
countermeasure sets for the system from their AT models [6]. Though, the basic construct of AT
does not take into account defense mechanisms. Roy et al. developed a novel attack tree model
called attack countermeasure tree (ACT). The ACT is built on following deliberation, A) defense
mechanisms not restricted to not just at the leaf B) Mincuts are used to generate and analysis of
the attack and the attack countermeasure scenarios. C) Security analysis using various measures is
performed in an integrated manner [1].
The remainder of this paper is organized as follows. A brief related work is presented in section
II. In Section III, Split-protocol architecture presented. In Section IV, describe Split-protocol
configurations V. Describes attack countermeasure tree (ACT). VI. Present quantitative and
qualitative analysis. Some simulation results and the impact Split-protocol using ACT on
security analysis are discussed. Finally, we conclude the paper in Section VII.
2.RELATED WORK
A security risk is being modeled using a graphical, mathematical, decision tree structure called an
attack tree. There is cause to believe that attack trees are widely patronized by the intelligence
community. Fault trees were invented in the early 1960s for use in the Minuteman Missile System
[16]. Weiss described the threat logic trees [17]. Amoroso [18] detailed a modeling concept he
called threat trees. Then, Schneider [19] (noted security expert) promoted the idea though he
called it attack trees (AT). Moore et.al [20] prolonged Schneider’s AT by familiarizing attack
scenarios and attack profiles. Mauw et.al [21] developed an alternative formalism for AT where
the goal was associated with the set of all mincuts. When applied to complex case studies, AT
often became large and unwieldy. Therefore, Daley [22] proposed a layered approach to partition
attack tree nodes with respect to their functionality.
To the best of our knowledge, a technique for splitting an HTTP-based TCP connection in this
manner has not been proposed previously. Splitting is similar to mask failures in TCP [23] and to
use the M-TCP (Migratory TCP) protocol to migrate TCP connections from one server to another
[24]. However, connection migration in M-TCP does not involve splitting an HTTP request and
TCP connection or operating in split mode. Moreover, the client needs to initiate the migration
process in M-TCP, whereas TCP connection splitting is transparent to the client. TCP connection
splitting is also different from migrating Web applications in mobile computing [25], which
moves applications from one node to another on the Web by using virtualization techniques.
Likewise, connection splitting is different from process migration in [26], which requires that an
agent be dispatched from one system to run on another at a remote location.
3.SPLIT-PROTOCOL ARCHITECTURE
The basic split protocol architecture used for the experiments described in [2] is reproduced here
for illustration as shown in Figure 1. The http request is splited at the GET command between a

International Journal of Computer Networks & Communications (IJCNC) Vol.7, No.4,
CS and a DS. The CS handles the connections, and the DS handles the data transfer.
connections, the CS also handles
knowledge of the requested file, its nam
the file itself. However, the DS has the file and serves the data to the client.
After the GET command is received by CS
delegate message DM1 to DS. The message DM1
that is stored in CS in the configuration
When DM1 reaches the DS, it creates its TCB entry and starts processing this request as if it
initiated itselfin the DS. When a DS sends data to the
hadreceivedan FIN-ACK from the
server packet referred to as DM2 to DS. The DM2 received by DS will close the state of the
request in DS. These DM1 and DM2 inter
at DS. More details of the design and imple
The CS and DS architecture illustrated in Figure 1 provides a variety of delegation configurations
of given requests. A request received by
DS. That is, some requests can
variation of delegation ratio. As CS and DS are identical functional units, they can also perform
any given role of CS or a DS. During these
percentage is 25% in both directions (CS and DS) [4], we achieved the maximum throughput.
The measurements indicated that the optimal split server performance was
(maximum can be 2.0). Thus, the two
server). These initial results provided us motivation to construct split protocol based servers as
described below. These novel splitting techniques and associated Web server architecture
introduced in this section also showed some p
server reliability.
Figure 1. Split
CS and a DS. The CS handles the connections, and the DS handles the data transfer.
handles the data ACKs and the connection closing. The CS has complete
knowledge of the requested file, its name, size, and other attributes, but it may or may not have
the file itself. However, the DS has the file and serves the data to the client.
the GET command is received by CS, it sends an ACK to the client and also sends a
delegate message DM1 to DS. The message DM1 contains the state infprmation
configuration of an entry in the TCP table (known as a TCB entry).
DM1 reaches the DS, it creates its TCB entry and starts processing this request as if it
. When a DS sends data to the client, it uses the CS’s IP. After CS
from the client to signify connection closing, it sends another inter
request in DS. These DM1 and DM2 inter-server packets serve as the start and end of the
at DS. More details of the design and implementation can be found in [3].
A request received by CS can be either processed wholly at CS or delegated to
can be handled at CS and some can be transferred to DS resulting in a
any given role of CS or a DS. During these experiments, we found that when the delegatio
The measurements indicated that the optimal split server performance was 1.8035 for two servers
(maximum can be 2.0). Thus, the two-server system suffers only 20% capacity (10% for each
). These initial results provided us motivation to construct split protocol based servers as
introduced in this section also showed some potential in distributed computing and improving
Figure 1. Split-protocol Architecture [3]
101
CS and a DS. The CS handles the connections, and the DS handles the data transfer. Also to
the data ACKs and the connection closing. The CS has complete
e, size, and other attributes, but it may or may not have
and also sends a
infprmation of the request
as a TCB entry).
DM1 reaches the DS, it creates its TCB entry and starts processing this request as if it was
it uses the CS’s IP. After CS
it sends another inter-
server packets serve as the start and end of the request
CS can be either processed wholly at CS or delegated to
to DS resulting in a
we found that when the delegation
1.8035 for two servers
capacity (10% for each
). These initial results provided us motivation to construct split protocol based servers as
otential in distributed computing and improving

5.SPLIT CONFIGURATIONS
Configuration 2 in Fig. 2 shows a single CS with two or more DSs in the system with
partial or full delegation. In partial delegation mode, clients designated as non
request clients (NSRCs) send requests to the CS, and these requests are processed
completely by the CS as usual. The connections between the NSRCs and the CSs are
shown as dotted lines. With full
(SRCs) make requests to the CS, and these requests are delegated to DSs. For full
delegation, there are no NSRCs in the system. When requests are delegated to DSs, we
assume that they are equally dist
fashion. It is also possible to employ other distribution strategies
Figure 2. Split
Figure 3 shows a general configuration for connecting one DS, one or more
clients (Configuration 2.). It shows
configuration, we used small file sizes to avoid overloading the single DS[
Figure 3. Split
CONFIGURATIONS
In partial delegation mode, clients designated as non
shown as dotted lines. With full delegation, clients designated as split-request clients
assume that they are equally distributed among DS1, DS2 and DS3 in round
fashion. It is also possible to employ other distribution strategies [4].
Figure 2. Split-protocol Architecture Configuration 1[4]
Figure 3 shows a general configuration for connecting one DS, one or more CSs and one or more
. It shows two CSs and one DS with both SRCs and NSRCs. For this
configuration, we used small file sizes to avoid overloading the single DS[4].
Figure 3. Split-protocol Architecture Configuration 2 [4]
102
In partial delegation mode, clients designated as non-split
request clients
ributed among DS1, DS2 and DS3 in round-robin
CSs and one or more
and one DS with both SRCs and NSRCs. For this

Figure 4 describes a general configuration for
or more clients (Configuration 3.). Various split configurations are useful according to the need of
system functionality. Example if we need a faster data transfer for
/Multi Server (MC/MS) configuration best ch
a connection with one connection server, and concurrent
data servers located on different subnets from each other. The data can be
clients anywhere on the network, which then reassemble or otherwise process the data. MC/MS
distributes the data of large file across multiple servers without any redundancy. The separation
of data transfer from a connection establishment is entirely transparent to the
Figure 4. Split
6.ATTACK COUNTERMEASURE
6.1Parameters
Ak is an attack event
Dk is a detection event
Mk isa mitigation event
CMk is a countermeasure
ACT = {V, ψ , E} (V: set of all
set of all gates in ACT, E: aset of all
where V= {∀k, vk: vk ∈ {Aj}|| vk
{Ml}} where A1, A2, D1, D2, M1, M2, are the
events of the ACT, = {∀k, k: k
gate}}, E= {∀k, ek: ek ∈ (vi, ψ
and X = (xA1xA2 ...xD1xD2 ...xM1x
vector for the ACT where xAk, x
variables associated with events A
Φ(X) structure function of an ACT
pAkodds of occurrence of attack event A
a general configuration for connecting multiple CSs, and various
system functionality. Example if we need a faster data transfer for alarge file, then Multi Client
/Multi Server (MC/MS) configuration best choice. In MC/MS architecture, one client establishes
a connection with one connection server, and concurrent data transfer dispatched from multiple
data servers located on different subnets from each other. The data can be sent to the multiple
ere on the network, which then reassemble or otherwise process the data. MC/MS
of data transfer from a connection establishment is entirely transparent to the client.
Figure 4. Split-protocol Architecture Configuration 3[5]
OUNTERMEASURE TREE (ACT)
, E} (V: set of all vertices in ACT, :
set of all boundaries in ACT)
{Aj}|| vk ∈ {Di}|| vk ∈
{Ml}} where A1, A2, D1, D2, M1, M2, are the
k, k: k ∈ {AND, OR, k-of-n
(vi, ψ j ) || ek ∈ (ψ i , ψ j )}
xM2 ...) is a state
, xDk, xMk are the boolean
variables associated with events Ak, Dk,Mk respectively.
(X) structure function of an ACT
of occurrence of attack event Ak
103
various DSs and one
file, then Multi Client
oice. In MC/MS architecture, one client establishes
dispatched from multiple
to the multiple
ere on the network, which then reassemble or otherwise process the data. MC/MS
client.

pDkodds of success of detection event D
pMkodds of success of mitigation event M
Pgoal odds of attack success at the ACT goal
pUDodds of undetected attack at the ACT goal
pDUModds of detected but unmitigated attack at
the ACT goal
In this subsection, the basic formalism of ACT is reproduced.
classes of events: attack events (e.g., install a keystroke logger),
keystroke logger) and mitigation
ACT for a regular server system (non
Figure 5(a) shows simple ACT with a single
probability of success a successful attack at goal node
PGoal =PA
In figure 5(b), one attack event and one detection mechanism are applied. The corresponding
expression for the probability of an undetected attack at goal node is shown Eq.(1)..
PGoal = PA (1
For n, detection mechanisms are being used to detect one attack event equation becomes.
The corresponding PGoal is:
PGoal = PA (1-
In figure 5(c), one attack event, one detection mechanism
corresponding expression for the probability of an undetected attack at goal node is shown Eq.(4).
PGoal = PA
Figure 5. ACT without Split
(
of success of detection event Dk
of success of mitigation event Mk
of attack success at the ACT goal
of undetected attack at the ACT goal
of detected but unmitigated attack at
In this subsection, the basic formalism of ACT is reproduced. In ACT, there are three
classes of events: attack events (e.g., install a keystroke logger), discovery events (e.g.,
keystroke logger) and mitigation activities (e.g., get rid of keystroke logger).
ACT for a regular server system (non-Split system)
Figure 5(a) shows simple ACT with a single-attack event. The corresponding expression for the
probability of success a successful attack at goal node is shown Eq. (1).
(1)
(1- PD) (2)
- PD1) (1- PD2) (1- PD3) …(1- PDn) (3)
figure 5(c), one attack event, one detection mechanism, and mitigation events are applied. The
A (1- PD PM) (4)
Figure 5. ACT without Split-protocol [1]
(a) (b) (c)
104
In ACT, there are three different
events (e.g., sense
attack event. The corresponding expression for the
and mitigation events are applied. The

Figure 5(a), representACT with one attack event
event and 5(c) represents ACT with one attack,one detection event and onemitigation event
ACT tree withoutSplit-protocol and t
successful attack at goal node is shown
PGoal = ½ PA
PGoal = ½ PA (1- PD)
PGoal = ½PA (1- PD PM)
Figure 6, represent the equivalent
protocolimplementationFigure 5(a), representACT with one attack event
5(b),ACT with two attacks , two
with two attacks, two detectionand two
Figure 6. Equivalent ACT
Figure 7. Equivalent
representACT with one attack event 5(b),ACT with one attack and one detection
ACT with one attack,one detection event and onemitigation event
protocol and the corresponding expression for the probability of success a
successful attack at goal node is shown are equation 5, 6 and 7.
)
) (7)
Figure 6, represent the equivalent ACT for figure 5(a) and 5(b) and
Figure 5(a), representACT with one attack event on CS and DS
s , two detections an two mitigation events. Figure 7, represents
and two mitigation eventsfor ACT tree with Split-protocol.
Equivalent ACT for figure 5(a) &5(b) with Split-protocol
Equivalent ACT for figure 5(c) with Split-protocol
105
ACT with one attack and one detection
ACT with one attack,one detection event and onemitigation event for
he corresponding expression for the probability of success a
(5)
(6)
(7)
and with Split-
on CS and DS
represents ACT
protocol.

Figure 8. shows one attack event, n detection
expression 8 displays the probability of a successful
attack was detected but not successfully mitigated.
PGoal = ܲ‫ܣ‬ሺ1 െ ሺ1 െ ∏ ሺ1 െ௠
௜ୀ଴
Figure 8. shows aSplit system with one attack event, n detection
The corresponding expression 9
detected, or attack was detected but not successfully mitigated.
PGoal= (1/2) {ܲ‫ܣ‬ሺ1 െ ሺ1 െ ∏ ሺ1௠
௜ୀ଴
Figure 9.
one attack event, n detection events, and n mitigation events. The corresponding
the probability of a successful attack, either attack was not
attack was detected but not successfully mitigated.
ሺ െ ܲ‫݅ܦ‬ሻሻሻ ൈ ሺ1 െ ∏ ሺ1 െ ܲ‫݅ܯ‬ሻሻሻ௡
௜ୀ଴
Figure 8. ACT without Split-protocol for multiple detection and mitigation events [
system with one attack event, n detection events, and n mitigation
The corresponding expression 9 shows the probability of a successful attack; either attack
or attack was detected but not successfully mitigated.
ሺ1 െ ܲ‫݅ܦ‬ሻሻሻ ൈ ሺ1 െ ∏ ሺ1 െ ܲ‫݅ܯ‬ሻሻሻ௡
௜ୀ଴ } 9
Figure 9. Equivalent ACT for figure 8 with Split-protocol
106
and n mitigation events. The corresponding
attack was not detected or
(8)
events [1].
mitigation events.
either attack was not
} 9

Figure 10. shows one attack event, n pairs of detection event and mitigation event. The
corresponding expression is displayed
Figure 11. shows ACT structure with
event and mitigation event. The corresponding expression 11
successful attack.
PGoal = ½ܲ‫ܣ‬ ∏ ሺ1 െ ܲ‫݅ܦ‬ ൈ ܲ‫݅ܯ‬௡
௜ୀ଴
Figure 11. ACT with Split
displayed by the equetion 10 for the probability of successful attack.
PGoalܲ‫ܣ‬ ∏ ሺ1 െ ܲ‫݅ܦ‬ ൈ ܲ‫݅ܯ‬ሻ௡
௜ୀ଴ 10
Figure 10. ACT without Split-protocol formultiple pairs of detection and mitigation events [
Figure 11. shows ACT structure with asplit system with one attack event, n pairs of detection
event. The corresponding expression 11 displays the probability of
ܲ‫݅ܯ‬ሻ11
Figure 11. ACT with Split-protocol for multiple pairs of detection and mitigation events.
107
the probability of successful attack.
events [1].
, n pairs of detection
the probability of
pairs of detection and mitigation events.

7.QUALITATIVE AND
7.1 Probabilistic Analysis
Table I.
Gate Type
AND gate
OR gate
k/n gate*
*for identical inputs [1]
Figure 12, illustartes adirect attack tree for resetting the BGP session
top eventis connected with the set of all mincuts.Mincuts ofAT represent attack scenarios [
whereas those of an ACT, represent attack
Figure 12. A straightforward
Top= {{[(A1121+A1122+A1123) + (A
AND QUANTITATIVE ANALYSIS WITH
Table I. methods for probability of attack success
Gate Type Prob. of attack success
AND gate
ෑ ‫݌‬ሺ݅ሻ
௡
௜ୀଵ
1 െ ෑሺ1 െ ‫݌‬ሺ݅ሻሻ
௡
௜ୀଵ
෎ ൬
݊
݆
൰ ‫݌‬௝
∗ ሺ1 െ ‫݌‬ሻ௡ି௝
௡
௃ୀ௞
*for identical inputs [1]
attack tree for resetting the BGP session. In both AT and ACT, the
with the set of all mincuts.Mincuts ofAT represent attack scenarios [
whereas those of an ACT, represent attack-countermeasure scenarios
straightforward attack tree for resetting the BGP session [1]
(A111)]}. [(D12).M12). (A12)].(D1.M1)} + {(D2.M2).A2)}}
108
WITH ACT
In both AT and ACT, the
with the set of all mincuts.Mincuts ofAT represent attack scenarios [1, 26]
)}}

Some of the Boolean algebra laws are
i. Commutative Law
(a) A + B = B + A
(b) A B = B A
ii. Associative Law
(a) (A + B) + C = A + (B + C)
(b) (A B) C = A (B C)
iii. Distributive Law
(a) A (B + C) = A B + A C
(b) A + (B C) = (A + B) (A + C)
iv. Identity Law
(a) A + A = A
(b) A A = A
v.Redundancy Law
(a) A + A B = A
(b) A (A + B) = A
Figure 13. An attack
laws are.
(b) A + (B C) = (A + B) (A + C)
attacktree with Split-protocol for resetting the BGP session
109

110
TOP = {{[(A1121+A1122+A1123)+(A111)]}.{[(D12).M12).(A12)].(D1.M1)}+{(D2.M2).A2)}}
*{{[(A1121+A1122+A1123)+(A111)]}.[(D12).M12).(A12)].(D1.M1)} +{(D2. M2).A2)}}
vi. Applying the Identity Law
A.A = A
= {{[(A1121+A1122+A1123) + (A111)]}. [(D12).M12). (A12)].(D1.M1)} + {(D2.M2).A2)}}
Split Protocol: Failure Rate and Survival Function
The distribution of each protocol component is the binomial distribution. With the number
protocol of component n approach infinity, distribution of each component is poison distribution
[4] with arrival rate, λ that is equal to the receiving rate of each component. Thus, service time or
failure rate (FR) of each element is an exponential distribution [3].
1. All the n components service time X is exponentially distributed:
F (T) = P {X≤T} = 1- e-λT; f (T) = λe-λT
2. Each ith
component1 ≤ i ≤ n, Failure Rate (FR) is constant.(,λi(t)= λi)
3. All n components are identical. Then FR of each element is equal to λ (λi = λ; 1 ≤ i ≤ n)
4. All n components are independent. Then
5. P {X1, X2...Xn> T} = P{X1>T} P{X2>T}…P{Xn>T}
6. The reliability of each component, Ri(T) is
a. Ri (T) = P {Xi> T} = e-λT
b. λ = -ln (Ri (T))/T
c. T denoted system mission time.
7. System failure rate is 1 – R (T) where R (T) indicated the reliability of the whole system.
There are supposed "n” protocol components and probability of non-failure (of each
component(x1, x2, x3, x4...) are exponentially distributed: For simplicity we will assume, every
i^th component 1 ≤ i ≤ n probability of failure is equal to all component, i.e. Failure Rate (FR) for
each component is same and (θi (τ) = θi). For given operational time and all system, components
are identical and their failure time is independent. Therefore, the reliability of, any ithcomponent
(1 <i< n) reliability "Π i(τ)": "Π i(τ)" = P(Xi> τ) = ݁஘୧த
=>θi =-ln(Π i(τ))/ τ .
First, we have assumed identical components, which are identical DS in a cluster System, and
they have same FR. Also, they are independent components those whose failure does not affect
the performance of any other system component [2].

111
Reliability of Parallel identical components:
Π S =1-(1 - Π 1) × (1 - Π 2) ×... (1 - Π n); if the component reliabilities differ
Π S =1-(1 - Π) × (1 - Π) ×... (1 - Π); if the component with similar reliability
1-[1 - Π ]n
For example system of two parallel components (CS, DS)
Π τ =1-{(1-Π 1(τ)) (1-Π 2(τ))}
= 1- {(1- ݁ିθଵτ
) ( 1 − ݁ିθଶτ
)}
= ݁ିθଵτ
+݁ିθଶτ
- ݁ି(θଵାθଶሻτ
And MTTF = µ =‫׬‬ ߨ (ܶ
∝
଴
ሻ݀τ= ‫׬‬ (݁ିθଵτ
+ ݁ିθଶτ
− ݁ି(θଵାθଶሻτ∝
଴
ሻ݀τ
=
ଵ
θଵ
+
ଵ
θଶ
-
ଵ
θଵାθଶ
And FR = θs = Density Function / Survival Function
= -
ௗ
ௗ௧
ߨ (τሻ / Π (τ)
=(θ1݁ିθଵτ
+ θ2݁ିθଶτ
– ( θ1 + θ2ሻ݁ି(θଵାθଶሻτ
/(݁ିθଵτ
+ ݁ିθଶτ
− ݁ି(θଵାθଶሻτ
)(22)
This system hazard rate θs(τ) can be calculated as a function of any mission time τ [3].
7.2 Mincut Analysis
According to Roy, et al, the mincuts (attack alleviationscenarios) of the ACT in Figure10 are
{(A111, CM1, A12, CM12), (A1121, CM1, A12, CM12), (A1122, CM1, A12, CM12), (A1123, CM1, A12,
CM12), (A2, CM2)} (where CM1= (D1M1), CM12= (D12M12), M2= (D2M2))[1]. Each of the 5
mincuts corresponds to a permutation of actionseach of happening will result in attack hit at the
target. For example, the mincut (A1122, CM1, A12, CM12) indicates that iftogether the attack events
A1122 and A12 were to take place and if both the defense activity CM1 and CM12 fail, attack will
be successful. From the mincut (A1122, CM1, A12, CM12) we also observe that the pair of attack
events (A1122, A12) is covered by either of the countermeasures CM1 or CM12 [1].
7.3 Qualitative Analysis:
Minimal cut set (mincut): a minimum combination of primary events that induce the top event
Introducing Split-protocol increase length of mincut, which signals low vulnerability. The split
does not introduce additional new cut sets. This implies that the inclusion of split system does not
introduce additional vulnerability in the overall system. Split- protocol reduces thechance of a
single point of failure. Spilt –protocol introduces n parallel components, to fail system all n
component must be faulty. The splitprotocol offers inbuilt architecture reliability and fault
tolerance against DoS/DDoS attack [8].

112
8.CONCLUSION
In this paper,we have presented the attack countermeasure trees (ACT) with implementing, a non-
state-space representation that permits us to perform qualitative and probabilistic analysis of the
security of the system. ACT takes into account attacks as well as countermeasures (in the form of
detection mechanisms and mitigation techniques). The detection and mitigation can be placed not
just at the leaf node but also at any intermediate node. When we implement the Split-protocol in
thesystem, it reduces the probability of system failure by 50%. If thesystemis made of n split unit,
system reliability will improve by n times. The innovative splitting system and associated Web
server architecture introduced in this paper have potential applications in distributed computing
and improving server reliability.
REFRENCES
[1] Roy, Arpan, Dong Seong Kim, and Kishor S. Trivedi. "Attack countermeasure trees (ACT): towards
unifying the constructs of attack and defense trees. “Security and Communication Networks 5.8
(2012): 929-943.
[2] Roy, Arpan, Dong Seong Kim, and Kishor S. Trivedi. "Scalable optimal countermeasure selection
using implicit enumeration on attack countermeasure trees." Dependable Systems and Networks
(DSN), 2012 42nd Annual IEEE/IFIP International Conference on. IEEE, 2012.
[3] B.Rawal, R. Karne, and A. L. Wijesinha. “Splitting HTTP Requests on Two Servers.” The Third
International Conference on Communication Systems and Networks: COMPSNETS 2011, January
2011, Bangalor, India.
[4] B. Rawal, R. Karne, and A. L. Wijesinha. “Mini Web Server Clusters based on HTTP Request
Splitting” HPCC 2011 : The 13th IEEE International Conference on High Performance Computing
and Communications ,Sep 2, 2011- Sep 4, 2011, Banff, Canada.
[5] Rawal, Bharat, et al. "Split-Encoding: The Next Frontier Tool for Big Data."Advanced Computing,
Networking and Informatics-Volume 1. Springer International Publishing, 2014. 501-510.
[6] START Understanding Series and Parallel Systems Reliability, Selected Topics in Assurance Related
Technologies, volume 11, Number 5 (34)
[7] Sheldon M. Ross, A First Course In Probability, Eighth Edition
[8] Resistant Augmented Split Architecture,” IEEE 10th HONET- CNS , EMU, Famagusta, Cyprus
2013.
[9] Espedalen, Jeanne H. "Attack trees describing security in distributed internet-enabled metrology."
(2007).
[10] Leveson, N. G. 1995. Safeware: System Safety and Computers. Addison-Wesley,Reading MA.
[11] Viega, J. & McGraw, G. 2002. Building Secure Software: How to Avoid Security Problems the
Right Way. Addison-Wesley.
[12] Ortalo R, Deswarte Y, Kaˆaniche M. Experimenting with quantitative evaluation tools for
monitoring operational security. IEEE Trans. on Software Engineering 1999; 25(5):633–
650.
[13] Schneier B. Secrets and Lies: Digital Security in a Networked World. John Wiley and Sons Inc., New
York, NY, USA, 2000.
[14] Trivedi KS, Kim DS, Roy A,Medhi D. Dependability and security models. Proc. DRCN, IEEE,
2009; 11– 20.
[17] J.D. Weiss, A System Security Engineering Process, Proceedings of the 14th National Computer
Security Conference, 1991

113
[18] Edward G. Amoroso, Fundamentals of Computer Security Technology, pp 15-29, Prentice-Hall,
1994, ISBN01310892935
[19] B. Schneier, Attack Trees, Dr. Dobb's Journal, v. 24, n. 12, December
[20] Moore AP, Ellison RJ, Linger RC. Attack Modelingfor Information Security and
Survivability.CMU/SEI-2001-TN-001 2001.
[21] Mauw S, Oostdijk M. Foundations of Attack Trees.LNCS2006; 3935:186–198
[22] Daley K, Larson R, Dawkins J. A StructuralFramework for Modeling Multi-stage Network Attacks.
Proc. ICPPW, 2002; 1530–1536.
[23] D. Zagorodnov, K. Marzullo, L. Alvisi and T.C. Bressourd, “Practical and low overhead masking of
failures of TCP-based servers,” ACM Transactions on Computer Systems, Volume 27, Issue 2,
Article 4, May 2009.
[24] K. Sultan, D. Srinivasan, D. Iyer and L. lftod. “Migratory TCP: Highly Available Internet Services
using Connection Migration,” Proceedings of the 22nd International Conference on Distributed
Computing Systems, July 2002.
[25] G. Canfora, G. Di Santo, G. Venturi, E. Zimeo and M.V.Zito, “Migrating web application sessions in
mobile computing,” Proceedings of the 14th International Conference on the World Wide Web, 2005,
pp. 1166-1167.
[26] T. Venton, M. Miller, R. Kalla, and A. Blanchard, “A Linux-based tool for hardware bring up, Linux
development, and manufacturing,” IBM Systems J., Vol. 44 (2), IBM, NY, 2005, pp. 319-330.Gan Z,
Tang J, Wu P, Varadharajan V. A NovelSecurity Risk Evaluation for Information Systems.Proc.
FCST, 2007; 67–73.
Author
Dr. Bharat Rawal, has conducted research in the area of computer networks, including
wireless networks, Split- protocol designs and analyzes, and network performance
evaluations, HPC and Network security. He was the author and co-author in several papers in
networking and security area. Currently, he has focused on solving a big integers and data
compression in Split-protocol infrastructure. He is now server as Assistant Professor in IST
department at Penn State Abington and Visiting Assistant Professor in the department of
Electrical and Computer Engineering at Duke University.

Attack countermeasure tree (act) meets with

More Related Content

What's hot (20)

Viewers also liked (19)

Similar to Attack countermeasure tree (act) meets with (20)

More from IJCNCJournal (20)

Recently uploaded (20)

Attack countermeasure tree (act) meets with