![](https://csdnimg.cn/release/download_crawler_static/21749679/bg4.jpg)
2020 Cai et al. / Front Inform Technol Electron Eng 2017 18(12):2017-2034
2014; Prakash and Wellman, 2015; Vadlamudi et al.,
2016), but application of game theory on any other
aspect is still absent.
Jia et al. (2006) proved that the interaction be-
tween attacker and defender is a non-cooperative dy-
namic game with incomplete information. Therefore,
we use incomplete information dynamic game theory
to investigate the defense mechanism of MTD from
an abstract point of view. Specifically, we use incom-
plete information dynamic game models to model the
interaction between the defender and the attacker
while deploying MTD technology, and obtain the
equilibria and their conditions. At the same time
we take the incomplete information dynamic game
model for the traditional defense and its equilibrium
proposed by Shi et al. (2009) as a baseline for com-
parison. Thereafter, we compare the equilibria and
their conditions when deploying MTD technology to
the equilibrium and its condition when deploying the
traditional defense approaches, and after further dis-
cussion, validate the proactivity and effectiveness of
MTD technology. Also, we identify the factors that
would influence the effect of MTD.
We now describe the game theory scenario con-
sidered in this study and the baseline game for
comparison.
3.1 Study scenario
The scenario we have examined is as follows.
There are two players: one is a defender who can
equip a server with different MTD approaches to im-
prove the server’s security while enabling it to pro-
vide a specific service such as web service, and the
other is a visitor who can be a normal user or an
attacker that attempts to launch attacks.
Existing MTD approaches can be categorized
into three main categories, namely, software trans-
formations (ST) (Jajodia et al., 2011), dynamic
platform techniques (DPT) (Okhravi et al., 2014),
and network address shuffling (NAS) (Carroll et al.,
2014). Each of these attempts to use its own method-
ology for designing various approaches to defend
against the attacker. Their running patterns should
follow the ‘hidden’ pattern or the ‘variation’ pattern
(Cai et al., 2016b). The major difference between the
two patterns is that the MTD approaches with ‘hid-
den’ pattern would make the attacker lose the target
and thus break off the connection with the target,
while the MTD approaches with ‘variation’ pattern
can disrupt attack but cannot prevent the attacker
from re-connecting with the target. The approaches
in the categories of ST and DPT conform to the ‘vari-
ation’ pattern, and the approaches in the category of
NAS conform to the ‘hidden’ pattern. In the study
scenario in this study, we consider that the MTD
approaches deployed by the defender can be of three
types: the ‘hidden’ MTD, the ‘variation’ MTD, and
‘mixed’ MTD (i.e., the combination of the ‘hidden’
and ‘variation’ MTDs). The relationship between
them is shown in Fig. 4.
MTD
approaches
Software
transformations
Dynamic platform
techniques
Network address
shuffling
‘variation’
‘hidden’
‘mixed’
Fig. 4 The relationship b etween the categories and
the running patterns of major MTD approaches
The MTD approaches with different running
patterns can dynamically change a certain server
property (i.e., shifting the attack surface) in differ-
ent ways, and their running patterns would exhibit
in the process of serving. Therefore, the external
service mode would be different when different types
of MTD techniques are deployed. Here, according
to the types of MTD deployed, we consider that the
server can provide three kinds of service, hid-service,
var-service, and the combination of the two. More-
over, the defense is included in each type of service.
Furthermore, because different types of MTD tech-
niques would influence the attacker in different ways,
the utilities for the interaction between the defender
and the attacker would be different when deploying
different types of MTD, and we will discuss them in
detail in Section 4.
3.2 Baseline game
We take the game under the situation of tradi-
tional defense proposed by Shi et al. (2009) as the
baseline game for comparison.
The scenario taken in the baseline game is that
a server provides the web service and is deployed
with a traditional defense approach, and an attacker
attempts to launch a DoS attack. Through mod-
eling the interaction between the defender and the
attacker, Shi et al. (2009) arrived at an equilibrium