A Novel Computer Virus Model and Its Stability€¦ · virus, Trojan horses, worm, logic bomb and...
Transcript of A Novel Computer Virus Model and Its Stability€¦ · virus, Trojan horses, worm, logic bomb and...
A Novel Computer Virus Model and Its Stability
Mei Peng and Huajian Mou College of Mathematics and Computer Science, Yangtze Normal University, Chongqing, China
Email: [email protected], [email protected]
Abstract—Computer virus is a malicious code which can
causes great damage. A SEIR model is proposed which
considered that the newly entered computer in the network
has been infected by other ways. It is concerned with the
constant immigration, which includes susceptible, exposed
and infectious, the threshold and equilibrium of this model
are investigated. The locally and globally asymptotical
stable results of virus-free equilibrium and viral equilibrium
were being proved. Finally, some numerical examples are
given to demonstrate the analytical results.
Index Terms—Computer Virus; Equilibrium; Locally
Asymptotically; Simulation
I. INTRODUCTION
Computer virus is a malicious code which including
virus, Trojan horses, worm, logic bomb and so on. It is a
program that can copy itself and attack other computers.
And they are residing by erasing data, damage files, or
modifying the normal operation. In recent years, some
models of the computer virus is proposed [1-3, 5-12]. The
action of computer virus throughout a network can be
studied by using epidemiological models for disease propagation [4]. Based on the Kermack and Mckendrick
SIR classical epidemic model, dynamical models for
malicious objects propagation were proposed [5-11].
These dynamical modeling of the spread process of
computer virus is an effective approach to the
understanding of behavior of computer viruses because
on this basis, some effective measures can be posed to
prevent infection. The computer virus has latent ability. An infected
computer which is in latency, called exposed computer.
Based on these characteristics, delay is used in some
models to describe the latent ability of computer virus [7-
9]. And some models proposed SEIR or SLBS to describe
the latent of the computer virus [5-6, 10-11]. They are all
assuming that the computer which is newly entered in the
network are all infectious. In this paper, a novel model of computer virus, known as SEIR model, is put forward to
describe before enter in the network, the computer has
been infected computer virus by other ways such as
removable media etc. ( ), ( ), ( ), ( )S t E t I t R t denote the
classes of , , ,S E I R at time t . The total number of new
infectious at a time is given by SI , with as the mass
action coefficient and is used to incidence. In this paper,
we assume that a fraction a and b of the computers
which newly entered in the network flux into the exposed
class E and the infected class .I The computers which
have been infected computer virus flux into exposed class
E is given by aN and flux into infectious class I is
given by bN . The susceptible computer which flux into
the susceptible class S is given by (1 )aN bN . If
0 1a b , the model has only one viral equilibrium
and it is globally asymptotically stable; If 0a b , the
model has only one virus-free equilibrium if 0 1R , and
if 0 1R the system has another viral equilibrium and
they are globally asymptotically stable; If 1a b , the
model has only one viral equilibrium and it is globally
asymptotically. In this paper, the dynamics of this model
has been fully studied through some numerical examples. This paper is organized as follows. Section 2
formulates a novel computer virus model. Section 3,
investigate the stability of the equilibriums. Section 4
some numerical examples are given to present the
effectiveness of the theoretic results. Finally, Section 5
summarizes this work.
II. MODEL FORMULATIONS
At any time, a computer is classified as internal and external depending on weather it is connected to internet
or not. At that time all to the internet computers are
further categorized into four classes: (1) Susceptible
computers, that is, uninfected computers and new
computers which connected to network; (2) Exposed
computers, that is, infected but not yet broken-out; (3)
Infectious computers; (4) Recovered computers, that is,
virus-free computer having immunity. Let
( ), ( ), ( ), ( )S t E t I t R t denote their corresponding numbers
at time t, without ambiguity, ( ), ( ), ( ), ( )S t E t I t R t will be
abbreviated as , , ,S E I R , respectively. Our assumption
on the dynamical transfer of the computers is depicted in
Fig. 1, and the model is formulated as following system
of differential equations:
(1 ) ,
( ) ,
( ) ,
.
S a b N SI dS
E aN SI r d E
I bN rE d I
R I dR
(1)
( ) ( ) ( ) ( ) ( ).N t S t E t I t R t (2)
We may see that first three equations in (1) are
independent of the fourth equation, and therefore, the
JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014 367
© 2014 ACADEMY PUBLISHERdoi:10.4304/jnw.9.2.367-374
fourth equation can be omitted without loss of generality.
Hence, system (1) can be rewritten as
(1 ) ,
( ) ,
( ) .
S a b N SI dS
E aN SI r d E
I bN rE d I
(3)
Figure 1. Schematic diagram for the flow of virus in computer
network.
where N denotes the rate at which external computers
are connected to the network; a denotes the rate of the
new computer which has been infected computer virus
but hasn’t been broken; b denotes the rate of the new
computer which has been infected computer virus and it
is has been broken; denotes the rate at which, when
having a connection to one infected computer, one susceptible computer can be become exposed; r denotes
the rate of which, the exposed computer is broken and
come into the infectious state; denotes the rate
recovery rate of infected computers are cured; denotes
the rate at which one computer is removed from the
network. All the parameters are nonnegative. When having connection to one expose computer, one
susceptible computer can be become exposed, and its
infectious rate is SI ; aN is the number of the new
enter the network has been infected but hasn’t broken;
bN is the number of the new enter the network has been
infected;
Moreover, all feasible solutions of the system (3) are
bounded and enter the region D , where
3{( , , ) | 0, 0, 0, }.N
D S E I R S E I S E Id
Refer to [12], we analysis the threshold 0R of the
system (3) .
Let
( , , ) ,Tx E I S then, the system (3) can be written as
( ) ( ),i ix F x V x where ( ) 0 ,
0
i
SI
F x
( )
( ) ( )
0
i
r d E aN
V x rE d I bN
.
The Jacobin matrices of ( )F x and ( )V x at the
0 (1 )( , , )
a b NP aN bN
d
are
0 0( ) ,
0 0
N
F DF P d
0,
r dV
r d
then
101
,( )( )
dV
r r dd r d
Thus
1 0
0 ( )( )( )
S rR FV
r d d
(1 ).
( )( )
r a b N
d d r d
Especially, when 0,a b obviously 0,a b the
system (3) can be revised as
,
( ) ,
( ) .
S N SI dS
E SI r d E
I rE d I
(4)
And there
0 .( )( )
rNR
d d r d
For the system (4), if 0 1R , there always exists the
virus-free equilibrium is 0 0 0
1 1 1( ,0,0) ( ,0,0)N
P S Pd
; if
0 1R , there also exists an virus-free equilibrium is
0
1 ( ,0,0)N
Pd
and a viral equilibrium is
* * *
1 1 1
( )( ) ( )( , , ),
r d d dP I I
r r
there
*
1
( )( ).
( )( )
N r d r d dI
r r d d
When 1,a b the system (3) can be revised as
,
( ) ,
( ) .
S SI dS
E aN SI r d E
I bN rE d I
(5)
For the system (5), there has only one viral equilibrium
*
2
( )(0, , ).
( ) ( )( )
aN bN r d aNrP
r d r d d
When 0 1a b , the system (2.3) has only a viral
equilibrium * *
* * 3 3
3 3 *
3
[ ]( , , ).
( ) ( )( )
aN S I ar br bd NP S
r d d r d rS
Let
* * * *
3 3 3 3( ) (1 ) ,G S a b N S I dS (6)
368 JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014
© 2014 ACADEMY PUBLISHER
when *
3
(1 )(0, ),
a b NS
d
discuss the roots of equation
*
3( ) 0G S .
For system (3), we have * *
3 3( )G S d I
*
3
[ ]
( )( )
ar br bd Nd
d r d rS
*
3
* 2
3
[ ] .0,
[( )( ) ]
ar br bd N rSd
d r d rS
thus
we get 3( )G S
monotone decreasing in
(1 )(0, ).
a b N
d
Case (1): If 0
(1 )1,
( )( )
r a b NR
d d r d
from (6), we get
(0) (1 ) 0G a b N ,
(1 )( )
a b NG
d
* *
3 3 ,S I
there
*
3 *
3
[ ]0,
( )( )
ar br bd NI
d r d rS
thus
(1 )( )
a b NG
d
>0.
3( ) 0G S has no real root in (1 )
(0, )a b N
d
if
0 1.R
Case (2): If 0
(1 )1,
( )( )
r a b NR
d d r d
*
3 *
3
[ ]0,
( )( )
ar br bd NI
d r d rS
thus
(0) (1 ) 0,G a b N
(1 )( ) 0.
a b NG
d
For that 3( )G S monotone decreasing in
(1 )(0, ),
a b N
d
the equation 3( ) 0G S has only one
positive root if 0 1.R
From the case, the system (3) has only one viral
equilibrium * *
* * 3 3
3 3 *
3
[ ]( , , )
( ) ( )( )
aN S I ar br bd NP S
r d d r d rS
if 0 1.R
III. STABILITY OF THE EQUILIBRIUMS
Theorem 3.1.1. 0
1P is locally asymptotically stale
if 0 1R . Whereas 0
1P is unstable if 0 1.R
Proof. The Jacobin matrix of system (2.4) about 0
1P is
given by
0
1
0 0
1 1
0
0 ( ) ,
0 ( )
d S
J r d S
r d
which equals to
2 0
1 2 1( ) ( )( ) 0,f d a a S r (7)
where
1 ( ) ( ),a r d d
2 ( )( ).a r d d
then, Eq. (7) has negative real parts roots
1
2
1 1 2 0
2
2
1 1 2 0
3
0,
4 (1 )0,
2
4 (1 ),
2
d
a a a R
a a a R
when 0 1,R 3 0 , and When
0 1R , 3 0.
Then there are no positive real roots of (7), thus 0
1P is
a locally asymptotically stable equilibrium if 0 1,R and
0
1P is unstable if0 1.R
The proof is completed.
Theorem 3.1.2. 0
1P is globally asymptotically stale
with respect to D if 0 1R .
Proof. Let ,( )
rL E I
d
obviously 0,L
thus
( )
rL E I
d
0
[ ( ) ] ( )( )
( )( )
( )
( )( 1)
0.
rSI r d E rE d I
r d
Nr r d d
d r d
d R
The proof is completed.
Theorem 3.1.3. *
2P is globally asymptotically stable.
Proof. Let ,L S
obviously 0,L
then
0.L SI dS
The proof is completed.
Theorem 3.1.4 The viral equilibrium * * * *
1 1 1 1( , , )P S E I
and * * * *
1 3 3 3( , , )P S E I is locally asymptotically stable.
Proof. Denote * * * *
1 1 1 1( , , )P S E I and * * * *
1 3 3 3( , , )P S E I
as *P .
The Jacobin matrix of the system (3) about *P is given
by
JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014 369
© 2014 ACADEMY PUBLISHER
* *
* *
0
( ) * ,
0 ( )
I d S
J I r d S
r d
which equals is
3 2
0 1 2 3( ) 0,f A A A A (8)
where
0 1,A
*
1 {( ) [( ) ( )]},A I d r d d
* *
2 [( ) ( )]( ) [( )( ) ],A r d d I d r d d S r
* *
3 ( )( )( ) .A r d d I d S dr
Case (1): When 0a b and 0 1R ,
0 1 0,A
*
1 1{( ) [( ) ( )]} 0,A I d r d d *
2 1( )[( ) ( )] 0,A I d r d d
*
3 1( )( ) 0,A r d d I
then
1 2 3 0.A A A
According to the Hurwitz criterion, all roots of Eq. (8)
have negative real parts. Thus the equilibrium *
1P is
locally asymptotically stable.
Case (2): When 0 1a b and 0 1R ,
0 1,A *
1 3{( ) [( ) ( )]} 0,A I d r d d * *
2 3 3[( ) ( )]( ) [( )( ) ] 0,A r d d I d r d d S
* *
3 3 3( )( ) [( )( ) ] 0,A r d d I d r d d S r
then 1 2 3 0.A A A
According to the Hurwitz criterion, all roots of Eq. (8)
have negative real parts. Thus the equilibrium *
3P is
locally asymptotically stable. The proof is completed.
Now, let us examine the global stability of *P with
respect to the D by means of a classical geometric
approach [13]. For our proposed, the following obvious
result will be useful.
Lemma 3.1. If 0 1R , the system (3) is uniformly
persistent, i.e., there exists 1 0c (independent of initial
conditions), such that
1liminf ( ) ,t
S t c
1liminf ( ) ,t
E t c
1liminf ( ) .t
I t c
Remark 1. We have proved that the virus-free
equilibrium 0
1P is unstable if 0 1R when 0a b and
the system (3) has no virus-free equilibrium when
0 1a b . Furthermore, the instability of 0
1P , together
with 0
1P 0
1P D ( D denotes the boundary of D ),
imply the uniform persistence of the satiate variable. The
uniform persistence of system (3) is to show the existence
of a compact absorbing set in int D (int D denotes the
interior of D ), Which is a necessary condition for
proving the global stability by using the geometric
approach.
Theorem 3.1.4 *P is globally asymptotically stable if
0 1.R
Proof. The second compound matrix of the Jacobin
matrix *J can be calculated as follows [14-16]: * *
[2] *
*
( ) ( ) *
( ) ( ) 0 .
0 ( ) ( )
I d r d S S
J r I d d
I r d d
Set P as the following diagonal matrix
( ) (1, , ).E E
P xI I
Denote
1 (0, , ),f
E I E IP P diag
E I E I
Therefore, the matrix 1 [2] 1
fB P P AJ A can be
written in the following block form:
11 12
21 22
,B B
BB B
with *
11 ( ) ( ),B I d r d
*
12 (1,1),I
B SE
21 ( ,0) ,TEB r
I
*
22
*
( ) ( ) 0
,
( ) ( )
E II d r d
E IB
E II r d d
E I
thus
1 1 11 12
2 1 22 21
( ) ,
( ) ,
g B B
g B B
(9)
there *
1 11( ) ( ) ( ),B I d r d
* *
1 22( ) max{ ( ) ( ) ,E I
B I d r d IE I
( ) ( )}E I
r d dE I
= ( ) .E I
r d dE I
The vector norm in
3
23R R
is chose as
( , , ) {max , }.u v W u v w
The Lozinskii measure ( )B with respected to as
follows (17)
1 2( ) sup{ , }B g g ,
there
* *
1
2
( ) ( ) (1,1) ,
( ) ( ,0) .T
Ig I d r d S
E
E I Eg d d r
E I I
(10)
370 JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014
© 2014 ACADEMY PUBLISHER
Figure 2. Dynamical behavior of system (2.4) with 0.a b Time series of ( ), ( ), ( )S t E t I t with 0 1R .
Figure 3. Dynamical behavior of system (2.4) with 0.a b Time series of ( ), ( ), ( )S t E t I t with 0 1R .
Figure 4. Dynamical behavior of system (5) with 1a b . Time series of ( ), ( ), ( )S t E t I t .
JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014 371
© 2014 ACADEMY PUBLISHER
Figure 5. Dynamical behavior of system (5) with 0 1a b , Time series of ( ), ( ), ( )S t E t I t if
0.3, 0.3, 0.0001, 100, 0.8, 0.01, 0.8a b N r d n and then 0 4.9875 1R
Figure 6. Dynamical behavior of system (5) with 0 1a b , Time series of ( ), ( ), ( )S t E t I t if
0.1, 0.1, 0.0001, 100, 0.8, 0.001, 0.8a b N r d n and then 0 9.9750 1.R
Figure 7. Dynamical behavior of system (5) with 0 1a b , Time series of ( ), ( ), ( )S t E t I t if
0.1, 0.1, 0.0001, 100, 0.6, 0.001, 0.6a b N r d n and then 0 13.2890 1.R
372 JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014
© 2014 ACADEMY PUBLISHER
From the system (2.3), we find that
* ( ),
( ).
I E aNS r d
E E E
I bN Er d
I I I
(11)
Thus
*
1
2
( ),
.
E aNg I d
E E
E bNg d
E I
(12)
Relations (9)-(12) imply
( ) ,E
B dE
thus
0 0
1 1 1 ( )( ) ( ) ln .
(0)
t t E E tB d d d d
t t E t E
(13)
If 0 1R , then the virus-frees equilibrium is unstable
by the Theorem 3.1.2, Moreover, the behavior of the local
near 0D as describe in Theorem 3.1.2 implies that the
system (3) is uniformly persistent in D , i.e. there exists a
constant 1 0c and 0T , such that t T implies
1
1
1
liminf ( ) ,
liminf ( ) ,
liminf ( ) ,
t
t
t
S t c
E t c
I t c
And
1liminf[1 ( ) ( ) ( )]t
S t E t I t c
For all ( ( (0), (0), (0))S E I D [18, 19]
0
1limsupsup ( ) 0.
2
t
t x K
dq B d
t
(14)
The claimed result follows by combining Theorem
3.1.1, Lemma 3.1 and the negativity of .q
The proof is completed.
IV. SIMULATION
Now, we would analysis the locally asymptotically
stable of the equilibriums through some numerical
examples.
Case 1: For the system (4), if 0 1R , there exists the
virus-free equilibrium 0
1P . Let
0, 0, 0.0001, 10, 0.8, 0.01,a b N r d 0.8,n
and then 0 0.0122 1R , Fig. 2 shows the solution of
system (4) if 0 1R . Case 2: For the system (4), if 0 1R , there exists the
viral equilibrium *
1P . Let
0, 0, 0.001, 10, 0.5, 0.01, 0.5,a b N r d n
and then 0 19.92 1R . Fig. 3 shows the solution of
system (4) if 0 1R .
Case 3: For the system (5), there exists the viral
equilibrium *
2P , Let
0.5, 0.5, 0.01,a b 0.05, 0.001,r d 0.1,n Fig.
4 shows the solution of system (5) if 0a b .
Case 4: For the system (3), there exists the viral
equilibrium *
3P . Fig. 5-Fig. 7 shows the solution of
system (3) if 0 1a b .
From the Fig. 2, We can see that the virus-free
equilibrium 0
1P of system (4) is locally asymptotically
stable if 0 1R .
From the Fig. 3, We can see that the viral equilibrium *
1P of the system (4) is locally asymptotically stable
0 1R .
From the Fig. 4, We can see that the viral equilibrium *
2P of the system (5) is locally asymptotically stable.
From the Fig. 5-Fig. 7, We can see that the viral
equilibrium *
3P of the system (3) is locally asymptotically
stable.
For the system (3), from the Fig. 5 and Fig. 6, we can
see that the number of computer virus is not affected by
the parameters a and b . If we’re keeping a and b the
same, from the Fig. 6 and Fig. 7, we can see that the
number of computer virus actually decreases as we
reduce the 0R .
V. CONCLUSION
We considering the fact, the computers which is newly
entered in the network computer has been infected by
other ways such as removable media, a novel computer
virus model is proposed. Denotes a is the rate of the new
computer which has been infected computer virus but
hasn’t been broken; b denotes the rate of the new
computer which has been infected computer virus and it
is has been broken. If 0 1a b , the system (3) has
only one viral equilibrium and it is locally asymptotically
stable; If 0a b , the system (3) can rewritten as
system (4) and it has only one virus-free equilibrium if
0 1R , and if 0 1R the system has a viral equilibrium
and which are locally asymptotically stable; If 1a b ,
the System (3) can be rewritten as system (5) and it has
only one viral equilibrium which is locally asymptotically.
In this paper, the dynamics of this model has been fully
studied through some numerical examples.
REFERENCES
[1] Wei Yang, Gui-RAN Cheng, Yu Yao, Xiao-meng Shen,
Stability Analysis of P2P Worm Propagation Model with Dynamic Quarantine Defence, Journal of networks, Vol. 6, No. 1 (2011), pp. 153-162, Jan 2011.
[2] Juan Du, Su-li Chen, Research of Improved Network Malicious Information Filtering model Based on SVM, Journal of Networks, Vol. 8, No. 5 (2013), pp. 1012-1018, May 2013.
JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014 373
© 2014 ACADEMY PUBLISHER
[3] Hui Xu, Xiang Gu, Extensive Design for Attack’s Recognition and Resistance of Survivable Network.
Journal of Networks, Vol. 7, No. 2 (2012), 222-228, Feb 2012.
[4] C Sun, YH Hsieh Global analysis of an SEIR model with varying population size and vaccination. In: Applied Mathematical Modelling 34 (2010) pp. 2685-2697.
[5] LP Song, Z Jin, Gq Sun, Modelling and analysing of botnet interactions. In: Physical A 390 (2011) pp. 347-358.
[6] BK Mishra, SK Pandey Dynamic model of worms with vertical transmission in computer network. In: Applied Mathematics and Computation 217 (2011) pp. 8438-8446.
[7] JG Ren, XF Yang, LX Yang, YH Xu, FZ Yang, A delayed
computer virus propagation model and its dynamics, Chaos, Solutions& Fractals 45 (2012) pp. 74-79.
[8] X Han, Q Tan, Dynamical behavior of computer virus on Internet, Applied Mathematics and Computation xxx (2010) xxx-xxx.
[9] Q Zhu, X Yang, J Ren, Modeling and analysis of the spread of computer virus, Communications Nonlinear Science Numerical Simulation 17 (2012) pp. 5117-5124.
[10] LX Yang, X Yang, Q Zhu, L Wen. A computer virus model with graded cure rats, Nonlinear Analysis: Real World Applications.
[11] LX Yang, X Yang, L W, J L, A novel computer virus
propagation model and its dynamics, International Journal of Computer Mathematic, Vol. 00, No. 00, Month 200x, 1-8.
[12] P. Van den Driessche, J watmough, Reproduction numbers and sub-threshold endemic equilibrium for compartmental models of disease transmission, Mathematical Biosciences 180 (2002) pp. 29-48.
[13] Michael Y. Li and James S. Muldowney, A Geometric Approach to Global Stability Problems, SIAM J. Math. Anal, Vol. 27, No. 4, pp. 1070-1083, July 1996.
[14] M. Fiedler, Additive compound matrices and inequality for eigenvalues of stochastic matrices, Czechoslovak Math. J
99 (1974) pp. 392-402. [15] J. S. Muldowney, Compound matrices and ordinary
differential equations, Rocky Mountain J. Math 20 (1990) pp. 857-872.
[16] Martin Jr R H, Logarithmic norms and projections applied to linear differential system, J Math Anal App., 1974(45) pp. 432-454.
[17] G. J. Butler, P. Waltman, Persistence in dynamical systems, Proc. Amer. Math, Soc 96 (1986) pp. 425-430.
[18] H. I. Freedman, M. X Tang, S. G. Ruan, Uniform persistence and flows near a closed positive invariant set,
Journal of Dynamics and Differential Equations 6 (1994) pp. 583-600.
[19] P. Waltman, A brief survey of persistence in dynamical systems, In: S. Busenberg, M. Martelli (Eds.), Delay Differential Equations and Dynamical Systems, Springer-Verlag, New York, 1991, pp. 31-40.
Mei Peng is a lecturer in College of Mathematics and Computer Science, Yangtze Normal University, Chongqing, China. She received her master degree in stochastic system analysis from
the Chongqing Normal University, China. Her interest in the area of the computer virus propagation model.
Huajian Mou is a assistant in College of Mathematics and Computer Science, Yangtze Normal University, Chongqing, China. He interest in the area of the computer Software Programming.
374 JOURNAL OF NETWORKS, VOL. 9, NO. 2, FEBRUARY 2014
© 2014 ACADEMY PUBLISHER