Synchronization in Integer and Fractional Order Chaotic Systems

Synchronization in Integer and Fractional Order Chaotic Systems

Ahmed E. Matouk (Mansoura University, Egypt; Hail University, Saudi Arabia)
DOI: 10.4018/978-1-61520-737-4.ch007

Abstract

In this chapter, the author introduces the basic methods of chaos synchronization in integer order systems, such as Pecora and Carroll method and One-Way coupling technique, applying these synchronization methods to the modified autonomous Duffing-Van der Pol system (MADVP). The conditional Lyapunov exponents (CLEs) are also calculated for the drive and response MADVP systems which match with the analytical results given by Pecora and Carroll method. Based on Lyapunov stability theory, chaos synchronization is achieved for two coupled MADVP systems by finding a suitable Lyapunov function. Moreover, synchronization in fractional order chaotic systems is also introduced. The conditions of Pecora and Carroll method and One-Way coupling method in fractional order systems are also investigated. In addition, chaos synchronization is achieved for two coupled fractional order MADVP systems using One-Way coupling technique. Furthermore, synchronization between two different fractional order chaotic systems is studied; the fractional order Lü system is controlled to be the fractional order Chen system. The analytical conditions for the synchronization of this pair of different fractional order chaotic systems are derived by utilizing the Laplace transform theory. Numerical simulations are carried out to show the effectiveness of all the proposed synchronization techniques.
Chapter Preview
Top

1. Synchronization In Integer Order Chaotic Systems

1.1 Pecora and Carroll Method

Let us consider the system

978-1-61520-737-4.ch007.m01
(1)

By dividing system (1) into two parts arbitrarily as

978-1-61520-737-4.ch007.m02
Where xD are the drive subsystem variables, 978-1-61520-737-4.ch007.m03, and xR are the response subsystem variables, 978-1-61520-737-4.ch007.m04. Now, equation (1) is written as
978-1-61520-737-4.ch007.m05
(2) where 978-1-61520-737-4.ch007.m06 and 978-1-61520-737-4.ch007.m07

If we make a copy of the response subsystem called 978-1-61520-737-4.ch007.m08 which driven by the variables xD (from the original system), it follows that; the system

978-1-61520-737-4.ch007.m09
(3) drives the response system

978-1-61520-737-4.ch007.m10
(4)

Following the suggestion of Pecora and Carroll, as the variation of the time the 978-1-61520-737-4.ch007.m11 variables will converge asymptotically to the xR variables and remain with it in step instantaneously. Our goal is to make the response system a stable subsystem in order to obtain synchronization. Suppose we have trajectory 978-1-61520-737-4.ch007.m12 with initial condition 978-1-61520-737-4.ch007.m13 then if a trajectory started at a nearby point 978-1-61520-737-4.ch007.m14 at t = 0, then the conditions needed for synchronization implies that

978-1-61520-737-4.ch007.m15

This leads to

978-1-61520-737-4.ch007.m16
(5) where 978-1-61520-737-4.ch007.m17 is the Jacobian of the response vector field with respect to the response variables and 978-1-61520-737-4.ch007.m18 represents the higher order terms. For small 978-1-61520-737-4.ch007.m19equation (5) becomes:

978-1-61520-737-4.ch007.m20
(6)

In Ref. (10), Pecora and Carroll used a matrix Z in place of 978-1-61520-737-4.ch007.m21 in Eqn. (6), such that Z(0) is equal to identity matrix. Thus

978-1-61520-737-4.ch007.m22
(7)

The solution Z(t) is called the transfer function or the principal matrix solution (Pecora & Carroll, 1991). Z(t) will determine whether perturbations will grow or shrink in any particular directions. Using the principal matrix solution Z(t) as 978-1-61520-737-4.ch007.m23 we can determine the Lyapunov exponents for 978-1-61520-737-4.ch007.m24 subsystem for a particular drive trajectory xD, which help us to estimate the average convergence rates and their associated directions.

Now, calculating Z(t) for large t from the variational equation (7), the Lyapunov exponents are given by:

978-1-61520-737-4.ch007.m25
(8) where 978-1-61520-737-4.ch007.m26 are the eigenvalues of Z(t), each associated with an eigenvector 978-1-61520-737-4.ch007.m27 (Pecora & Carroll, 1991).

  • Definition 1. The Lyapunov exponents for 978-1-61520-737-4.ch007.m28 subsystem (see equation (4)) are called sub Lyapunov exponents or conditional Lyapunov exponents (CLEs).

Now, we have the following theorem:

  • Theorem 1.(Pecora and Carroll) According to Pecora and Carroll, the coupled chaotic systems can be regarded as drive and response systems which will perfectly synchronize only if CLEs are all negative (Carroll & Pecora, 1991; Pecora & Carroll, 1990).

The above theorem is necessary, but not sufficient condition for synchronization. It says nothing about the set of initial conditions between the drive and response systems.

However, synchronization can be achieved even with positive CLEs (Güémez et al, 1997). Intermittent synchronization can occur when CLEs are very small positive or negative values close to zero, while permanent synchronization occurs when CLEs take sufficiently large negative values.

Now, suppose that 978-1-61520-737-4.ch007.m29 and xR are subsystems under perfect synchronization. Hence, if we define 978-1-61520-737-4.ch007.m30 then 978-1-61520-737-4.ch007.m31.

Assume that the subsystem is linear, then

978-1-61520-737-4.ch007.m32
(9) where J is a (m×m) constant matrix. The real parts of the eigenvalues of the matrix J defined in equation (9), are the conditional Lyapunov exponents (CLEs) we seek.

In the following we are going to apply Pecora and Carroll method to a modified autonomous Duffing-Van der Pol system (Matouk & Agiza, 2008; Matouk, 2005), and known here as MADVP. This system is described by the following equations:

978-1-61520-737-4.ch007.m33
(10)

The parameters 978-1-61520-737-4.ch007.m34 are all positive real parameters and 978-1-61520-737-4.ch007.m35. System (10) exhibits chaotic behaviors at the parameter values 978-1-61520-737-4.ch007.m36 and 978-1-61520-737-4.ch007.m37 (see Figure 1).

Figure 1.

Phase portrait of system (10), showing the double-band chaos, using the parameter values 978-1-61520-737-4.ch007.m38 and 978-1-61520-737-4.ch007.m39.

978-1-61520-737-4.ch007.f01

1.1.1 Pecora and Carroll Method for MADVP System

The x-drive system is given by equations (10) will be used to drive the following response system:

978-1-61520-737-4.ch007.m40
(11)

The difference system for 978-1-61520-737-4.ch007.m41 and 978-1-61520-737-4.ch007.m42 in matrix form is

978-1-61520-737-4.ch007.m43
(12) Or in vector notation
978-1-61520-737-4.ch007.m44
where

978-1-61520-737-4.ch007.m45 and 978-1-61520-737-4.ch007.m46

The real parts of the eigenvalues of the matrix J are the conditional Lyapunov exponents (CLEs). Then the response system synchronizes if all the real parts of the eigenvalues of J are negative.

The characteristic equation of (12) is

978-1-61520-737-4.ch007.m47
then the roots are given as follows

978-1-61520-737-4.ch007.m48

It is clear that the term 978-1-61520-737-4.ch007.m49 is always less than 978-1-61520-737-4.ch007.m50 then we have two cases;

Case (i) if 978-1-61520-737-4.ch007.m51

This means that the two roots will be real and negative and the system with x-drive configuration does synchronize.

Case (ii) if 978-1-61520-737-4.ch007.m52

This is the case used in observing the chaotic motion of this system. The eigenvalues will be 978-1-61520-737-4.ch007.m53

Hence all eigenvalues have negative real parts, which imply that all solutions tend to zero as t tends to ∞. The conditional Lyapunov exponents (CLEs) are 978-1-61520-737-4.ch007.m54 and as expected, 978-1-61520-737-4.ch007.m55 and 978-1-61520-737-4.ch007.m56 then the response system with x-drive configuration does synchronize.

1.1.2 Numerical Results

By solving the drive and response systems (10) and (11) numerically and using the above-mentioned parameter values, we find that in spite of the differences in initial conditions of 978-1-61520-737-4.ch007.m57 and 978-1-61520-737-4.ch007.m58, the response system synchronizes so that for 978-1-61520-737-4.ch007.m59 and 978-1-61520-737-4.ch007.m60 (see Figure 2).

Figure 2.

(a-b) shows that the synchronization error tends to zero using Pecora and Carroll method

978-1-61520-737-4.ch007.f02

Now, we will calculate the conditional Lyapunov exponents (CLEs) numerically. We used a MATLAB code based on the Eckmann and Ruelle QR decomposition technique to calculate the exponents (Eckmann & Ruelle, 1985). Table (1) shows a calculation of the CLEs for various subsystems. From the table we see that, the subsystem (y, z) driven by x is stable subsystem as we expected and the numerical values of its CLEs are close to the analytical values shown above 978-1-61520-737-4.ch007.m61. We also deduce that synchronization can not occur for z-drive configuration because the largest CLE of the (x, y) subsystem is sufficiently positive (see Table 1).

Table 1.
Conditional Lyapunov exponents for various drive-response configurations for the MADVP system with the parameter set 978-1-61520-737-4.ch007.m62, 978-1-61520-737-4.ch007.m63.
DriveResponseCLEs
x(y, z)-0.7999, -0.8004
y(x, z)0.0021, -12.053
z(x, y)+6.0177, -19.4448

Complete Chapter List

Search this Book:
Reset