IEICE TRANS. FUNDAMENTALS, VOL. E89-A, No.

5 May 2006

1500-1504

Letter on FUNDAMENTALS

2-D Laplace-Z Transformation

Yang Xiao , member and Moon Ho Lee, nonmember
Summary
Based on recent results for 2-D continuous-discrete systems, this paper develops 2-D Laplace-z transform, which can be used to analyze 2-D continuous-discrete signals and system in Laplace-z hybrid domain. Current 1-D Laplace transformation and z transform can be combined into the new 2-D s-z transform. However, 2-D s-z transformation is not a simple extension of 1D transform, in 2-D case, we need consider the 2-D boundary conditions which dont occur in 1-D case. The hybrid 2-D definitions and theorems are given in the paper. To verify the results of this paper, we also derived a numerical inverse 2-D Laplace-z transform, applying it to show the 2-D pulse response of a stable 2-D continuous-discrete system. Laplace-z transform to treat 2-D continuous-discrete signals and systems independent of the particular field of application and taking into account both 1-D results and 2-D invariant results.

2. Definitions and properties of 2-D s-z transformation

For 2-D continuous-discrete signals and systems, we establish the following definitions. Definition 1: For a 2-D continuous-discrete signal x(t , n) , its 1-D Laplace transformation for the first variable of defined as

x(t , n)

is

Key words:
2-D Laplace-z transform, 2-D continuous-discrete signals, 2-D continuous-discrete systems.

X ( s, n) = x(t , n)e st dt
0

(1)

1. Introduction
Classical multidimensional (M-D) signals and systems divided M-D continuous signals and systems and M-D discrete ones. Corresponding 1-D continuous signals and systems, we have 1-D Laplace transform and z transform. Extending 1-D Laplace and z transformation to the M-D case, we have M-D Laplace transform and M-D z transform [1-3], and many papers such as [4-10] have studied the systems theory with only the continuous or discrete case. Unfortunately, no paper considered the M-D signals or M-D systems to be hybrid: partial variables of some dimensions of the M-D signals or systems are continuous, and others are discrete, until the work of Kaczorek [15-17], in which 2-D continuous-discrete systems were introduced. However, Kaczorek limited his analysis about the solutions, controllability and observability of 2-D continuous-discrete systems in 2-D time-discrete space domain, thus he did not consider the 2-D Laplace-z transformation. The question of getting the results in 2-D Laplace-z transform domain was explored in [18-21] directly. Applying 2-D Laplacez transformation in system analysis, Ref [18]-[21] got the results of stability, robust stability, controllability and observability of the systems. In this paper, we try to make 2-D Laplace-z transform to be perfect as presented 1-D Laplace or z transform, since 2-D continuous-discrete systems arise also in signal processing whenever the output of a 2-D filter is sampled in one direction on the discrete plane. With this motivation we will show 2-D
Manuscript received October 15,2005 The author is with Institute of Information Science, Beijing Jiaotong University, Beijing 100044, China, E-mail: yxiao@center.njtu.edu.cn. The author is with Institute of Information & Communication, Chonbuk National University, Jeonju 561-765, Korea. E-mail: moonho@chonbuk.ac.kr *This project is supported by the National Natural Science Foundation of China under grant: 60572093 and Brain Pool Program of Korea under grant: 051S-3-5.

We express 1-D partial Laplace transformation of

x(t , n)
(2)

X ( s, n) = Lt [ x(t , n)]
Definition 2: For a 2-D hybrid signal
+ j

X ( s, n) , its inverse 1-D

partial Laplace transformation is defined as

x(t , n) =
where the

X ( s , n)e
j

st

ds

(3)

> 0 is an arbitrary real number which is greater than

part of any singularity of

real

X ( s, n)

and

j = 1 denotes the imaginary unit.

We express the inverse 1-D Laplace transformation of for the first variable of

X ( s, n)
(4)

X ( s, n) x(t , n) = L [ X ( s, n)]
1 t

Definition 3: For a 2-D continuous-discrete signal 1-D z transformation for the second variable defined as

x(t , n) , its of x(t , n) is

(5)

X (t , z ) = x(t , n)z n
n =0

We express 1-D z transformation of

x(t , n)
(6)

X (t , z ) = Z n [ x(t , n)]
Definition 4: For a 2-D hybrid signal z transformation is defined as

X (t , z ) , its inverse 1-D

(7)

x(t , n) =

1 X (t , z ) z n 1 dz 2j r

IEICE TRANS. FUNDAMENTALS, VOL. E89-A, No. 5 May 2006

We express the inverse 1-D z transformation of

X (t , z )
(8)

x(t , n) = Z [ X (t , z )]
Definition 5: For a 2-D continuous-discrete signal 2-D Laplace-z transformation is defined as

1 n

Property 5: If the 2-D Laplace-z transformation of continuous-discrete signal x(t , n) is X ( s, z ) ,

2-D and

x(t , n) , its

x(t ,0) 0 , then Lt Z n [x(t , n + 1)] = zX ( s, z ) X ( s,0)

Proof: Similarly the proof of the Property 2, we have

(18)

X ( s, z ) = x(t , n)e st dtz n

n =0 0

(9)

Z n [x(t , n + 1)] = x(t , n + 1) z n1dz

r

(19)

We express 2-D Laplace-z transformation of

x(t , n)
(10)

X ( s, z ) = Lt Z n [ x(t , n)]
Definition 6: For a 2-D continuous-discrete signal inverse 2-D Laplace-z transformation is defined as
+ j

= zX (t , z ) x(t ,0)
Further take Laplace transformation to (19), we obtain

x(t , n) , its

Lt [zX (t , z ) x(t ,0)] = zX ( s, z ) X ( s,0)

x(t , n) = X ( s, z )

1 X ( s, z )e st z n1dsdz 2j r j

(11)

QED. Property 6: If the 2-D Laplace-z transformation of continuous-discrete signal x(t , n) is X ( s, z ) , then

2-D

We express the inverse 2-D Laplace-z transformation of

x(t , n) = ( Lt Z n ) 1[ X ( s, z )]
x1 (t , n) x 2 (t , n)

(12)

x(t , n + 1) Lt Z n t = szX ( s, z ) sX ( s,0) zX (0, z ) + x(0,0)

Proof: Using partial integration approach, we have

(20)

Following properties will be used. Property 1: If the 2-D Laplace-z transformations of 2-D continuous-discrete signal and are

x(t , n + 1) st x(t , n + 1) Lt e dt = t t 0 = x(t , n + 1)e st

0

X 1 ( s, z ) and X 2 ( s, z ) , respectively, then Lt Zn [ax1 (t, n) + bx2 (t, n)] = aX1 (s, z) + bX 2 (s, z)

(13)

+ s x(t , n + 1)e st dt
0

(21)

Property 2: If the 2-D Laplace-z transformation of 2-D continuous-discrete signal x(t , n) is X ( s, z ) , then

= sX ( s, n + 1) x(0, n + 1)
Further apply z transformation to (21), we obtain

Lt Z n [ x(t , n)] = Z n Lt [ x(t , n)]

(14)

Z n [sX ( s, n + 1) x(0, n + 1)]

Property 3: If the 2-D Laplace-z transformation of 2-D continuous-discrete signal x(t , n) is X ( s, z ) , then

= szX ( s, z ) sX ( s,0) zX (0, z ) + x(0,0)

QED. Due to limited space, we could not provide all the properties of 2-D Laplace-z transformation. However, other properties can be derived by similar procedures.

( Lt Z n ) 1[ X ( s, z )] = ( Z n Lt ) 1[ X ( s, z )] x(t , n) Lt Z n = sX ( s, z ) X (0, z ) t
Proof: Using partial integration approach, we have

(15)

Property 4: If the 2-D Laplace-z transformation of 2-D continuous-discrete signal x(t , n) is X ( s, z ) , then (16)

3. Laplace-z Transformation of 2-D continuous-discrete systems

Some practical systems need to be expressed by the model of 2-D continuous-discrete systems [15-21]. Similar to 1-D systems, there are also controllability and reachability [15-17, 21]. Consider following the Fornasini-Marchesini model of 2-D Continuous-Discrete Systems [3,15-17,21]

x(t, n) x(t, n) st Lt = e dt 0 t t = x(t, n)e

st 0

+ s x(t, n)e dt = sX(s, n) x(0, n)

st 0

(17)

Further take z transformation to (17), we obtain

Z n [sX ( s, n) x(0, n)] = sX ( s, z ) X (0, z )

QED.

x(t , n) x(t , n + 1) = A 0 x(t , n) + A 1 t t + A 2 x(t , n + 1) + Bu(t , n) y (t , n) = Cx(t , n)

(22a)

(22b)

Yang Xiao and Moon Ho Lee : 2-D Laplace-z transform

where

x(t , n) R N is state vector u(t , n) R M is the input vector y (t , n) R K is output input vector C R K N .

of of

the the

system, system,

Step 2:

Apply the numerical inversion of Laplace

( s, n) , for transform to the first variable of X

0 t < 2T selecting h = 2T / N , t = kh, k = 1,..., N ,
we have

of the system,

(t, n) x = ( , n) et N 1 k kt X Re X ( + j , n) exp( j ) N k =0 T T 2
(28)

Take 2-D Laplace-z transformation to the variables t and n of (22) according to Property 6, we get

szX( s, z ) sX( s,0) zX(0, z ) + x(0,0)

= A 0 X( s, z ) + A1[ sX( s, z ) X(0, z )] + A 2 [ zX( s, z ) X( s,0)] + BU ( s, z ) Y( s, z ) = CX( s, z )

X( s, z ) = (szI A 0 sA1 zA 2 ) 1 [sX(s,0) + zX(0, z ) x(0,0)] + ( szI A 0 sA1 zA 2 ) BU( s, z )
1

(23a)

Substitute (27) into (28), we get final result for the numerical

inversion of 2-D Laplace-z transform, (t, n) x

(23b)

Rearrange (23a), we get the s-z domain solution of the 2-D continuous-discrete system (22) due to Property 1-Property 6,

(24)

2m k kt 2mn et N 1 M 1 )) exp(j ) +j Re X ( + j , exp(j M T M T N k =0 m=0 2m 2mn 1 M 1 ( , exp(j ) exp(j ) Re X M 2 m=0 M (29) =

It is easy to extend the above result to the case of vector

Take inverse 2-D Laplace-z transformation to (23), we can get the t-n domain solution of the 2-D continuous-discrete system (22)

(t , n) for X( s, z ) . x
Now, we can use it as a tool for analysis of the 2-D invariant continuous-discrete systems. Applying 2-D Laplace-z transformation to (22), we can get the characteristic polynomial of system (22) to be

x(t , n) =

1 X( s, z )e st z n1dsdz 2j r j

+ j

(25)

B ( s, z ) = det[ szI A 0 sA 1 zA 2 ]
It is important for the characteristic polynomial stable. We construct a transfer function based on

(30)

where X( s, z ) is given by (24).

4. The numerical algorithm of inversion of 2D Laplace-z transforms

To solve (25) we need to extend the presented numerical algorithms for 1-D Laplace transform such as [22] and 1-D z transform to 2-D hybrid case. For simplicity, we consider the x(t , n) and X ( s, z ) . We note that the variables t and n in

B ( s, z ) to B ( s, z ) ,

be

H ( s, z ) =

1 B ( s, z )

(31)

If there are no second nonsingularities for the system (22), the stability of B ( s, z ) is a necessary and sufficient condition for the system (22). Take inverse 2-D Laplace-z transformation to (31), we have a 2-D impulse response

x(t , n) are independent each other, so are the variables s and z in x( s, z ) . Thus, the inversion of X ( s, z ) can be performed
in two steps. Step 1: Take the inversion z transform to the second variable of X ( s, z ) , we have
r=|z|=1 1 1 X(s, n) = X(s, z)zn1dz = X(s, e j )e j(n1)dej 2j r 2j 0 2

1 1 ( LZ ) [ H (s, z )] = e st z n1dsdz 2j r j B(s, z )

1

+ j

(32)

= h(t , n)

1 = X(s,e j )e jnd 2 0
Let

(26)

5. Examples
In this section, we will demonstrate the applications of the results of last sections. Example 1: Consider the 2-D continuous-discrete signal and its 2-D Laplace-z transformation:

2m , then the above integration can be approximated M 2m j 1 M 1 2mn ) (27) X ( s, n) = X ( s, e M ) exp( j M M m =0

IEICE TRANS. FUNDAMENTALS, VOL. E89-A, No. 5 May 2006

h(t , n) = 0.5 n cos(0.3t ) sz H ( s, z ) = 2 ( s + 0.09)( z + 0.5)

(33)

and apply the proposed inverse 2-D Laplace-z transformation algorithm to H ( s, z ) in (33), we get the h(t , n) as shown

in

Fig.

1,

which
n

is

agreed

with

the

analysis

Remarks: Different from [17], we get the solution of the 2-D system by using 2-D Laplace-z transform instead of partial differential and difference equation (35). The obtained 2-D pulse response in Fig.2 is the numerical solution of the system equation (35) with the initial state h(0,0) = 1 . Here, instead of solving (35) directly, we apply the proposed results in previous sections. The 2-D Laplace-z transform of this paper provides a tool to

result h(t , n) = 0.5 cos(0.3t ) in (33). Thus, it verifies the proposed the inverse 2-D Laplace-z transformation to be
valid.

obtain analysis and numerical solutions of 2-D continuous and discrete systems.

Fig.2 Fig.1 The result of proposed inverse 2-D Laplace-z

2-D pulse response of the system (35)

6. Conclusions
transformation for

sz . 2 ( s + 0.09)( z + 0.5)

The following example will be more complicated, since there is no easy approach to get the analysis solution of such 2-D continuous-discrete system in (35). Example 2: Assume that a 2-D continuous-discrete system with following 2-D hybrid transfer function [20],

H (s, z) =

1 (12 + 6 s) + (10 + 5 s)z-1 + (2 + s)z 2

(34)

determining the partial differential and difference equation and the 2-D pulse response of the system. The 2-D hybrid transfer function in (34) is corresponding to the following system equation

Based on recent results for 2-D continuous-discrete systems, this paper develops 2-D s-z transformation, the definitions, theorems and numerical inversion of 2-D Laplace-z transform inverse are given in the paper, which can be used to analyze 2-D continuous-discrete signals and system in s-z hybrid domain. Current 1-D Laplace transformation and z transformation can be combined into the new 2-D s-z transformation on consider the 2-D boundary conditions. An example for 2-D pulse response of a 2-D continuous-discrete system is given to show the results of the paper to be valid. The application of the results of this letter to 2-D Markov chain systems [23-25] is a subject for further research.

References
[1] [2] N.K. Bose, Editor, Multidimensional Systems: Theory and Applications, IEEE Press, N.Y.,1979. X. Du, Y. Xiao, et al. Multidimensional Digital Filters. National Defense Industry Publishing House, Beijing, 1995. Y. Xiao, Stability Analysis of Multidimensional Systems, Shanghai Science and Technology Press, Shanghai, 2003. Y. Xiao, X. Du, R. Unbehauen, Improved 2-D stability margin test for 2-D discrete systems, Journal of Systems Science and Systems Engineering, Vol.7, No.2, pp.228-233. 1998. Y. Xiao, R. Unbehauen and X. Du, Sufficient conditions of robust Schur stability for 2-D polynomials, Journal of Systems Science and Systems Engineering,Vol.8, No.3, pp.368-374, 1999. Y. Xiao, R. Unbehauen, X. Du.,Schur stability of polytopes of bivariate polynomials, IEEE Transactions on Circuits and

h(t , n) h(t , n 1) 12 + 6 + 10h(t , n 1) + 5 t t h(t , n 2) + 2h(t , n 2) + =0 t

(35)

[3] [4]

Applying the algorithm of the numerical inversion of 2-D Laplace-z transform given by (29), we can get the 2-D pulse response of the 2-D continuous-discrete system, shown in Fig.2. We can see that the 2-D system is stable since the 2-D pulse response h(t , n) of the system is finite and towards to zero when [20].

[5]

t , n , the conclusion is agreed with that of

[6]

Yang Xiao and Moon Ho Lee : 2-D Laplace-z transform

Systems I: Fundamental Theory and Applications, Vol. 49, No. 7, pp. 1020 1023, July 2002.

[7]

[8]

[9]

[10]

[11]

[12] [13]

[14] [15]

[16]

[17] [18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

Y. Xiao, R. Unbehauen, X. Du. Robust Hurwitz stability conditions of polytopes of bivariate polynomials, Proceedings of 38th IEEE Conference on Decision and Control, Phoenix, pp.50345035, AZ, 1999. Y. Xiao, Schur stability of interval bivariate polynomials, Proceedings of IEEE International Symposium on Circuits and Systems (ISCAS2000), pp.I-527-530, Geneva, 2000. Y. Xiao and R. Unbehauen, New stability test algorithm for twodimensional digital filters, IEEE Trans. Circuits and Systems I: Fundamental Theory and Applications, Vol. 45, No., pp. 739-740, 1998. D. L. Davis. A correct proof of Huang's theorem on stability, IEEE Trans. on Acoustics, Speech, and Signal Processing, Vol. ASSP-24, No.10, pp. 425-426, 1976. W. Lu and E. Lee, Stability analysis for two-dimensional systems, IEEE Transactions on Circuits and Systems, Vol. 30, No. 7, pp. 455-461, 1983. M. Bisiacco, On the algebraic theory of 2D state observers, Systems and Control Letters, vol. 5, pp. 347353, 1985. M. Bisiacco, State and Output Feedback Stabilizability of 2D Systems, IEEE Trans. Circuit and Systems, vol. 32, pp. 1246 1254, 1985, M. Bisiacco and E. Fornasini, Optimal control of two dimensional systems, SIAM J. Contr. and Optim., vol. 28, pp. 582601, 1990, T. Kaczorek, Stabilization of singular 2D continuous-discrete systems by state-feedbacks controllers, IEEE Trans. Automatic Control, AC-41, No.7, pp.1007-1009, 1996. T. Kaczorek, Stabilization of singular 2D continuous-discrete systems by output-feedbacks controllers, SAMS, Vol.28, pp.21-30, 1997. T. Kaczorek, Positive 2D continuous-discrete linear systems, in Advances in Control, Springer, London, 1999, pp.309-316. Y. Xiao, Robust Hurwitz-Schur stability conditions of polytopes of 2-D polynomials. Proceedings of 40th IEEE Conference on Decision and Control, Orlando, pp. 3643-3648, 2001. Y. Xiao, Hurwitz-Schur stability of interval of bivariate polynomials, Proceedings of IEEE International Symposium on Circuits and Systems (ISCAS2001), Sydney, pp. 829-832, 2001. Y. Xiao. Stability test for 2-D continuous-discrete systems, Proceedings of 40th IEEE Conference on Decision and Control, Orlando, pp. 3649-3654, 2001. Y. Xiao. Stability, controllability and observability of 2-D continuous-discrete systems, Proceedings of IEEE International Symposium on Circuits and Systems, Bangkok, Thailand, Vol. 4, pp.468-471, 2003. A. Yonemoto, T. Hisakado, L. Okumura, Accuracy improvement of the FFT-based numerical inversion of Laplace transforms Circuits, Devices and Systems, IEE Proceedings Vol.150, No. 5, pp. 399-404, 2003. N. F. Mir, Analysis of fault-tolerant multipath crossbars, IEEE Global Telecommunications Conference, San Francisco, Vol. 5, pp.2908 2912, 2003. M. Yang, P. H. J. Chong, Performance analysis of a cellular system supporting two types of services with different reuse factors using channel partitioning , Proceedings of Wireless Telecommunications Symposium 2004, California, pp.39 43, 2004 J So, D-H Cho, Performance analysis of a DS/SSMA unslotted ALOHA system with two user classes, IEEE Transactions on Vehicular Technology, Vol. 51, No. 6, pp. 1628 1639, 2002.