I

IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 39, NO. 5, MAY 1994 995

Application of the results of [SI yields the bounds [7] E. Yaz, “Deterministic and stochastic robustness measures for discrete
systems,” IEEE Trans. Automat Contr., vol. 33, pp. 952-955, 1988.
I ~ :P max{lp~zl,1 ~ 2 1 1 ,1 ~ 3 2 1 ) < 0.212 [8] S. R. Kolla, R. K. Yedavalli, and J. B. Farrison, “Robust stability
bounds on time-varying perturbations for state-space models of linear
discrete-time systems,” Znt. J. Contr., vol. 50, pp. 151-159, 1989.
I2s : [(pi’ F 0.085)’ + (pzi 0.352)’ [9] J. W. Brewer, “Kronecker products and matrix calculus in system
+ ( p x 7 0.059)2]”2 < 0.668 (6.8) theory,” ZEEE Trans. Circuits Syst., vol. CAS-25, pp. 772-780, 1978.
[IO] M. E. Sezer and D. D. siljak, “A note on robust stability bounds,” ZEEE
where the constants in 0 s have the same sign as the corresponding Trans. Automat Contr., vol. 34, pp. 1212-1214, 1989.
parameters. Note that [8] has no counterpart of 20 and 52p is inferior [I I] M. Araki, “Stability of large-scale systems: Quadratic order theory of
composite-system method using M-matrices,” ZEEE Trans. Automat
to ap. Also, the largest sphere with its center at the origin which is Contr.: vol. AC-23, pp. 129-142, 1978.
included in 0 s is given by [ 121 D. D. Siljak, Large-Scale Dynamic Systems. New York North-Holland,
1978.
(6.9)
and is smaller than a?.
On the other hand, the composite Lyapunov function approach of
[IO] yields the aggregate matrix

II’ = -5.S641p21
1
I
-3.5981P121
1
;] (6.10)
Design of Robust Controllers for Time-Delay Systems

[ 0 -1.8861p321 1 Magdi S. Mahmoud and Naser F. AI-Muthairi

Note that the block-triangular structure of 2, is reflected in the

structure of W.The robustness region is specified by the AI-matrix Abstract- The problem of stabilizing linear dynamical time-delay
conditions on I T ’ as systems subject to bounded uncertainties is investigated. Two memo-
ryless feedback controllers are considered. It is established that when
(2’’ : lp12p211 < 0.047. (6.1 1) the matching conditions are met and certain bounding relations are
satisfied, then the linear controller renders the zero-responseof the system
When the -11-matrix conditions are replaced by the stronger condition asymptotically stable. Saturation-typecontrollers are shown to guarantee
in (5.16), we obtain the estimate that all system responses are uniformly ultimately bounded.

0: : 3.5981p121 + 5.8641p211 + 1.8861p321 < 1. (6.12)

I. INTRODUCTION
We note that 5ILv is independent of p32 as 52 in (6.3) is, and in this A major problem in the analysis of linear dynamical systems
sense, is superior to the closed regions in (6.5) and (6.7). However, with time-delay is related to their stabilization using linear feedback
a further majorization as in (5.16) eliminates this advantage, as can with or without memory. Several results are readily available in
be observed from (6.12). the literature; see [1]-[6] and the references cited therein. Some of
We also note that each of the estimates in (6.5)-(6.7) can be further the results have been successfully extended to include the effect
expanded by repeated application of the robustness analysis to a of bounded uncertainties [7]-[ 101. Preliminary investigations on
modified system obtained by moving the nominal system to a point on discrete-time systems with state delay are found in [11]-[13]. A
the boundary of the robustness region and redefining the perturbation thorough review of the major past works of the deterministic approach
parameters accordingly. However, since this process destroys the to uncertain system based on the constructive use of Lyapunov
subsystem versus interconnection structure, it may not be suitable function can be found in [14], [15]. If the uncertainty fits a certain
for interconnected systems. characterization (often termed the matching condition), then a class of
We finally note that a scaling of the perturbation parameters and feedback controllers can be designed based only on the upper bound
the corresponding perturbation matrices may be useful in obtaining of the uncertainty [16]. When dealing with time-delay systems, it
improved robustness bounds, as noted in [IO]. turns out [7], [9] that an additional assumption is needed to guarantee
smooth behavior of the closed-loop state trajectories.
REFERENCES Perhaps the work of Thowsen [7] and Yu [SI were among the
early investigations to include the effect of bounded uncertainties
R. E. Kalman and J. E. Bertram, “Control system analysis and design
via the second method of Lyapunov,” Trans. Amer. Soc. Mech. Eng. J. on dynamical systems with time delay. The results of [7] required
Basic Eng., vol. 82, pp. 371-393, 1960. a bounding assumption to hold for all possible solutions not just
R. V. Patel and M. Toda, “Quantitative measures of robustness for multi- the nominal. This restriction was relaxed in [9] and applied to
variable systems,” in Proc. Joint Automat. Contr. Con$, San Francisco, river pollution control. Linear uncertain systems with state delay
CA, 1980, paper ”8-A.
R. K. Yedavalli, “Improved measures of stability of robustness for
were treated in [8] and the stabilizing controller had limited gain
linear state space models,” IEEE Trans. Automat Contr., vol. AC-30, adjustment.
pp. 577-579, 1985. This work adds to the further development of stabilizing controllers
R. K. Yedavalli and Z. Liang, “Reduced conservatism in stability ro- for continuous-time uncertain systems with state delay. The major
bustness bounds by state transformation,” IEEE Trans. Automat Contr.,
vol. AC-31, pp. 863-866, 1986. Manuscript received July 21, 1992; revised December 4, 1992 and April 15,
K. Zhou and P. P. Khargonekar, “Stability robustness bounds for linear 1993. This work was supported by Kuwait University Research Administration
state space models with structured uncertainty,” ZEEE Trans. Automar under Grant Number EE-049.
Contr.. vol. AC-32, pp. 621-623, 1987. The authors are with the Department of Electrical and Computer Engineer-
M. E. Sezer and D. D. Siljak, “Robust stability of discrete systems,” ing, Kuwait University, Safat, 13060 Kuwait.
Int. J. Contr., vol. 48, pp. 2055-2063, 1988. IEEE Log Number 92 16455.

0018-9286/94$04.00 0 1994 IEEE

996 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 39, NO. 5, MAY 1994

contributions of the present paper are as follows. The linear time- other ways to characterize the uncertainties like rank I-decomposition
delay uncertain system is reformulated in an extended state-space and [18], [19] and norm-bounded method [20], [21]. With focus on robust
by initially focusing on the nominal system, an additional assumption design of time-delay systems, our approach is systematic and starts by
is placed on the norm of the state vector. Two feedback controllers are stabilizing the nominal system, under Assumption 1, in an extended
considered: one is linear memoqless and the other is of saturation- state space. Then we turn to the uncertain system by exploiting
type. Under certain conditions on the norms of system matrices, it is Assumption 3 and introducing Assumption 4 as a norm-bounding
proved that the zero response of the system is asymptotically stable condition on the state vector. It will be shown that inequality (12)
by linear memoryless feedback controller. When the saturation-type provides a flexibility in the feedback gain adjustment. Our work is
controller is applied, it turns out that the resulting state trajectories based on the constructive use of Lyapunov functions and guarantees
remain within a calculable neighborhood around the origin. The size quadratic stabilizability and robustness of the closed-loop trajectories.
of this neighborhood can be adjusted by changing a parameter in
the controller gain. The theoretical results are illustrated by a typical 111. ANALYSIS
OF THE NOMINAL
SYSTEM
example.
By setting A A l ( . ) = 0, A A , ( . ) = 0, A B ( . ) = 0 and U(.) = 0 in
system (2), we obtain the free nominal system, which is described by
U. UNCERTAIN SYSTEM WITH TIME-DELAY
Consider the general functional differential equation
Z(t) = A , z ( t ) + A z z ( t - 7). (4)
To examine the stability of system (4), we define a Lyapunov function
Z(t) = f(t, zt) (1) candidate V ( . ) :R" x R" x R+ ---$ R+ as
where zt: = {z(t+B),0 E [-T, 01) defines the state of the dynamic
systems governed by functional differential equations with delays in
the argument of the dependent variables bounded by T > 0. Let
C = E([-., 01, R") be the space of all continuous functions which
V ( t )= z " t ) P z ( t ) +

where P = P t > 0 and R = Rt > 0.

lT + zt(t B)Rz(t + B)d0 (5)

map the real interval [-T, 01 into R". A particular class of these The following lemma summarizes a preliminary stability result.
systems is the class of linear uncertain delay systems described by Lemma 1: Suppose Assumption 1 holds. Then the free nominal
the following: system (4)is asymptotically stable if there exist (n x n ) matrices
P = fl > 0 and R = Rt > 0 such that
+AA~(u)]z(~)
Z ( t ) = [Ai
+ [A2 + A A 2 ( ~ ) ] z-( t T)
&f1 = - ( P A ~ + A ; P + R ) > o (64

+ [B+ AB(n)]u(t); t 2 0 (24

+
Mz = R AiP(PA1 Ai P R)-'PAz + + >0 (6b)

z ( t )= d t ) , t E [-T, 01 (2b) Proof: It follows from (5) that the time derivative of V ( . )along
the trajectory of system (4)is given by
where z ( t ) E R" is the current value of the response of (2),
u ( t ) E R" is the control, ~ ( tE) Rp is the uncertain element, V ( t )= Z"t)Pz(t) + z"t)PZ(t)
and T is the fixed delay. The quantities A I , A z , and B are constant +Z"t)Rz(t) - z t ( t - T ) R z ( t - 7). (7)
matrices with appropriate dimensions. System (1) is uncertain because
In terms of
the system and the input matrices depend on the uncertain parameter
U ( . ) through the functions AA1 (-), A A z ( . ) ,and A B ( . ) .The quantity Z(t) = [&(t) d ( t - T)]t (8)
4(.) is a vector-valued initial function on [-T, 01. We shall use
W t , X(W), X,(W), XM(W) to denote, respectively, the transpose, the substitution of (4) into (7) with some manipulations yields
eigenvalue, minimum eigenvalue, and maximum eigenvalue of a V ( t )= - . z t ( t ) H z ( t )
square matrix W . The notation 11 . 11 stands for the Euclidean norm of
a vector and its induced norm of a matrix. We let W > 0 (W < 0) to
signify positive-(negative-) definite matrix W . Finally, the solution
at time t of (1) with initial conditions zto = d(-) E C is written In view of (9). a sufficient condition for the asymptotic stability of
as z(to, 4 ) ( t ) .Sometimes, the subscripts will be omitted when no (4) is that V ( t )< 0 for all ~ ( t#)0 and V ( t )5 0 for z ( t ) = 0.
confusion arises. This is guaranteed if H > 0. Since P = > 0 and R = & > 0,
fl
The objective of this work is to design memoryless feedback then based on Assumption 1 it is easy to note that Mi = Mi and
controllers which can guarantee that, irrespective of the uncertain M2 = M z . A simple block diagonalization procedure [17] would
element, the zero-response is asymptotically stable, otherwise, all then convert (9b) to the following form:
system responses are globally bounded.
We now make the following assumptions.
Assumption I : a) The matrix A1 is Hurwitz. b) The pair ( A I ,B)
ao= :[ 4 (10)

is controllable. from which we see that the positive-definiteness of H is equiv-

Assumption 2: The uncertain vector U ( . ) : R + R is Lebesgue alent to conditions (6). We therefore conclude that system (4)is
measurable, where R E Rp is a prescribed compact subset of R p . asymptotically stable.
Assumption 3: There exist matrix functions of appropriate dimen-
sions D ( . ) ,E ( . ) ,F ( . ) ,continuous on RP, such that for all U E R w. STABILIZATION OF THE UNCERTAIN SYSTEM
In terms of (3), we rewrite system (2) in the following form:
A A i ( U )= B D ( u ) , A A z ( 0 )= B E ( u ) ,
AB(u)= BF(o). (3) +
Z ( t ) = A i ~ ( t ) A z ~ (-t T ) + Bu(t)
In the terminology of uncertain systems [14]-[16], the foregoing
+
+ B [ D ( a ) z ( t ) E ( o ) z ( t- T ) + F(U)U(t)] (11)
conditions are the so-called matching conditions. Indeed, there are and introduce the following Assumption.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 39, NO. 5 , MAY 1994 997

Assumption 4: For all 0 E [-7,01, there exists a scalar q > 0 When condition (16) is satisfied, then (20) reduces to
such that
v5 -@ 11 z [I2< 0 (21)
II z ( t + 0) I I I Q II z ( t ) II . (12)
and we conclude that the closed-loop trajectories of the uncertain
It should be emphasized that (12) is not restrictive since we allow q system (1 1) under the action of the linear controller (13) are asymp-
to be any value, greater or smaller than 1. In the sequel, we shall treat totically stable.
q as an adjustable parameter at the disposal of the designer. Thus, he Remark I : It is interesting to observe that condition (16) is struc-
will have the freedom to change q in a way to produce satisfactory tural in nature and does not impose any restriction on the gain factor
system performance, a useful feature that'will become clear in the p. The relevant information emerges from the system matrices and
simulation. can be adjusted by manipulating q. Hence, it is less conservative than
To stabilize ( I l ) , we consider two memoryless feedback con- the condition of [8] wherein a gain factor is required to satisfy an
trollers. inequality constraint.
1) Linear-type: Corollary I : Suppose that Assumption 5 is violated and we have
11 > 1. It follows from (20) that
U1 (t) = - p P P z ( t ) (13)
2) Saturation-type:

where P = P t > 0. Note that (13) represents a high-gain controller and we conclude that I 0 in the region
and the case ( p = 1) stands for the linear optimal controller [17] of
@
the nominal undelayed system of (1). The controller (14) for small p I po: = 2(r/ - 1)w2' (23)
n has a magnitude p . Both (13) and (14) can therefore provide
robustness of (1) in face of the uncertainties; see [14] for further In this case, the linear controller (13) renders the uncertain system
details. To facilitate further development, we define the following (1 1) asymptotically stable where the gain factor p is limited by (23).
norms: Now, turning attention to the behavior of system (1 1) under the
action of saturation-type controller (14), we observe from (8) and
J BtP 11; P=
(4:s 11 E(U) 11 $
:+
: 11 D(') 11).
(15)
(11) that the time derivative of V(.) can be written as
v = - 2 H z + 2UtBtPZ + 2UtFt(U)BtPZ + 2 z t K t ( u ) B t P z
A final assumption is in order.
(244
Assumption 5:
where
11 = It F(u) +2 F t ( U ) II< 1. K ( u )= [ D ( u ) E ( @ ) ] . (24b)

We now examine the stability behavior of the closed-loop systems we note that
formed by either (11) and (13) or (11) and (14). In both cases, we (2"B)(BtPZ) -
-
11 BtPZ 112 >I1 BtPz 11 -CY (25)
use the Lyapunov function V(.) given in (5). The following lemma II B t P z II +a! II B t P z II +a! -
provides our stability result concerning the linear controller (13). and define
Lemma 2: Suppose Assumptions 1, 4, and 5 are met. Then the
uncertain system (1 1) under the action of the linear controller (13) is $ = max
UER
11 K ( u ) 11 . (26)
asymptotically stable provided that
The following lemma summarizes our stability result.
@: = [ X , ( H ) - 2wp1 > 0. (16) Lemma 3: Suppose that Assumptions 1 , 4 , and 5 are satisfied and
proof: The time derivative of V(.) along the c~osed-~oop tra-
that > 2w$. Then the saturation-type renders the
closed-loop trajectories of system (1 1) uniformly ultimately bounded
jectories of (1 1) and (13) with the aid of (8) and (9a) is given
with a final attractor r(z)given by
by
P ( t ) = -z"(t)Hz(t) + 22"t)PBD(u)z(t)
+ 2 z ' ( t ) P B E ( u ) z ( t- 7)

- 2 p [ B t P z ( t ) l t ( 1+ F ) [ B t P z ( t ) ] . (17) Proof: The substitution of (14) into (24) with the aid of (25)
and Assumption 5, results in
In view of (8) and (12) we get

II z ( t ) 1 1 ~ 1 1z ( t ) II (18)
V 1- z t H z - 2p(ll B t P z 11 -a) + 2pq(II B t P z I( -CY)
+ 2 z t K t B t P z . (28)
which when used in (17) with some algebraic manipulation, it yields
Since (1 - 11) > 0, then the use (15) and (26) in (28) reduces it to
5 - X m ( H ) II z 112
i- +2 II B II P
~ P II z 112
-2p(BtPz)t(l+ F ) ( B t P Z ) . (19) i~I - X m ( H ) II z 11' + 2 ~ ( 1 - 7 ) ~2 + II z II II z II (29)
In view of (18), we get from (29)
By virtue of Assumption 5, it follows for all U E R that 0 <
+ +
1 [ F ( u ) F t ( u ) ] / 25 21, which is equivalent to the condition V 5 - [ X , ( H ) - 2~74111 [I2 +2p(1- 7 ) ~ . (30)
+ + +
that ([I F(cr)]' [ I F ( u ) ] ) / 2is positive-definite. Proceeding
Examination of (30) reveals that sufficient conditions for V < 0 are
further, we obtain from (19)
I ) X m ( H ) - 2 ~ 7 4> 0
F 5 -[X,(H) -2 4 11 z [I2 -2p(l - V)(JBtPZ(l*. (20) 2) II = 112> ( 2 4 1 - 11b)/(Ln(fc - 2w4).
998 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 39, NO. 5, MAY 1994

4-

4-

Fig. 1. Closed-loop state trajectories with linear-type controller ( p = 5). Fig. 3. Closed-loop state trajectories with saturation-type controller
(J =l 0.1, a = 0.001).

z
ii

r i(I&)
rune (sec)
Fig. 2. Control trajectory with linear-type controller ( p = 5).
Fig. 4. Control trajectory with saturation-type controller ( p = 0.1,
0 = 0.001).
Since 1) z I)<)) t 1) from ( I Q , then condition (2) above is
necessarily satisfied if
For purpose of illustration, p is a constant which can take any arbitrary
II z I(*> (W1- v)cy)/(Lz(H)- 24J). (31) value provided that Lemmas 2 and 3 are satisfied. In our example, p
Based on the results of [15], it is readily seen that the =trajectories can take values in the range from 0.01 to 3.5 given that q = 0.1. Also,
are uniformly ultimately bounded, and the final attractor is given by it has been observed that if the value of q is decreased below 0.1,
(27). then the conditions of previous lemmas are still satisfied. However,
R e m a r k 2 : As cy + 0, then p* + 0. Also, p + 0 results in if the value of q is increased to 10, then the conditions of Lemma
p* + 0 provided that X , ( H ) > 2w+. Hence, small values of cy 3 are violated and the design of a saturation-type controller is not
reduces the tolerable zone around the origin, but the control signal guaranteed.
would approach a stair-case form. The other parameters values are selected as
Next, we present the results of computer simulation. [xl(O) 22(0)lt = [5.5 3.5It, p = 3.5, = I . (35)
U
4
V. EXAMPLE To ensure that the conditions (6a) and (6b) are satisfied, we choose
-I A t , p = -I2 and = 0‘512’ Then’ the solution Of the
To illustrate the preceding results, we consider a linear uncertain
Lyapunov matrix equation yields the following:
delay system of type (2) with
(36)

(32) A. Linear-Type Controller Design

D ( u ) = [O psin (U)] The conditions of Lemma 2 are met with the following values:
E ( u ) = [p sin ( U ) pcos (U)]
F(cr) = [0.5 sin (a)]
(33) q = 0.5,
/3 = 3.9950,
X,(H) = 0.4293,
= 0.1468 > 0.
w = 0.0354, q = 0.1
} (37)
and
A linear-type controller of the form (13) is designed, and the results
7 = 3. (34) of the simulation using SIMULAB package are shown in Figs. 1
IEEE TRANSACTIONS ON AUMMATIC CONTROL, VOL. 39, NO. 5, MAY 1994 999

REFERENCES

M. Ikeda and T. Ashida, “Stabilization of linear systems with time-

varying delay,” IEEE Trans. Automat. Contr., vol. AC-24, pp. 369-370,
1979.
T. Mori, E. Noldus and M. Kuwahara, “A way to stabilize linear systems
with delayed state,” Automatica, vol. 19, pp. 571-573, 1983.
A. Hmamed, “Stability conditions of delay-differential systems,” Int. J.
VI Contr., vol. 43, pp. 455-463, 1986.
W. H. Kwon and A. E. Pearson, “A note on feedback stabilization
of differential-difference system.” IEEE Trans. Automat. Contr., vol.
AC-22, pp. 468470, 1977.
G. J. Nazaroff, “Stability and stabilization of linear differential delay
systems,” IEEE Trans. Automat. Contr., vol. AC-18, pp. 317-318, 1973.
A. Feliachi and A. Thowsen, “Memory stabilization of linear delay-
I differential systems,” IEEE Trans. Automat. Contr., vol. AC-26, pp.
I 2 3 4 5 6 586589, 1981.
T i (ret) A. Thowsen, “Uniform ultimate boundedness of the solutions of un-
certain dynamic delay systems with state-dependent and memoryless
Fig. 5 . Closed-loop state trajectories with saturation-type controller ( p = 5, feedback control,” Int. J. Contr., vol. 37, pp. 1135-1 143, 1983.
Q = 0.001). Y. Yu, “On stabilizing uncertain linear delay systems,” J. Optimiz.
Theory Appl., vol. 41, pp. 503-508, 1983.
C. S. Lee and G. Leitmann, “Continuous feedback guaranteeing uniform
ultimate boundedness for uncertain linear delay systems: An application
to river pollution control,” Comput. Math. Appl., vol. 16, pp. 929-938,
1988.
Y. H. Chen and C. S. Lee, “On the control of an uncertain water quality
system,” Optimal Contr. Appl., Methods, vol. 8, pp. 3-9, 1987.
M. S. Mahmoud, “Robustness of discrete-time systems with delayed
perturbations,” Kuwait Univ. Tech. rep. ECE-049-R1, submitted for
publication.
“Control of uncertain discrete-time systems with state-delay,’’ Kuwait
Univ. Tech. rep. ECE-049-R2, submitted for publication.
“Design of stabilizing controllers for uncertain discrete-time systems
with state-delay,’’ Kuwait Univ. Tech. rep. ECE-049-R3, submitted for
publication.
A. A. Bahnasawi and M. S. Mahmoud, Control of Partially-Known
Dynamical Systems. New York: Springer-Verlag, 1989.
I M. Corless and G. Leitmann, “Deterministic control of uncertain sys-
1 2 3 4 5 6
tems,” in: Modelling and Adaptive Control, I. Bymes and A. Kurzhanski,
Timc (wc) eds., New York: Springer-Verlag. 1988, pp. 108-133.
B. R. Barmish, M. Corless, and G. Leitmann, “A new class of stabilizing
Fig. 6. Control trajectory with saturation-type,/controller (fi = 5, controllers for uncertain dynamical systems,” SIAM J. Contr. Optim.,
Q = 0.001).
vol. 21, pp. 246-255, 1983.
T. Kailath, Linear Systems. New York Prentice-Hall, 1980.
and 2. It should be noted that the closed-loop uncertain system is I. R. Petersen and C. V. Hollot, “A Riccati equation approach to
the stabilization of uncertain linear systems,” Automatica, vol. 22, pp.
asymptotically stable using any positive value of the controller gain
397411, 1986.
p. However, one can choose a particular value of p such that other J. C. Shen, B. S. Chen, and F. C. Kung, “Memoryless stabilization
design specifications are met. In ow simulation, we set /L = 5. of uncertain dynamic delay systems: Riccati equation approach,” IEEE
Trans. Automat. Contr., vol. 36, pp. 638-640, 1991.
B. Saturation-Type Controller Design I. R. Petersen, “A stabilization algorithm for a class of uncertain linear
systems,” Syst. Contr., vol. 8, pp. 351-357, 1987.
The conditions of Lemma 3 are also satisfied with the following L. Xie, M. Fu, and C. E. de Souza, “Hm control and quadratic
results: stabilization of systems with parameter uncertainty via output feedback,”
IEEE Trans. Automat. Contr., vol. 31, pp. 1253-1256, 1992.
$I = 6.0622, X,(H) - 2 ~ =46.2862
~ x > 0. (38)

Simulation results for the closed-loop system using a saturation-type

controller of the form (14) are shown in Figs. 3 4 where the value of
the controller parameter cy is set to 0.001. In this simulation, we tried
two values for the controller gain p. Also, it has been observed that
for fixed a,the radius of attractor pLIas defined in (27) is proportional
to Jr;,so for p = 0.1 and 5, p* = 0.3988, and 2.8203 respectively.

VI. CONCLUSIONS
Design of linear-type and saturation-type controllers to stabilize
continuous-timedelay systems has been developed in this work. It has
been established that the linear controller yields asymptotic behavior
of the closed-loop system whereas the saturation controller produces
uniformly bounded behavior.