
the matrices occurring in equation (31) and the covariance matrix
of x
~
(
t|t) are found after simple calculations. We have, for all t ≥
0,
Φ(t; t + 1)∆*(t) =
− )()(
)(
)(
)(
1
21
2
2
1
ttCtC
tC
ttC
tC
Φ(t; t + 1)Φ*(t + 1; t)Φ(t + 1; t)
=
++−+−
−−−
−−−
)()()()()(
)()()()(
)()()()()(
)(
1
2332121
233212
2333121
1
ttCttCtCttCtC
tCtCtCtC
ttCttCtCttCtC
tC
φ
φ
φ
and
cov x
~
(
t|t) = E x
~
(t|t) x
~
'(t|t) =
−−
−
−
22
22
1
2
1
)(
ttt
tt
ttt
tC
b
To gain some insight into the behavior of this system, let us
examine the limiting case
t → ∞ of a large number of observa-
tions. Then
C
1
(t) obeys approximately the differential equation
dC
1
(t)/dt ≈ C
2
2
(t) (t >> 1)
from which we find
C
1
(t) ≈ (1 –
φ
33
)
2
t
3
/3 +
φ
33
(1 –
φ
33
)t
2
+
φ
33
2
t + b
2
/a
2
(
t >> 1) (39)
Using (39), we get further,
Φ
–1
Φ*Φ ≈
−− 011
010
011
and Φ
–1
∆* ≈
1
0
0
(t >> 1)
Thus as the number of observations becomes large, we depend
almost exclusively on
x
1
*(t|t) and x
2
*(t|t) to estimate x
1
*(t + 1|t +
1) and
x
2
*(t + 1|t + 1). Current observations are used almost
exclusively to estimate the noise
x
3
*(t|t) ≈ y
1
*(t) – x
1
*(t|t) (t >> 1)
One would of course expect something like this since the prob-
lem is analogous to fitting a straight line to an increasing number
of points.
As a second check on the reasonableness of the results given,
observe that the case
t >> 1 is essentially the same as prediction
based on continuous observations. Setting
φ
33
= 0, we have
E
3/
)|(
~
322
222
2
1
tab
tba
ttx
+
≈
(t >> 1;
φ
33
= 0)
which is identical with the result obtained by Shinbrot [11],
Example 1, and Solodovnikov [14], Example 2, in their treat-
ment of the Wiener problem in the finite-length, continuous-data
case, using an approach entirely different from ours.
Conclusions
This paper formulates and solves the Wiener problem from the
“state” point of view. On the one hand, this leads to a very gen-
eral treatment including cases which cause difficulties when at-
tacked by other methods. On the other hand, the Wiener problem
is shown to be closely connected with other problems in the
theory of control. Much remains to be done to exploit these
connections.
References
1 N. Wiener, “The Extrapolation, Interpolation and Smoothing of
Stationary Time Series,” John Wiley & Sons, Inc., New York, N.Y.,
1949.
2 L. A. Zadeh and J. R. Ragazzini, “An Extension of Wiener's The ory
of Prediction,” Journal of Applied Physics, vol. 21, 1950, pp. 645–655.
3 H. W. Bode and C . E. Shannon, “A Simplified Derivati on of Linear
Least-Squares Smoothing and Prediction Theory,” Proceedinqs IRE, vol.
38, 1950, pp. 417–425.
4 R. C. Booton, “An Optimization Theory for Time-Varying Linear
Systems With Nonstationary Statistical Inputs,” Procee dings IRE, vol. 40,
1952, pp. 977–981.
5 J. H. Laning and R. H. Battin, “Random Processes in Automatic
Control,” McGraw-Hill B ook Co mpa ny, Inc., New Yor k, N. Y., 1956.
6 W. B. Davenport, Jr., and W. L. Root, “An Introduction to the
Theory of Random Signals and Noise,” McGraw-Hill Book Company,
Inc., New York, N. Y., 1958.
7 S. Darlington, “Linear Least-Squares Smoothing and Prediction,
With Applications,” Bell System Tech. Journal, vol. 37, 1958, pp.1221–
1294.
8 G. Franklin, “The Optimum Synthesis of Sampled-Data Systems”
Doctoral dissertation, Dept. of Elect . En gr ., C ol umbia U niv ersit y, 1 955.
9 A. B. Lees, “Interpolation and Extrapolation of Sampled Data,”
Trans. IRE Prof. Grou.p on Information Theory, IT-2, 1956, pp. 173–175.
10 R. C. Davis. “On the Theory of Prediction of Nonstationary
Stochastic Processes,” Journal of Applied Physics, vol. 23. 1952, pp.
1047–1053.
11 M. Shinbrot , “Optimization of Tirne-Varying Linear Systems With
Nonstationary Inputs,” T
RANS. ASME, vol. 80, 1958. pp. 457–462.
12 M. Blum, “Recursion Formulas for Growing Memory Digital
Filters,” Trans. IRE Prof. Group on Information Theory. IT-4. 1958, pp.
24–30.
13 V. S. Pugachev, “The Use of Canonical Expansions of Random
Functions in Determining an Optimum Linear System,” Automatics and
Remote Control (USSR), vol. 17, 1956, pp. 489–499; translation pp. 545–
556.
14 V. V. Solodovnikov and A. M. Batkov. “On the Theory of Self-
Optimizing Systems (in German and Russian),” Proc. Heidelberg
Conference on Automatic Control, 1956. pp. 308–323.
15 J. L. Doob, “Stochastic Processes,” John Wiley & Sons, Inc., New
York, N. Y., 1955.
16 M. Loève, “Probability Theory,” Van Nostrand Company, Inc.,
New York, N. Y., 1955.
17 R. E. Bellman, I. Glicksberg, and 0. A. Gross, “Some Aspects of
the Mathematical Theory of Control Processes,” RAND Report R-313,
1958, 244 pp.
18 R. E. Kalman and R. W. Koepcke, “Optimal Synthesis of Linear
Sampling Control Systems Using Generalized Performance Indexes,”
T
RANS. ASME, vol. 80, 1958, pp. 1820–18 26.
19 J. E. Bertram, “Effect of Quantization in Sampled-Feedback
Systems,” Trans. AlEE, vol. 77, II, 1958, pp. 177–182.
20 R. E. Kalman and J . E. Bertram, “General Synthesi s Procedure for
Computer Control of Single and Multi-Loop Linear Systems” Trans.
AlEE, vol. 77, II, 1958, pp . 602–609.
21 C. W. Merriam, III, “A Class of Optimum Control Systems,”
Journal of the Franklin Institute, vol. 267, 1959, pp. 267–281.
22 R. E. Kalman and J. E. Bertram, “A Unified Approach to the
Theory of Sampli ng Systems,” Journal of the Fr anklin Institute, vol. 26 7,
1959, pp. 405–436.
23 R. E. Kalman and R. W. Koepcke, “The Role of Digital Com-
puters in the Dynamic Optimization of Chemical Reactors,” Proc.
Western Joint Computer Conference, 1959, pp. 107–116.
24 R. E. Kalman, “Dynamic Optimization of Linear Control Sys-
tems, I. Theory,” to appear.
25 S. Sherman, “Non-Mean-Square Error Criteria,” Trans. IRE Prof.
Group on Information Theory, IT-4, 1958, pp. 125–126.
26 V. S. P ugac hev, “On a Possi ble General Sol ution of the Problem of
Determining Optimum Dynamic Systems,” Automatics and Remote
Control (USSR), vol. 17, 1956, pp. 585–589.
27 G. C. Newton, Jr., L. A. Gould, and J. F. Kaiser, “Analytical
Design of Linear Feedback Controls,” John Wiley & Sons, Inc., New
York, N. Y., 1957.
28 0. J. M. Smith, “Feedback Control Systems,” McGraw-Hill Book
Company, Inc., Ne w Yor k, N. Y., 1 9 58 .
29 R. E. Kalman, “On the General Theory of Control Systems,”
Proceedings First International Conference on Automatic Control,
Moscow, USSR, 1960.
Transactions of the ASME–Journal of Basic Engineering, 82 (Series D): 35-45. Copyright © 1960 by ASME