See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/51955025

Extensions to the Theory of Widely Linear Complex Kalman Filtering

Article in IEEE Transactions on Signal Processing · May 2011

DOI: 10.1109/TSP.2012.2214213 · Source: arXiv

CITATIONS READS
13 393

2 authors:

Wenbing Dang Louis L. Scharf

Halliburton Colorado State University
8 PUBLICATIONS 543 CITATIONS 320 PUBLICATIONS 12,424 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Wenbing Dang on 01 April 2014.

The user has requested enhancement of the downloaded file.


Widely Linear Kalman Filtering
Wenbing Dang, Student Member, IEEE, and Louis L. Scharf, Life Fellow, IEEE
Abstract—For a zero mean, proper, complex random vector
x, the Hermitian covariance ExxH is a complete second-order
characterization. However, if the vector x is improper, it is
correlated with its complex conjugate, meaning ExxT 6= 0. This
improper or complementary covariance must be accounted for
in a complete second-order characterization.
The improper covariance has been exploited for widely linear
arXiv:1105.5432v1 [cs.SY] 27 May 2011
(WL) Wiener filters and WL minimum mean squared error
(MMSE) estimators, and the improvement in performance of
the WLMMSE estimator over the LMMSE estimator has been
quantified. In this paper we consider the design of the widely
linear Kalman filter (WLKF). We analyze the WLKF, extended
WLKF, and unscented WLKF. The key idea of this paper is
to modify the error covariance matrices and the construction
of effective sigma points in the WLKF in a systematic way
that exploits the Hermitian and complementary covariance of
improper states and noises.
Index Terms—Widely linear transformation, complementary
covariance, Kalman filter, unscented Kalman filtering, sigma
points.
I. I NTRODUCTION
Complex signals are ubiquitous in science and engineering,
arising as they do as complex representations of two real
channels or of two dimensional fields. Consider a zero mean
complex random vector x. The usual covariance matrix defined
as ExxH describes its Hermitian second order covariance.
But when x and its complex conjugate x∗ are correlated, the
complementary covariance matrix ExxT does not vanish, so
it carries useful second order information about the complex
random vector x. We call a complex random vector proper
as long as its complementary covariance matrix vanishes and
improper otherwise. Proper complex vectors have a statisti-
cal description similar to real vectors, but improper random
vectors do not. A comprehensive second order analysis of
improper random vectors and processes is considered in [1]–
[6].
For any improper random vector x, for which x is correlated
with its complex conjugate x∗ , intuition suggests that a good
estimator of x should depend on x∗ . This requires a method-
ology of widely linear processing instead of strictly linear
processing [1], [2]. The merit of widely linear processing has
been exploited in various estimation and filtering problems.
The widely linear minimum mean squared error (WLMMSE)
This work is supported by the National Science Foundation under grants
CCF-1018472 and CCF-0916314, and the Air Force Office of Scientific
Research under contract FA9550-10-1-0241.
W. Dang is with the Department of Electrical and Computer Engineering,
Colorado State University, Fort Collins, CO 80523, USA, Phone: (970) 817-
0249 (email: dwb87514@gmail.com).
L. L. Scharf is with the Department of Electrical and Computer Engi-
neering, and the Department of Statistics, Colorado State University, Fort
Collins, CO 80523, USA, Phone: (970) 491-2979, Fax: (970) 491-2249 (e-
mail: Louis.Scharf@colostate.edu).
1
estimators for random complex signals are developed in [4],
[6], [7]. Also the WLMMSE estimators for second order
stationary process are given by [8]. WL Wiener filters are
developed in [3] and [4]. And a widely linear Kalman filter
is studied in [9]. The problems of detection and equalization
in communication networks are also addressed in [10]–[16]
using WL processing.
It turns out that widely linear processing brings improve-
ment in performance over strictly linear processing [4], [17],
[18]. For instance, in [4] it has been shown that by assuming
proper white measurement noises, the maximum performance
advantage of WLMMSE estimation over LMMSE estimation
is a factor of two.
In the past few decades the Kalman filter [19] has been
modified to apply to nonlinear problems, producing Extended
Kalman filters [20], Unscented Kalman filters [21], and parti-
cle filters [22]–[25]. The motivation of this paper is to make
use of widely linear processing to develop novel Kalman filters
and their nonlinear versions for improper complex states.
We show that for improper complex states, complementary
covariance matrices may be used to create widely linear KFs,
as well as extended and unscented widely linear KFs. Widely
linear Kalman filters (WLKF) and extended WLKFs were
derived in [9], [26]. However, in these papers, the authors
have studied WLKFs for dynamical systems with very special
state and measurement matrices. In other words, the updates
of random states and measurement do not depend on the
conjugates of states and noises.
In this paper we show that the WLKF can address more
general dynamical models that may have widely linear state
and measurement equations. The duality between widely linear
and dual real channel adaptive filtering [27] motivates us
to demonstrate that the WLKF can be used for LMMSE
estimation in dual channels when those channels have ar-
bitrarily coupled state equations. In other words the WLKF
may be implemented into real equations, or by using complex
augmented notation, into complex equations. Moreover, by
studying a complex scalar random state with proper noises,
we show that compared to the conventional KF, the WLKF
reduces MSE by a factor of two. This advantage may be more
significant for more general models.
We propose the extended WLKF for the widely linear
approximation of nonlinear state and measurement equations,
which is equivalent to the results in [9], [26]. For the EWLKF,
however, the estimated posterior means and covariances are
precise only to first order in a Taylor expansion. To obtain
accuracy to second order, we study the unscented WLKF. The
Unscented WLKF of [9] uses sigma points that represent the
mean and Hermitian covariance of states. However, the sigma
points should also represent the corresponding complementary
covariance. In this paper we develop a systematic way to
 2

construct modified sigma points with the property that they The matrix H = 12 Tm MTH n is called an augmented matrix
can preserve the complete first and second order statistical with the property that its southeast block is the conjugate of
information of complex random vectors. its northwest block, and its southwest block is the conjugate
The remainder of this paper is organized as follows. In sec- of the northeast block:
tion II we briefly review some rudiments of complex random 
H1 H2

vectors, including widely linear transformations, augmented H= , (8)
H∗2 H∗1
covariances, and the definition of improper random signals.
In section III the widely linear minimum mean squared error where
estimator is introduced. In section IV we derive the widely 1
H1 = [M11 + M22 + j (M21 − M12 )] , (9)
linear Kalman filter (WLKF). We give two representations of 2
WLKF: the augmented WLKF and the Cartesian WLKF. In 1
H2 = [M11 − M22 + j (M21 + M12 )] . (10)
addition, we give an example demonstrating the performance 2
enhancement of the WLKF over the KF. In section V, we de- Therefore the augmented matrix H rules the widely linear
rive the extended WLKF. In section VI, we obtain the effective transformation
sigma points for any improper complex random vector. Then
y = H1 x + H2 x∗ ⇔ a + jb = H1 (u + jv) + H2 (u − jv)
we give an algorithm for constructing an unscented WLKF
(UWLKF). In a simulation for real phase demodulation, we = (H1 + H2 )u + j(H1 − H2 )v
show that an UWLKF outperforms a complex, unscented (11)
Kalman filters that assumes proper states and noises in making We see that if and only if H2 = 0, the widely linear transfor-
real estimates from complex baseband measurements. mation is a strictly linear transformation. This corresponds to
the special case M11 = M22 and M12 = −M21 in the linear
II. B RIEF R EVIEW OF C OMPLEX R ANDOM V ECTORS transformation of u, v into a, b.
Let Ω be the sample space of a random experiment that The augmented mean vector of the complex random vector
generates two channels of real signals u, v ∈ Rn defined on Ω. x is    
From this we obtain the real composite random vector z ∈ R2n µx µu + jµv
µx = Ex = = , (12)
as zT = [uT , vT ]. We also have the complex random vector µ∗x µu − jµv
x ∈ Cn , obtained by composing u and v into its real and and the augmented covariance matrix of x is
imaginary parts: " #
x = u + jv. (1) H Rxx R e xx
Rxx = E(x − µx )(x − µx ) = e ∗ . (13)
Rxx R∗xx
The complex augmented random vector x corresponding to x
is defined as   The matrix Rxx is the conventional Hermitian covariance
x matrix
x= ∗ . (2)
x Rxx = E(x − µx )(x − µx )H = RHxx , (14)
From here the complex augmented random vector will always
and the matrix R
e xx is the complementary covariance matrix
be underlined. It’s easy to check that the real composite vector
z and the complex augmented vector x are related as e xx = E(x − µx )(x − µx )T = R
R eT . (15)
xx
x = Tn z. (3) Definition 1: If the complementary covariance matrix R e xx
The real-to-complex transformation Tn is is zero, then x is called proper, otherwise x is improper.
  The random vector x = u + jv is proper if and only if
I jI Ruu = Rvv and Ruv = −RTuv , where u and v are the real
Tn = , (4)
I −jI and imaginary parts of x respectively.
which is unitary within a factor of 2:
Tn TH H III. W IDELY L INEAR MMSE E STIMATION
n = Tn Tn = 2I. (5)
The general linear MMSE estimation problem is to estimate
In fact, it is equation (3) that governs the equivalence between a complex random vector x from a complex random measure-
dual channel filtering for z and complex filtering for x. Given a ment vector y. The objective is to minimize the mean square
real linear transformation M ∈ R2m×2n and a composite real error between the signal x and its estimator x̂. Suppose the
vector z ∈ R2n , then the most general linear transformation output of the linear estimator is x̂ = Wy. The orthogonality
of the real channels u and v into the real channels a, b is principle says the estimator error e = Wy − x is orthogonal
    
a M11 M12 u to the measurement y:
= , (6)
b M21 M22 v
E[(Wy − x)yH ] = 0. (16)
Call y = a + jb. Then the corresponding complex augmented
vector y is We shall assume that both the signal x and the measurement
        y have zero means, since this doesn’t change the structure of
y a 1 u
y = ∗ = Tm = Tm MTH n Tn = Hx, the optimum linear estimator. This yields the normal equations
y b 2 v
(7) WRyy = Rxy , (17)
 3

where Ryy = EyyH and Rxy = ExyH . We wish to extend IV. W IDELY-L INEAR K ALMAN F ILTER
this result to a widely linear MMSE estimator. From section
II we know that the widely linear MMSE estimator should be Consider the widely linear state and measurement equations
of the form
xt = A1 xt−1 + A2 x∗t−1 + B1 wt−1 + B2 wt−1
∗
, t = 1, 2, ...,
x̂ = W1 y + W2 y∗ . (18) (27)
This is equivalent to yt = C1 xt + C2 x∗t + nt , t = 0, 1, ..., (28)
x̂ = Wy, (19) where xt ∈ Cn and yt ∈ Cp , A1 , A2 ∈ Cn×n , B1 , B2 ∈
Cn×m , and C1 , C2 ∈ Cp×n . The initial state vector x0 is
where     assumed to have zero mean. Our procedure is to augment these
y W1 W2 equations as
y= , and W = . (20)
y∗ W2∗ W1∗
xt = Axt−1 + Bwt−1 , t = 1, 2, ..., (29)
Since the mean square error may be written
1 yt = Cxt + nt , t = 0, 1, ..., (30)
Ekx̂ − xk2 = Ekx̂ − xk2 , (21)
2
it is enough to minimize the mean-squared error between where xT = [xT , xH ], yT = [yT , yH ], and
x̂ and x. Applying the orthogonality principle, we have the      
corresponding augmented normal equations A1 A2 B1 B2 C1 C2
A= ,B = ,C = . (31)
A∗2 A∗1 B∗2 B∗1 C∗2 C∗1
WRyy = Rxy (22)
Suppose Ex0 = 0, and the augmented covariance of x0 is
Therefore the remaining problem is to specify the matrices  H T
 " #
W1 and W2 . Actually the key question is to determine the Ex x
0 0 Ex x
0 0 Π 0 Π
e 0
Ex0 xH0 = Ex∗ xH = = Π0 . (32)
inverse of Ryy . Using the matrix-inversion lemma, the matrix 0 0 Ex∗0 xT0 e ∗ Π∗
Π 0 0
R−1
yy may be written as
Assume that Eut = 0 and Ent = 0 for all t. Further assume
" #−1 " # the real and imaginary parts of x0 are uncorrelated with the
R R P−1 −P−1 −∗
yy Ryy Ryy real and imaginary parts of ut and nt , meaning Ex uH = 0
R−1 = e yy
e yy yy
e
yy = , 0 t
R∗yy R∗yy −Q−1 e ∗ −1
yy Ryy Ryy Q−1 and Ex0 nH
yy t = 0 for t ≥ 0. Using the representation advocated
(23) in [28], the joint augmented second-order characterization of
where the matrix Pyy = Ryy − R e yy R−∗ Re ∗ is the (x , u , n ) is given by
yy yy 0 t t
Schur complement of Ryy . It is the error covariance matrix
for linearly estimating y from y∗ . And Qyy = R∗yy −
   
x0  Π0 0 0 0
e ∗ R−1 Re yy = P∗ . Thus R−1 may be simplified to the E wn  xH H

R yy yy yy yy 0 wHm nH
m 1 = 0 δnm Q δnm S 0 .
augmented matrix nn 0 δnm SH δnm R 0
" # (33)
H H
P −1
−P −1 e
R R
yy yy
−∗ For n ≥ m, Ew n x m = 0 and En n xm = 0. For n > m,
R−1
yy =
yy
−∗ e ∗ −1
yy
−∗
(24) Ew yH = 0 and En yH = 0. This is the same setup as that
−Pyy Ryy Ryy Pyy n m n m
of the usual Kalman filter, but with covariances augmented to
account for non-zero complementary covariance.
From the Woodbury identities we know that P−1 −∗
yy Ryy Ryy =
e
Assume the estimator of xt−1 from measurements YTt−1 =
R−1 −∗ −∗ e ∗ −1
yy Ryy Pyy , and Pyy Ryy Ryy = (Pyy Ryy Ryy )
e −1 e −∗ ∗
=
−1 (yT1 , . . . , yt−1 )T is known to be x̂t−1|t−1 . Here we require
R−∗ e ∗ −1
yy Ryy Pyy . Thus Ryy may also be written as x̂t−1|t−1 to be a complex augmented vector. Then the LMMSE
"
−1 −1 e −∗
# prediction of xt from YTt−1 is x̂t|t−1 = Ax̂t−1|t−1 , and
P −R R P
yy yy T
R−1
yy =
yy yy
. (25) the prediction of yt from Yt−1 is ŷt|t−1 = Cx̂t|t−1 . These
−R−∗ e ∗ −1
yy Ryy Pyy P−∗
yy augmented prediction updates are equivalent to widely linear
prediction updates x̂t|t−1 = A1 x̂t−1|t−1 + A2 x̂∗t−1|t−1 and
So the widely linear estimator is [4]
ŷt|t−1 = C1 x̂t|t−1 + C2 x̂∗t|t−1 .
x̂ = (Rxy −R e xy R−∗ Re ∗ )P−1 y+(R e xy −Rxy R−1 R e yy )P−∗ y∗ . Suppose we are given the error covariance matrix Pt−1|t−1
yy yy yy yy yy
(26) for êt−1|t−1 = x̂t−1|t−1 − xt−1 . Then the error covariance
The linear MMSE estimator is a special case of this widely matrix Pt|t−1 for êt|t−1 = x̂t|t−1 − xt is
linear MMSE estimator. In [4] it has been shown that when es- " #
timating an improper complex signal in additive white proper H H Pt|t−1 P e t|t−1
Pt|t−1 = APt−1|t−1 A + BQB = e ∗ ,
noise, the maximum performance advantage of WLMMSE Pt|t−1 Pt|t−1
estimation over LMMSE estimation is a factor of 2. (34)
 4

where the NW block of Pt|t−1 is the Hermitian error covari- where the NW block of Pt|t is the Hermitian error covariance
ance and the NE block is the complementary error covariance: and the NE block is the complementary error covariance:

Pt|t−1 =A1 Pt−1|t−1 AH e∗ H Pt|t = (I − Kt C1 − Ke t C∗ )Pt|t−1 − (Kt C2 + Ke t C∗ )P

e∗
1 + A2 Pt−1|t−1 A1 2 1 t|t−1 ,
e t−1|t−1 AH + A2 P∗ AH P e t C∗ )P
e t|t = (I − Kt C1 − K e t C∗ )P∗
e t|t−1 − (Kt C2 + K
+ A1 P 2 t−1|t−1 2 2 1 t|t−1 .
e∗ H ∗ H (42)
+ B1 QBH e H
1 + B2 Q B1 + B1 QB2 + B2 Q B2 ,
e∗ These widely linear Kalman filter equations are summarized
P
e t|t−1 =A1 Pt−1|t−1 AT + A2 P
2 AT
t−1|t−1 2
as Algorithm 1.
+ A1 Pt−1|t−1 A1 + A2 P∗t−1|t−1 AT1
e T

+ B1 QBT2 + B2 Qe ∗ BT B1 QB
e T + B2 Q∗ BT . Algorithm 1 Widely Linear Kalman Filter
2 1 1
(35) Initialize with x̂0|0 = 0, P0|0 = Π0 and P e 0|0 = Πe 0 as defined
in equation (32);
The error covariance matrix St|t−1 for the innovation n̂t|t−1 = At the t-th iteration, t = 1, 2, · · · ,
ŷt|t−1 − yt is 1. Update the predicted state as x̂t|t−1 = A1 x̂t−1|t−1 +
" # A2 x̂∗t−1|t−1 ;
S S 2. Update the predicted Hermitian and complementary error
= CPt|t−1 C + R = e t|t−1
H
e t|t−1
St|t−1 , (36)
S∗t|t−1 S∗t|t−1 covariance matrices Pt|t−1 , P e t|t−1 , St|t−1 , and Se t|t−1 by
using equations (35) and (37);
where the NW block of St|t−1 is the Hermitian error covari- 3. Apply equation (40) to derive the widely linear Kalman
ance and the NE block is the complementary error covariance: gains Kt and K e t , wherein the Schur complement is PS =
St|t−1 =C1 Pt|t−1 CH e∗ H H St|t−1 − St|t−1 S−∗
e e∗
t|t−1 St|t−1 ;
1 + C2 Pt|t−1 C1 + C1 Pt|t−1 C2
e
4. Use the predictions x̂t|t−1 and ŷt|t−1 = C1 x̂t|t−1 +
+ C2 P∗t|t−1 CH
2 + R, C2 x̂∗t|t−1 to generate the widely linear estimator of xt
S T
=C1 Pt|t−1 C + C2 Pe∗ T e t|t−1 CT
t|t−1 C2 + C1 P
e t|t−1
2 1

+ C2 P∗t|t−1 CT1 + R. e t yt − ŷt|t−1 ∗ ;

e x̂t|t = x̂t|t−1 + Kt yt − ŷt|t−1 + K
(37)
5. Update the Hermitian and complementary error covariance
Recall that the normal equation for the Kalman gain is matrices Pt|t and P
e t|t by using equation (42).

Kt St|t−1 = Pt|t−1 CH . (38)

Remark 1: In the state and measurement equations (27)-(28),
From Section III we know that S−1 has the augmented form the new state xt depends on xt−1 , x∗t−1 , wt−1 , and wt−1 ∗
.
t|t−1 ∗ ∗
And measurement yt depends on xt , xt , nt , and nt . For
the state and measurement equations of the WLKF proposed
" #
−1 P−1
S −S−1 −∗
t|t−1 St|t−1 PS
e
St|t−1 = , (39) in [9], the new state xt depends only on xt−1 and wt−1 ,
−S−∗ e∗ −1
t|t−1 t|t−1 PS
S P−∗S and measurement yt depends only on xt and nt . Thus the
e t|t−1 S−∗ S e∗ WLKF in [9] can be obtained as a special case of the WLKF
where PS = St|t−1 − S t|t−1 t|t−1 is a Schur comple- considered here by simply letting matrices A2 , B2 , and C2
ment, namely the error covariance for estimating n̂t|t−1 from
in (27)-(28) be zero.
n̂∗t|t−1 . Thus the augmented Kalman gain may be written as
Remark 2: An insightful interpretation of the widely lin-
" # ear KF is that the augmented Kalman gain determines the
K K
Kt =Pt|t−1 CH S−1 t t
e
t|t−1 = e ∗ ; WLMMSE estimator of the prediction error êt|t−1 = xt −
Kt K∗t x̂t|t−1 from the innovation n̂t|t−1 = yt − ŷt|t−1 . Thus the
Kt =(Pt|t−1 CH H −1 widely linear KF reduces to the linear KF if and only if
1 + Pt|t−1 C2 )PS
e
e t|t−1 CT )S−∗ S e∗ −1 (40) êt|t−1 − Kt n̂t|t−1 is orthogonal to n̂∗t|t−1 , where Kt =
− (Pt|t−1 CT2 + P t|t−1 t|t−1 PS ,
1
Pt−1|t−1 CH S−1 t|t−1 . That is,
e t =(Pt|t−1 CT + P
K e t|t−1 CT )P−∗
2 1 S
E(êt|t−1 − Kt n̂t|t−1 )n̂Tt|t−1
− (Pt|t−1 C1 + Pt|t−1 C2 )S−1
H e H −∗
t|t−1 St|t−1 PS .
e
= Eêt|t−1 (Cêt|t−1 + nt )T − Kt E n̂t|t−1 n̂Tt|t−1
When complementary covariances P e t|t−1 and Se t|t−1 vanish,
=Pe t|t−1 CT − Pt−1|t−1 CH S−1 S e t|t−1
t|t−1
and when C2 is zero, we have Kt = diag(Kt , K∗t ), where
Kt = Pt|t−1 CH −1 = 0.
1 St|t−1 (the usual KF). Finally, the WLKF is
x̂t|t = x̂t|t−1 + Kt n̂t|t−1 and the error covariance matrix for One special case is that the error covariance of the predictor
êt|t = x̂t|t − x̂t is êt|t−1 is proper, P
e t|t−1 = 0, and the innovation n̂t|t−1 is
" # proper, St|t−1 = 0. This is true when êt−1|t−1 , ut , and
e
Pt|t Pe t|t nt are all proper. Another special case is that n̂t|t−1 is
Pt|t = (I − Kt C) Pt|t−1 = e ∗ , (41)
P t|tP∗ t|t maximally improper, i.e., n̂t|t−1 = αn̂∗t|t−1 with probability
 5

1 for constant scalar |α| = 1. This is irrespective of whether connection between the complex augmented and real compos-
êt|t−1 is improper. For instance, assume yt = Re(xt ) + nt ite representations is
with nt real. We know that yt is a noisy widely linear " #
Rxx R e xx
transformation of xt and ŷt|t−1 = Re(x̂t|t−1 ), meaning the Rxx = e ∗ = ExxH = E[(Tz)(Tz)H ] = TRzz TH ,
innovation n̂t|t−1 is real and hence maximally improper. One Rxx Rxx
can also readily see that no widely linear processing is needed (47)
because yt is only dependent on the real part of xt . and hence
Example 1 (dual channel KF using WLKF): This example Rxx = Ruu + Rvv + j(RTuv − Ruv ),
extends the duality between widely linear and dual channel e xx = Ruu − Rvv + j(RT + Ruv ).
R uv
adaptive filtering made by Mandic et. al. [27] to Kalman filters.
Start with two real channels worth of random states ut and By applying the above connection between real composite and
vt , with the same dimension. Denote zTt = [uTt vtT ] as the complex augmented vectors, the conventional Kalman filter
corresponding real composite state. Suppose the composite [19], [28] of the dual real channel model (43)-(44) can be
state and measurement equations are written in the following complex augmented form:
x̂0|0 = 0, P0|0 = Π0 , (48)
 
u
zt = t = Ezt−1 + Fω t−1
vt H
Pt|t−1 = APt−1|t−1 A + BQB , H
     
E11 E12 ut−1 F11 F12 µt−1
= + , t = 1, 2, ..., x̂t|t−1 = Ax̂t−1|t−1 ,
E21 E22 vt−1 F21 F22 σ t−1
(43) ŷt|t−1 = Bx̂t|t−1 ,
and St|t−1 = CPt|t−1 CH + R,
      
ψt =
ξt
= Gzt + η t =
G11 G12 ut ρ
+ t , Kt = CPt|t−1 CH S−1
t|t−1 ,
κt G21 G22 vt φt
x̂t|t = x̂t|t−1 + Kt (yt − ŷt|t−1 ).
t = 0, 1, ...,
(44) Pt|t = (I − Kt C)Pt|t−1 .

where ω Tt = [µTt σ Tt ] and η Tt = [ρTt φTt ] are the composite The matrices Pt|t−1 , Pt|t , St|t−1 , Q, and R are augmented
driving and measurement noises, and ψ Tt = [ξ Tt κTt ] is the covariance matrices of the augmented errors xt − x̂t|t−1 and
composite measurement. This dynamical model allows the xt − x̂t|t , innovation yt − ŷt|t−1 , and noises wt and nt
states and measurements on the respective real channels to respectively. The matrix Kt is the augmented Kalman gain.
be arbitrarily coupled. In the following we will show that a By checking Algorithm 1, we see that the equations in (48)
WLKF can be used as the KF for zt . For the real composite are identical with the WLKF for the complex augmented
vectors zt , ω t , η t , and ψ t , establish their complex augmented dynamical model in (45)-(46). This means that the KF for
representations as a dual real channel model can be always implemented as
a WLKF for the corresponding complex augmented model.
T
zTt = [uTt vtT ] −−→ xTt = [xTt xH
t ],
Conversely, the WLKF may be organized into a real KF
over dual channels. Therefore, the duality between dual real
T
ω Tt = [µTt σ Tt ] −−→ wTt = [wtT wtH ], channel KF and the single complex channel WLKF holds if
and only if the conditions of state and measurement matrices,
T
η Tt = [ρTt φTt ] −−→ nTt = [nTt nH
t ],
A = 12 TETH , B = 12 TFTH , and C = 21 TGTH are
satisfied. This is implicit in [1]–[4], [6], [7] and explicit in
T
ψ Tt = [ξ Tt κTt ] −−→ yTt = [ytT ytH ], [27]. The key reason for the duality observed here is that for
any augmented random vector x and its estimator error e, the
The resulting augmented state and measurement equations are orthogonality principle says ExeH = TEz(z − ẑ)T TH = 0.
This happens if and only if Ez(z − ẑ)T = 0.
xt = Axt−1 + Bwt−1 , t = 1, 2, ..., (45) Example 2 (Performance Comparison between WLKF and
KF): Here we are interested in the evolution of the aug-
yt = Cxt + nt , t = 0, 1, ..., (46)
mented covariance matrix Pt|t for estimator error êt|t . Given
with A = 12 TETH , B = 12 TFTH , and C = 21 TGTH . Pt−1|t−1 , the augmented covariance matrix Pt|t is
For a real composite random vector zT = [uT vT ] with Pt|t =(I − Kt Ct )Pt|t−1
zero mean, its covariance matrix Rzz is −1
=(I − Pt|t−1 CH
t St|t−1 Ct )Pt|t−1
EuuT EuvT
   
T Ruu Ruv h
−1
i
Rzz = Ezz =
EvuT EvvT
=
RTuv Rvv = I − Pt|t−1 CH H
t (Ct Pt|t−1 Ct + R) Ct Pt|t−1
= I − (At Pt−1|t−1 AH H H H

The composite covariance matrix Rzz is determined by three t + Bt QBt )Ct (Ct (At Pt−1|t−1 At
−1
+ Bt QBH H
Ct · (At Pt−1|t−1 AH H

real blocks: Ruu , Ruv , and Rvv . Recall the complex aug- t )Ct + R) t + Bt QBt )
mented representation for z is x = Tz. The second order (49)
 6

Suppose we want to estimate a complex random scalar xt . The Substituting (53) into (52), we have the minimum ξtWL :
corresponding state equations are
1
min ξtWL =


xt = at−1 xt−1 + bt−1 wt−1 , t = 1, 2, ..., (50) qt (2P0|0 ) + qt (0)
2
yt = ct xt + nt , t = 0, 1, ... (51) The ratio of min ξtWL to ξtSL is:
where at , bt , ct ∈ C for all t ≥ 0. The augmented matrices are min ξtWL qt (2P0|0 ) + qt (0)
At = diag(at , a∗t ), Bt = diag(bt , b∗t ), and Ct = diag(ct , c∗t ). θt = SL
=
ξt 2qt (P0|0 )
We make the assumption that Q = N1 I, and R = N2 I, where
N1 , N2 > 0. From (49) we may find the recursion for the 2 It’s obvious that 21 ≤ θt ≤ 1. This is because qt is concave, and
by 2 augmented covariance matrix Pt|t as qt (2P0|0 ) + qt (0) ≤ 2qt (P0|0 ) for any P0|0 . Also qt (2P0|0 ) ≥
Pt|t = I − |ct |2 (|at |2 Pt−1|t−1 + |bt |2 N1 I)(|ct |2 (|at |2 Pt−1|t−1
 qt (P0|0 ) for any P0|0 and qt (0) ≥ 0. Actually the condition for
achieving the lower bound is N1  N2  1. This coincides
+ |bt |2 N1 I) + N2 I)−1 · (|at |2 Pt−1|t−1 + |bt |2 N1 I),

with the MMSE analysis in [4]. Here we have assumed: (i) the
t = 1, 2, ... complex state equations (50) and (51) are strictly linear, (ii)
the state is scalar, and (iii) the noises wt and nt are white and
Suppose the initial augmented state covariance is
" # proper for all t. Only the prior state is improper. For a vector-
P0|0 Pe0|0 valued state coupled with widely linear update equations and
P0|0 = e∗ ∗ , improper colored noises, the advantage of the WLKF over the
P0|0 P0|0
SLKF may be much greater.
with eigenvalues {λ01 , λ02 }. We can show that given the eigen-
values {λt−1 t−1
1 , λ2 } of matrix Pt−1|t−1 , the eigenvalues of
Pt|t are V. E XTENDED W IDELY L INEAR K ALMAN F ILTER
λti = gt (λt−1i ), i = 1, 2, In this section it is assumed that the complex state sequence
where the function gt is given by {xt } evolves according to the nonlinear model

|ct |2 (|at |2 λ + |bt |2 N1 )

 
xt = ft−1 (xt−1 , wt−1 ), t = 1, 2, ..., (54)
gt (λ) = 1 − (|at |2 λ + |bt |2 N1 )
|ct |2 (|at |2 λ + |bt |2 N1 ) + N2
N2 (|at |2 λ + |bt |2 N1 ) where ft−1 : Cn → Cn is a known time-varying nonlinear
= , t = 1, 2, ... transformation and wt is the complex random sequence of
|ct |2 (|at |2 λ + |bt |2 N1 ) + N2
driving noise with moments up to second order independent
Thus the eigenvalues {λt1 , λt2 } may be conveniently expressed of xt . The state sequence is observed through the memoryless
as the function recursion nonlinear transformation ht to produce the complex measure-
λti = gt ◦ gt−1 ◦ · · · ◦ g1 (λ0i ) ment sequence yt :
, qt (λ0i ) yt = ht (xt , nt ), t = 0, 1, ... . (55)
Observe that gt is an increasing concave function for each t.
Thus we conclude that qt is concave for each t. Next we want Here the measurement noise nt is again a sequence of random
to compute the widely linear minimum mean squared error vectors with moments up to second order independent of xt .
(WLMMSE) at the t-th iteration. This may be written The definition of complex Jacobians can be found in [4],
[29]. Suppose g : Cn → Cm is a vector valued function of a
1
ξtWL = E k êt|t k2 = tr(Pt|t ) complex vector x = u + jv. Then the complex Jacobian of g
2 (52) is the m by 2n matrix:
1
= (qt (λ01 ) + qt (λ02 )).
2  ∂g 
1

∂g1 ∂g1

Assume the initial scalar Hermitian covariance is P0|0 . Then  ∂x   ∂x ∂x∗ 
∂g 
=  ... 
  . .. 
the t-th LMMSE for the strictly linear Kalman filter is
 =  .. . 
 
∂x  ∂g   ∂g
ξtSL = qt (P0|0 ). m m ∂gm 
∂x ∂x ∂x∗
To achieve the maximum performance improvement of the   (56)
∂g1 ∂g1 ∂g1 ∂g1
widely linear Kalman filter over the strictly linear Kalman  ∂x1 · · · ∂xn ∂x∗1
···
∂x∗n 
filter for the special case discussed here, we shall minimize  
=  ... .. .. .. .. ..  ,

ξtWL with fixed P0|0 and variable Pe0|0 . It can be seen that at  . . . . . 
each t, ξtWL is a Schur-concave function with respect to all  ∂gm ∂gm ∂gm ∂gm 
··· ···
λ0i . Since λ01 + λ02 ≤ 2P0|0 [4], the minimum is achieved by ∂x1 ∂xn ∂x∗1 ∂x∗n
letting
∂
= 12 ( ∂u
∂ ∂
), and ∂x∂ ∗ = 12 ( ∂u
∂
 0  
λ1 2P0|0 where the operators ∂x − j ∂v +
= . (53) ∂
λ02 0 j ∂v ). Then we may approximate these nonlinear equations by
 7

their Taylor expansion around the state estimate x̂t−1|t−1 : Therefore we can derive the extended WLKF along the lines
of Section III and summarize the filter as Algorithm 2 on the
∂ft−1
xt =ft−1 (x̂t−1|t−1 , 0) + (x̂t−1|t−1 , 0)(xt−1 − x̂t−1|t−1 ) top of next page. It can be observed from Alg. 2 that the
∂x nonlinear functions f and h are used to predict xt and yt as
∂ft−1 in the conventional EKF. The complex Jacobians are used to
+ (x̂t−1|t−1 , 0)wt−1
∂w modify the Hermitian and complementary covariance matrices
∂ft−1 and Kalman gain. This treatment of the EWLKF is equivalent
= (x̂t−1|t−1 , 0)xt−1 + ft−1 (x̂t−1|t−1 , 0)
∂x to the original treatment of Mandic, et al. [9], [26]. The only
∂ft−1 ∂ft−1
− (x̂t−1|t−1 , 0)x̂t−1|t−1 + (x̂t−1|t−1 , 0)wt−1 difference is that the authors of [9], [26] define updates for
∂x ∂w augmented covariances, which of course can be resolved for
=Ft,1 xt−1 + Ft,2 x∗t−1 + v̂t−1 + Lt,1 wt−1 + Lt,2 wt−1 ∗ updates of Hermitian and complementary covariances, whereas
=Ft xt−1 + v̂t−1 + Lt wt−1 , our updates for Hermitian and complementary covariances can
be assembled into updates for augmented covariances.
where
  ∂ft−1
Ft = Ft,1 Ft,2 = (x̂t−1|t−1 , 0) VI. U NSCENTED W IDELY L INEAR K ALMAN F ILTER
∂x

∂ft−1 ∂ft−1
 The major defect of extended Kalman filters is that, since
= (x̂t−1|t−1 , 0) (x̂t−1|t−1 , 0) , the approximated model is obtained by linearizing state and
∂x ∂x∗
measurement equations, the posterior means and covariances
  ∂ft−1 are accurate only to the first order in a Taylor expansion.
Lt = Lt,1 Lt,2 = (x̂t−1|t−1 , 0)
∂w Unscented Kalman filters are motivated by the inaccuracy
 
∂ft−1 ∂ft−1 issues of the EKF. An UKF uses the unscented transformation
= (x̂t−1|t−1 , 0) (x̂t−1|t−1 , 0) ,
∂w ∂w ∗ (UT) to generate a fixed set of sigma points to represent the
distribution of a random variable [30], [31]. After propagating
and the term v̂t−1 = ft−1 (x̂t−1|t−1 , 0) − Ft x̂t−1|t−1 depends
sigma points through nonlinearities, the estimated posterior
only on the estimator x̂t−1|t−1 . Similarly we may approximate
mean and covariance are precise at least to second order in
yt as
a Taylor expansion. The common approach is to compose
∂ht complex random states and noises into a complex vector
yt =ht (x̂t|t−1 , 0) + (x̂t|t−1 , 0)(xt − x̂t|t−1 )
∂x sT = [xT wT nT ] and construct the sigma points of s. The
∂ht first and second order statistical information of s is
+ (x̂t|t−1 , 0)nt
∂n
∂ht µTs = [µTx µTw µTn ], (59)
= (x̂t|t−1 , 0)xt + ht (x̂t|t−1 , 0)
∂x    
Rxx 0 0 R
e xx 0 0
∂ht ∂ht
− (x̂t|t−1 , 0)x̂t|t−1 + (x̂t|t−1 , 0)nt Rss =  0 Q 0, R e ss = 
 0 Q
e 0 .
∂x ∂n
0 0 R 0 0 R
=Ht,1 xt + Ht,2 x∗t + ŵt + Mt,1 nt + Mt,2 n∗t
e

=Ht xt + ŵt + Mt nt , For the complex random vector s = u + jv ∈ CN , its cor-

where responding composite real random vector is zT = [uT vT ] ∈
R2N with mean µTz = [µTu µTv ]. We know that the augmented
  ∂ht
Ht = Ht,1 Ht,2 = (x̂t|t−1 , 0) covariance matrix is
∂x " #  
 
∂ht ∂ht Rss R e ss H Ruu Ruv
= (x̂t|t−1 , 0) (x̂t|t−1 , 0) , Rss = e ∗ = TR zz T = T TH .
∂x ∂x∗ Rss R∗ss Rvu Rvv
  ∂ht (60)
Mt = Mt,1 Mt,2 = (x̂t|t−1 , 0) The inversion is
∂n
  1 H
∂ht ∂ht Rzz = T Rss T. (61)
= (x̂t|t−1 , 0) (x̂t|t−1 , 0) . 4
∂n ∂n∗
Given the augmented covariance matrix Rss , there are mul-
The term ŵt = ht (x̂t|t−1 , 0) − Ht x̂t|t−1 depends only on
tiple ways to generate sigma points for the augmented ran-
the predictor x̂t|t−1 . Accordingly we rewrite the approximated
dom vector s. But a hidden restriction imposed here is that
widely linear state equations as
these sigma points should be augmented vectors. Otherwise
xt − v̂t−1 = Ft,1 xt−1 + Ft,2 x∗t−1 + Lt,1 wt−1 + Lt,2 wt−1
∗
, they cannot be propagated through the unscented WLKF.
t = 1, 2, ..., One approach for this is to start with sigma points of the
(57) composite random vector z. Using a Cholesky decomposition
the composite covariance matrix Rzz may be factored as
yt − ŵt = Ht,1 xt +Ht,2 x∗t +Mt,1 nt +Mt,2 n∗t , t = 0, 1, ... .
(58) Rzz = BBH
 8

Algorithm 2 Extended Widely Linear Kalman Filter

Initialize with x̂0|0 = 0, P0|0 = Π0 and P
e 0|0 = Πe 0 as defined in equation (32);
At the t-th iteration, t = 1, 2, · · · ,
1. Update the predicted state as
x̂t|t−1 = ft−1 (x̂t−1|t−1 , 0);
2. Update the Hermitian and complementary error covariance matrices as
Pt|t−1 = Ft,1 Pt−1|t−1 FH e∗ H H ∗ H
t,1 + Ft,2 Pt−1|t−1 Ft,1 + Ft,1 Pt−1|t−1 Ft,2 + Ft,2 Pt−1|t−1 Ft,2
e
+ Lt,1 QLH + Lt,2 Q e ∗ LH + Lt,1 QL
e H + Lt,2 Q∗ LH ,
t,1 t,1 t,2 t,2

e t|t−1 = Ft,1 Pt−1|t−1 FT + Ft,2 P

P e∗ T T ∗ T
t−1|t−1 Ft,1 + Ft,1 Pt−1|t−1 Ft,2 + Ft,2 Pt−1|t−1 Ft,2
e
t,1

+ Lt,1 QLT + Lt,2 Q e ∗ LT + Lt,1 QL

e T + Lt,2 Q∗ LT ,
t,1 t,1 t,2 t,2

St|t−1 = Ht,1 Pt|t−1 HH e∗

+ Ht,2 P H H ∗ H
t|t−1 Ht,1 + Ht,1 Pt|t−1 Ht,2 + Ht,2 Pt|t−1 Ht,2
e
t,1

+ Mt,1 RMH e∗ H e H ∗ H
t,1 + Mt,2 R Mt,1 + Mt,1 RMt,2 + Mt,2 R Mt,2 ,

S e∗
e t|t−1 = Ht,1 Pt|t−1 HT + Ht,2 P T T ∗ T
t|t−1 Ht,1 + Ht,1 Pt|t−1 Ht,2 + Ht,2 Pt|t−1 Ht,2
e
t,1

+ Mt,1 RMT + Mt,2 R e ∗ MT + Mt,1 RM e T + Mt,2 R∗ MT ,

t,1 t,1 t,2 t,2

3. Derive the widely linear Kalman gains Kt and K

e t as
−1 −∗ e ∗ −1
Kt = (Pt|t−1 HH e H T e T
t,1 + Pt|t−1 Ht,2 )PS − (Pt|t−1 Ht,2 + Pt|t−1 Ht,1 )St|t−1 St|t−1 PS ,

e t|t−1 HT )P−∗ − (Pt|t−1 HH + P

e t = (Pt|t−1 HT + P
K e t|t−1 HH )S−1 S −∗
t|t−1 t|t−1 PS ,
e
t,1 t,2 S t,2 t,1

4. Use the predictions x̂t|t−1 and ŷt|t−1 = ht (x̂t|t−1 , 0) to generate the widely linear estimator of xt
e t yt − ŷt|t−1 ∗ ,



x̂t|t = x̂t|t−1 + Kt yt − ŷt|t−1 + K
5. Update the Hermitian and complementary error covariance matrices
   
Pt|t = I − Kt Ht,1 − K e t H∗ Pt|t−1 − Kt Ht,2 + K e t H∗ P e∗
t,2 t,1 t|t−1 ,
   
P e t H∗ P
e t|t = I − Kt Ht,1 − K e t H ∗ P∗
e t|t−1 − Kt Ht,2 + K
t,2 t,1 t|t−1 .

Denote the vector bk as the k-th column of matrix B for augmented vectors {X k } as
k = 1, 2, ..., 2N . Then the sigma points {Z k } of z are [30]–  
Xk
[32] X k = TZ k =
X ∗k

Z 0 = µz , (62) µs , √
 k = 0,
√ = µs + 2N + λTbk , k = 1, ..., 2N,
Z k = µz + 2N + λbk , k = 1, ..., 2N,  √
√ 
µs − 2N + λTbk , k = 2N + 1, ..., 4N.
Z k = µz − 2N + λbk , k = 2N + 1, ..., 4N, (63)
We can show that all the X k compose the sigma points of the
corresponding to the following weights: augmented vector s, since X 0 = µs and

Rss = TRzz TH
(
λ/(2N + λ), k = 0,
mean weights: Wm (k) = 2n
1/[2(2N + λ)], k = 1, ..., 4N, X
= T( bk bH
k )T
H

covariance weights: k=1

(   H
λ/(2N + λ) + (1 − α2 + β), k = 0, = Tb1 Tb2 ··· Tb2N Tb1 Tb2 ··· Tb2N .
Wc (k) =
Wm (k), k = 1, ..., 4N,
Therefore we have obtained the sigma points {X k } of s
w.r.t weights {Wm (k), Wc (k)} from widely linear transfor-
where λ = α2 (2N + κ) − 2N , and α, β, and κ are parameters mation of the real composite sigma points {Z k } of z w.r.t
controlling the distribution of sigma points. Define a set of weights {Wm (k), Wc (k)}. Note that each sigma point X k
 9

Algorithm 3 Unscented Widely Linear Kalman Filter Hermitian covariance Rss , but also complementary covariance
Initialize with x̂0|0 = 0, P0|0 = Π0 and P e 0|0 = Πe 0 as defined Re ss . In [9] the authors proposed complex sigma points of s
in equation (32); which carry moments µs and Rss , but not R e ss .
At the t-th iteration, t = 1, 2, · · · , But what really concerns us is whether the modified sigma
Replace Pt−1|t−1 with Rxx , and P e t−1|t−1 with R e xx , and points will refine the propagation of mean and covariance
µTs,t = [x̂Tt−1|t−1 0 0] with µTs in (59). Construct effective through the nonlinearities f and h. Suppose we push each
sigma points {Xk,t T x
} = {[(Xk,t−1 )T (Xk,t−1
w
)T (Xk,t
n T
) ]} by modified sigma point X k through any non-linearity Y k =
using equations (59)-(63). g(X k ). Compute the following deterministic sample averages:
Prediction updates: 4N
X 4N
X
µ̂y = Wm (k)Y k = Wm (k)g(X k ),
X xk,t|t−1 = ft−1 (X xk,t−1 , X w
k,t−1 ) k=0 k=0
4N
X 4n
X
x̂t|t−1 = Wm (k)X xk,t|t−1 , Pyy = Wc (k)(Y k − µ̂y )(Y k − µ̂y )H ,
k=0 k=0
4N 4n
X
X
Pt|t−1 = Wc (k)(X xk,t|t−1 − x̂t|t−1 )(X xk,t|t−1 − x̂t|t−1 )H , P
e yy = Wc (k)(Y k − µ̂y )(Y k − µ̂y )T .
k=0 k=0

4N
X From the argument of [30], it can be verified that the sample
P
e t|t−1 = Wc (k)(X xk,t|t−1 − x̂t|t−1 )(X xk,t|t−1 − x̂t|t−1 )T , averages µ̂y , and Pyy , and Pe yy may be used to approximate
k=0 the mean E(f (s)), the Hermitian, and complementary covari-
Y k,t|t−1 = gt (X xk,t|t−1 , X nk,t ), ance matrices of f (s). The approximation is precise up to at
least second order in a Taylor expansion. The unscented widely
4N
X linear Kalman filter is described in Algorithm 3, in notation
ŷt|t−1 = Wm (k)Y k,t|t−1 , that follows [32].
k=0 Example 3 (Phase Demodulation Problem): Consider a
Measurement updates: scalar real random phase θt which is updated as
4N
X θt = aθt−1 + bwt−1 , t = 1, 2, ..., (64)
St|t−1 = Wc (k)(Y k,t|t−1 − ŷt|t−1 )(Y k,t|t−1 − ŷt|t−1 )H ,
k=0 where wt is a real driving noise. So, the phase is real. The
4N
measurement of θt is a noisy complex signal modulated by
X θt :
S
e t|t−1 = Wc (k)(Y k,t|t−1 − ŷt|t−1 )(Y k,t|t−1 − ŷt|t−1 )T ,
yt = eiθt + nt , t = 0, 1, ... (65)
k=0
4N
X where each nt is assumed to be a zero mean, scalar complex
Et|t−1 = Wc (k)(X xk,t|t−1 − x̂t|t−1 )(Y k,t|t−1 − ŷt|t−1 )H , Gaussian random variable [2], [4] with Hermitian variance
k=0 R and complementary variance R. e Suppose nt = ut + jvt ,
4N where ut and vt have variances Ruu and Rvv respectively and
covariance Ruv . The complex correlation coefficient between
X
E
e t|t−1 = Wc (k)(X xk,t|t−1 − x̂t|t−1 )(Y k,t|t−1 − ŷt|t−1 )T ,
k=0 nt and n∗t is

Kt = Et|t−1 P−1 −∗ e ∗ −1 R Ruu − Rvv + 2jRuv

S − Et|t−1 St|t−1 St|t−1 PS ,
e e
ρ= =
R Ruu + Rvv
K e t|t−1 P−∗ − Et|t−1 S−1 S
et = E −∗
t|t−1 t|t−1 PS , which describes the impropriety of nt . It can be shown that
e
S

e t (yt − ŷt|t−1 )∗ , |ρ| ≤ 1. When |ρ| = 1, nt is maximally improper. In the

x̂t|t = x̂t|t−1 + Kt (yt − ŷt|t−1 ) + K
following we let Ruu = Rvv and set ρ by changing the value
Pt|t = Pt|t−1 − Et|t−1 KH eH
t − Et|t−1 Kt ,
e of Ruv . The signal-to-noise ratio at the receiver is SNR =
R−1 . In simulation we set a = 0.98, b = 0.05. Each wt is
P
e t|t = P eT −E
e t|t−1 − Et|t−1 K e t|t−1 KT . a standard Gaussian real random variable, independent of all
t t
others.
Fig. 1 draws the outputs of the UWLKF over time at SNR =
30dB and |ρ| = 0.5. The UWLKF is constructed according to
Algorithm 3. The widely linear Kalman gain for the UWLKF
is an augmented vector. Thus it follows that the complex is a 2 by 2 matrix and the estimate θ̂t|t is always real. It can
set {X k }, generated by extracting the top halves of {X k }, be observed that for most iterations, the estimate θ̂t|t is close
is sufficient to capture both first and second order statistical to the phase θt .pAlso the true θt is almost confined by the
information of the augmented random vector s. We call {X k } envelope θ̂t|t ± Pt|t .
the modified sigma points of s. The impact of these modified Fig. 2 compares the performances of the UWLKF that
sigma points is that {X k } preserves not only mean µs and accounts for the impropriety of the noise, and the UKF that

Fig. 1. Phase estimated by UWLKF at each iteration, SNR = 30dB, |ρ| = 0.5.
Fig. 2. Normalized estimation error ξ of UWLKF and UKF vs SNRs, |ρ| =
0.7.
assumes the noise to be proper. Unlike the UWLKF above, the
UKF estimates θt from a real 2 by 1 measurement vector con-
sisting of the real and imaginary part of yt collected from dual
channels. At each iteration the UKF produces sigma points
from the real mean vector [θ̂t|t µw µu µv ]T = [θ̂t|t 0 0 0]T and
covariance matrix M = diag(Pt|t , 1, Ru , Rv ), and it has a 1
by 2 Kalman gain vector. The complex correlation coefficient
is |ρ| = 0.7. Define the normalized estimation error power
as ξ = ||e||22 /||θ||22 , where θ and e are vectors consisting
of phases and estimation errors in 500 iterations respectively.
In the plot each ξ is computed by averaging 1000 Monte-
Carlo simulations. It can be seen that in the low-medium SNR
regime, UWLKF requires about 2dB less SNR than the UKF.
Fig. 3 shows the performance improvement of UWLKF
over UKF vs the noise impropriety |ρ| at different SNRs. We
use the factor r = ξUKF /ξUWLKF to evaluate the advantage
of UWLKF. The normalized estimation error power ξUKF and
ξUWLKF are defined as above. Each r is computed by averaging
1000 Monte-Carlo simulations. When SNR is either 20dB or
40dB, the UWLKF outperforms the UKF for all 0 < |ρ| ≤ 1.
The performance improvement becomes more significant as
|ρ| increases.
VII. C ONCLUSION
In this paper we have designed widely linear, extended, and
unscented Kalman filters for complex noisy dynamical systems
10
Fig. 3. Performance improvement r of UWLKF over UKF vs impropriety
of nt .
with improper states and noises. The corresponding error
covariance matrices and modified sigma points are derived. An
example has demonstrated that two channels of real dynamical
state, with arbitrarily coupled state equations may be linearly
Kalman filtered in real variables or widely linearly Kalman
filtered in complex variables. A simulation for real phase
demodulation shows how a complex, augmented, unscented
Kalman filter produces real estimates from complex baseband
measurements and shows the improvement of its performance
over the complex, unscented Kalman filter assuming proper
states and noises.
R EFERENCES
[1] B. Picinbono and P. Bondon, “Second-order statistics of complex sig-
nals,” IEEE Trans. on Signal Processing, vol. 45, no. 2, pp. 411–420,
Feb. 1997.
[2] B. Picinbono, “Second-order complex random vectors and normal distri-
butions,” IEEE Trans. on Signal Processing, vol. 44, no. 10, pp. 2637–
2640, Oct. 1996.
[3] P. J. Schreier and L. L. Scharf, “Second-order analysis of improper com-
plex random vectors and processes,” IEEE Trans. on Signal Processing,
vol. 51, no. 3, pp. 714–725, Mar. 2003.
[4] ——, Statistical Signal Processing of Complex-Valued Data: The Theory
of Improper and Noncircular Signals. Cambridge University Press,
2010.
[5] E. Ollila and V. Koivunen, “Generalized complex elliptical distributions,”
in Sensor Array and Multichannel Signal Proc. Workshop, Jul. 2004, pp.
460–464.
[6] B. Picinbono and P. Chevalier, “Widely linear estimation with complex
data,” IEEE Trans. on Signal Processing, vol. 43, no. 8, pp. 2030–2033,
Aug. 1995.
[7] B. Picinbono, “Wide-sense linear mean square estimation and predic-
tion,” in Int. Conf. on Acoustics, Speech, and Signal Proc., vol. 2, May
1995, pp. 2032–2035.
[8] J. Navarro-Moreno, J. Moreno-Kayser, R. M. Fernandez-Alcala, and
J. C. Ruiz-Molina, “Widely linear estimation algorithms for second-
order stationary signals,” IEEE Trans. on Signal Processing, vol. 57,
no. 12, pp. 4930–4935, Dec. 2009.
[9] D. P. Mandic and S. L. Goh, Complex Valued Nonlinear Adaptive Filters:
Noncircularity, Widely Linear and Neural Models. Wiley, 2009.
[10] A. S. Aghaei, K. N. Plataniotis, and S. Pasupathy, “Maximum likelihood
binary detection in improper complex gaussian noise,” in Int. Conf. on
Acoustics, Speech, and Signal Proc., Mar. 2008, pp. 3209–3212.
[11] P. J. Schreier, L. L. Scharf, and C. T. Mullis, “Detection and estimation
of improper complex random signals,” IEEE Trans. on Information
Theory, vol. 51, no. 1, pp. 306–311, Jan. 2005.
[12] S. Buzzi, H. V. Poor, and A. Zappone, “Transmitter optimization, widely-
linear detection, and power control for energy efficient networks: a
game-theoretic approach,” IEEE Trans. on Signal Processing, vol. 43,
no. 8, pp. 2030–2033, Aug. 1995.
 11

[13] A. A. de Lima and R. C. de Lamare, “Adaptive detection using widely

linear processing with data reusing for DS-CDMA systems,” in Int.
Telecommunications Symposium 2006, Sep. 2006, pp. 187–192.
[14] S. Buzzi, M. Lops, and S. Sardellitti, “Widely linear reception strategies
for layered space-time wireless communications,” IEEE Trans. on Signal
Processing, vol. 54, no. 6, pp. 2252–2262, Jun. 2006.
[15] W. H. Gerstacker, R. Schober, and A. Lampe, “Equalization with widely
linear filtering,” in IEEE ISIT2001, Jun. 2001, p. 265.
[16] D. Mattera, L. Paura, and F. Sterle, “Widely linear MMSE equalizer for
MIMO linear time-dispersive channel,” in Int. Symposium on ISPA2003,
Sep. 2003, pp. 1176–1180.
[17] P. J. Schreier, L. L. Scharf, and C. T. Mullis, “A unified approach to
performance comparisons between linear and widely linear processing,”
in IEEE Workshop on Statistical Signal Processing, 2003, pp. 114–117.
[18] Y. Xia, B. Jelfs, M. V. Hulle, J. C. Prłncipe, and D. P. Mandic, “An
augmented echo state network for nonlinear adaptive filtering of complex
noncircular signals,” IEEE Trans. on Neural networks, vol. 22, no. 1,
pp. 74–83, Jan. 2011.
[19] R. E. Kalman, “A new approach to linear filtering and prediction
problems,” Trans. of the ASME-Journal of Basic Engineering, pp. 35–45,
Mar. 1960.
[20] F. Daum, “Nonlinear filters: Beyond the Kalman filter,” IEEE Aerospace
and Electronics Systems Magazine, pp. 57–69, 2005.
[21] S. J. Julier and J. K. Uhlmann, “Unscented filtering and nonlinear
estimation,” Proceedings of the IEEE, vol. 92, no. 3, pp. 401–422, 2004.
[22] M. S. Arulampalam, S. Maskell, N. Gordon, and T. Clapp, “A tutorial
on particle filters for online nonlinear/non-gaussian bayesian tracking,”
IEEE Trans. on Signal Processing, vol. 50, no. 2, pp. 174–188, 2002.
[23] B. Ristic and S. Arulampalam, Beyond the Kalman Filter: Particle
Filters for Tracking Applications. Artech House, 2004.
[24] D. Simon, “Kalman filtering with state constraints: a survey of linear
and nonlinear algorithms,” IET Control Theory and Applications, vol. 4,
no. 8, pp. 1303–1308, 2010.
[25] F. Gustafsson, “Particle filter theory and practice with positioning appli-
cations,” IEEE Aerospace and Electronics Systems Magazine, vol. 25,
no. 7, pp. 53–82, 2010.
[26] S. L. Goh and D. P. Mandic, “An augmented extended Kalman filter
algorithm for complex-valued recurrent neural networks,” in Int. Conf.
on Acoustics, Speech and Signal Proc., vol. 5, May. 2006, pp. 561–564.
[27] D. P. Mandic, S. Still, and S. C. Douglas, “Duality between widely linear
and dual channel adaptive filtering,” in Int. Conf. on Acoustics, Speech
and Signal Proc., vol. 5, Apr. 2009, pp. 1729–1732.
[28] T. Kailath, A. H. Sayed, and B. Hassibi, Linear Estimation. Prentice
Hall, 2000.
[29] E. A. Wan, “Time series prediction by using a connectionist network
with internal delay lines,” in Time Series Prediction. Addison-Wesley,
1994, pp. 195–217.
[30] S. J. Julier and J. K. Uhlmann, “Unscented filtering and nonlinear
estimation,” Proc. IEEE, vol. 92, no. 3, pp. 401–422, 2004.
[31] S. J. Julier, “The scaled unscented transformation,” in Proc. American
Control Conf., vol. 6, May 2002, pp. 4555–4559.
[32] E. A. Wan and R. van der Merwe, “The unscented Kalman filter for
nonlinear estimation,” in Adaptive Systems for Signal Proc., Comm.,
and Control Symp., Oct. 2000, pp. 153–158.

View publication stats