PowerPoint 簡報 - 清華大學電機系
Download
Report
Transcript PowerPoint 簡報 - 清華大學電機系
BLIND SOURCE SEPARATION BY KURTOSIS
MAXIMIZATION WITH APPLICATIONS IN WIRELESS
COMMUNICATIONS
Chong-Yung Chi (祁忠勇)
Institute of Communications Engineering &
Department of Electrical Engineering
National Tsing Hua University
Hsinchu, Taiwan 30013, R.O.C.
Tel: +886-3-5731156, Fax: +886-3-5751787
E-mail: [email protected]
http://www.ee.nthu.edu.tw/cychi/
Invited talk at I2R, Singapore, July 18, 2006.
Acknowledgments: The viewgraphs were prepared through Chun-Hsien Peng’s helps.
OUTLINE
1.
Introduction to Blind Source Separation (BSS)
2.
FKMA and MSC Procedure
3. Turbo Source Extraction Algorithm (TSEA)
4.
Non-cancellation Multistage Source (NCMS)
Separation Algorithms
NCMS-FKMA
NCMS-TSEA
5.
Simulation Results --- Part 1
6.
Turbo Space-time Receiver for CCI/ISI Reduction
7. Simulation Results --- Part 2
8.
Conclusions
FKMA: Fast Kurtosis Maximization Algorithm
MSC: Multistage Successive Cancellation
1
1. Blind Source Separation (BSS)
Instantaneous Mixture of Sources
Noise w1[n ]
(Mutually Indep.
but Colored)
Unknown
s1[n ]
s[n ]
P K
A
(Memoryless channel)
Measurements
x1[n ]
mixing matrix
sK [n ]
P Output
x[n ]
x P [n ]
wP [n ]
K
x[n] A s[n] w[n] ai si [n] w[n]
i 1
ai (ith column of A)
GOAL
is the basis vector that spans the subspace of si [n ]
Extract all the source signals si [n ] with only measurements x[n ].
Applications: array signal processing, wireless communications and biomedical
signal processing, etc [1-3].
2
Existing BSS Algorithms
SOS
Algorithm
Whitening
AMUSE
SOBI
FOBI
EFOBI
FastICA
HOS
MSC-FKMA
MSC-TSEA
NCMS-FKMA
NCMS-TSEA
Multistage
Statistically independent
sk [n ],
A PK
k 1,..., K
w[n ]
1. Statistically mutually uncorrelated
PK
2. Zero mean
3. Temporal colored with distinct
power spectra
1. Statistically mutually independent
2. Zero mean
3. Distinct fourth-order moments
PK
1. Statistically mutually independent
2. Zero mean
3. Non-Gaussian
(Non-zero fourth-order cumulants,
e.g.,
1. Zero
mea
n
2. Gaussian
kurtosis)
SOS: Second-order Statistics
HOS: Higher-order Statistics
3
AMUSE: Algorithm for Multiple Unknown Signals Extraction (Tong et al., 1990 [1])
SOBI: Second-order Blind Identification (Belouchrani et al., 1997 [2])
FOBI: Fourth-order Blind Identification (Cardoso, 1989 [12])
EFOBI: Extended Fourth-order Blind Identification (Tong et al., 1991 [1])
FastICA: Fast Independent Component Analysis (Hyvarinen et al., 1997 [13, 14])
MSC: Multistage Successive Cancellation
NCMS: Non-cancellation Multistage
FKMA: Fast Kurtosis Maximization Algorithm
TSEA: Turbo Source Extraction Algorithm
4
AMUSE and SOBI Algorithm Using SOS:
Step 1: Prewhitening by Eigenvalue Decomposition (EVD)
R x E {x[n ]x H [n ]} (PxP matrix)
EVD
{
1 , 2 , ..., K : largest K eigenvalues of R x
f1 , f 2 , ..., f K
: associated K eignevectors of R x
ˆ w2 : average of the other (smallest) P-K eigenvalues of R x
2
(assuming that R w E {w[n ]w H [n ]} w
I)
ˆ [
D
f1
1 ˆ w2
, ...,
fK
K ˆ w2
]H
ˆ x[n ] Us[n ] D
ˆ w[n ]
z[n ] D
U:
KxK unitary matrix
(whitening matrix)
(dimension-reduced whitening
spatial processing)
AMUSE: Algorithm for Multiple Unknown Signals Extraction (Tong et al., 1990 [1])
SOBI: Second-order Blind Identification (Belouchrani et al., 1997 [2])
5
Step 2: Estimation of the Unitary Matrix U from
R z [k ] E {z[n ]z H [n k ]}
x[n ]
Prewhitening
by EVD
(KxK matrix)
EVD of
z[n ]
Û
( R z [k ] R z [k ]) / 2
H
Û
Joint Diagonalization of
{R z [ki ] | i 1, ..., J }
(AMUSE)
(SOBI)
Step 3: Source Separation and Channel Estimation
ˆ H z[n ]
sˆ[n ] U
(spatial processing for simultaneous extraction of
all the K sources)
ˆ HD
ˆ (demixing matrix)
WU
ˆ W# D
ˆ #U
ˆ
A
(mixing matrix estimate)
( # : pseudo-inverse)
6
2. FKMA and MSC Procedure
Assumptions:
(A1) The unknown P K mixing matrix
A is of full column rank with P K .
(A2) si [n ], i {1, 2, ..., K } are modeled as
ui [n ]
Stable LTI System
bi [n ]
si [n ] ui [n ] bi [n ]
ui [n ] : zero-mean non-Gaussian independent identically distributed (i.i.d.)
process with C4{ui [n]} 0 ; ui [n ] is statistically independent of u j [n ]
for all i j.
(A3) w[n ] is zero-mean Gaussian and statistically independent of s[n ].
cum{z1 , z 2 , z 3 , z 4 }: fourth-order joint cumulant of random variables z1 , z 2 , z 3 , z 4
C4{z} cum{z, z, z , z} (referred to as kurtosis of z )
FKMA: Fast Kurtosis Maximization Algorithm (Chi and Chen, 2001 [4,5])
MSC: Multistage Successive Cancellation
7
Definition of HOS (i.e., Cumulants):
(Bartlett, 1955, Brillinger, 1975, etc)
M ln (1 ,, M )
cum{x1 ,, x M } ( j )
1 M
M
where
1 M 0
(1 ,, M ) E{e j (1x1 M x M ) }
is the characteristic function of random variables x1 ,, x M .
Assume that x1 , x 2 , x 3 , x 4 are zero-mean random variables. Then
cum{x1 , x 2 } E{x1x 2 }
cum{x1 , x 2 , x 3 } E {x1x 2 x 3 }
cum{x1 , x 2 , x 3 , x 4 } E {x1x 2 x 3 x 4 } E {x1x 2 }E {x 3 x 4 }
E {x1x 3 }E {x 2 x 4 } E {x1x 4 }E {x 2 x 3 }
4
2
C4 {x} cum{x, x, x* , x* } E{ x } 2( E{ x }) 2 E{x 2 }
2
(referred to as kurtosis of x )
8
Fast Kurtosis Maximization Algorithm (FKMA)
(Chi et al., 2001 [4, 5])
Criterion [7]:
J ( v) J (e[n])
C4 {e[n]}
2
Maximization
2
E { e[n] }
Optimum
v
e[n] v T x[n] k sk [n]
(noise-free case)
( k is an unknown complex scale
factor and k {1,, K } )
magnutude of normalized
kutorsis of e[n ]
Closed-form solution for v : Not existent
Gradient-type iterative algorithms for finding a local optimum v : Not very
computationally efficient
v (i ) v (i 1) Q(i 1) J ( v (i 1) )
where Q is a positive-definite matrix depending on the algorithm used, and
μ is the step size such that
J ( v (i ) ) J ( v (i 1) )
9
Fast Kurtosis Maximization Algorithm (FKMA)
(Chi et al., 2001 [4, 5])
Criterion [7]:
C4 {e[n]}
J ( v) J (e[n])
Maximization
2
2
E { e[n] }
magnutude of normalized
kutorsis of e[n ]
Optimum
v
e[n] v T x[n] k sk [n]
(noise-free case)
( k is an unknown complex scale
factor and k {1,, K } )
Algorithm:
e (i ) [n ]
At the i th iteration
Yes
Compute
x[n ]
e (i 1) [n ]
v
(i )
R 1d (i 1)
J ( v ( i ) ) J ( v ( i 1) )
R 1d (i 1)
?
No
Update v through a
gradient type optimization
algorithm such that
(i )
R R*X E{x*[n]xT [n]} (PxP matrix)
d(i1) cum{e(i1) [n], e(i1) [n], (e(i1) [n])* , x*[n]}
To the
(i 1) th
iteration
e (i ) [n ]
J ( v (i ) ) J ( v (i 1) )
9
Observations:
The FKMA itself is an exclusive spatial processing algorithm.
The smaller the value of ( si [n]), the worse the performance of the
FKMA for finite SNR and finite data length N .
By (A2)
J (si [n]) (si [n]) J (ui [n])
where
(absolute normalized kurtosis of si [n ])
J ( si [n])
(entropy measure of the stable
0 ( si [ n])
k
1
2
sequence bi [n ] )
J (ui [n])
2
bi [k ]
k
(equality holds only as bi [n ] [n ], i.e., minimum entropy of bi [n ] )
bi [k ]
4
(si [n]) can be thought of as a measure of distance of
si [n ] from a Gaussian
process, implying that the performance of the FKMA (which requires si [n ] to
be non-Gaussian [6]), depends on ( si [n]).
10
MSC Procedure
x[n ]
Each Stage of the Multistage Successive
Cancellation (MSC) Procedure
Estimate One Source
Signal Using FKMA
( ak : k th column of A )
sˆk [n ]
Obtain
aˆ k
x[n ]sˆk [n ]
| sˆk [n ] |2
E{
E{
}
}
âk
x[n ]
Update x[n ] by
x[n ] aˆ k sˆk [n ]
x[n ]
Next
Stage
NOTE
The estimated sources sˆk [n ] and âk columns of A obtained at later
stages in the MSC procedure may become less accurate due to error
propagation effects from stage to stage [6].
11
3. Turbo Source Extraction Algorithm
Source Separation Filter:
v T SEA [n ] vv [n ]
[ n] v
where
T
TSEA
(A bank of same temporal filters)
[n] x[n]
k
T
v TSEA
[k ]x[n k ]
v: P
1 vector for extracting a colored source signal sk [n ], i.e.,
removing spatial interference due to the mixing matrix A . (spatial filter)
v [n ]: single-input single-output (SISO) deconvolution (or higher-order whitening)
filter of order L to restore uk [n ] from sk [n ] . (temporal filter)
Design Criterion:
J ( [n])
Maximization
( [n]) J (uk [n])
J (e[n]) (e[n])
(Extracted Source)
[n] v TTSEA [n] x[n] v T y[n] e[n] v[n]
e[n] sˆk [n] v T x[ n] (spatial processing)
y[n] v[n] x[n] (temporal processing)
12
Turbo Source Extraction Algorithm (TSEA)
(Chi et al., 2003 [3])
Signal processing procedure at the i th cycle
Temporal Processing
Step 1
x[n ]
y
(i 1)
[n ]
x[n ] v[n ]
(a)
Spatial Processing
(b)
v [n ] vˆ ( i1) [n ]
FKMA(s)
[n ]
( vˆ (i ) )T y (i 1) [n ]
Step 1
vˆ ( i )
vˆ ( i ) [n ]
Step 2
(b)
[n ]
e[n ] vˆ [n ]
(i )
T e[n ]
FKMA(t)
(a)
e[n] v x[n]
T
v vˆ
Step 2
(i )
x[n ]
sˆk [n]
[v[0], v[1],
, v[ L]]T
e[n ] [e[n ], e[n 1], , e[n L ]]T (Extracted Source)
13
Why?
Performance of TSEA is superior to FKMA.
Interpretations:
1) Temporal
Processing:
[n ] e [n ] v (i ) [n ] sˆk [n ] v (i ) [n ]
( k uk [n ] bk [n ]) v (i ) [n ] k uk [n ] g k [n ]
(b)
Increasing J ( [n]) ( [n]) J (uk [n]) is equivalent to increasing
( [n])
m
m
2) Spatial
Processing:
(b)
g k [ m]
g k [ m]
4
2 2
( sk [n]), gk [n ] bk [n ] v (i ) [n ]
~ [n ]
y[n ] x[n ] v (i ) [n ] A~s [n ] w
~
s [n ] (s~1[n ], , s~k [n ] sk [n ] v (i ) [n ] [n ], , s~K [n ]) T
( [n]) (sk [n]) (sk [n]) and
(sl [n]), l k
14
Remarks:
TSEA is computationally efficient with super-exponential
convergence rate and P parameters for spatial processing
and L+1 parameters for temporal processing, respectively.
The performance gain of the TSEA reaches the maximum
as long as the order L (a parameter under our choice) of
the temporal filter is sufficiently large. On the other hand,
the asymptotic performance of FKMA approaches that of
the TSEA as N and SNR .
All the sources can be extracted through the MSC procedure.
The resultant BSS algorithm that uses the TSEA, is referred
to as MSC-TSEA, also outperforms the MSC-FKMA, at the
extra expense of the temporal processing at each stage.
15
4. Non-Cancellation Multistage Source Separation
Algorithms
NCMS-FKMA
Constrained
Criterion:
Unconstrained
Criterion:
Constraint
v arg max{J ( v) J (e[n]) : e[n] v T x[n], v T C 0T-1}
v
where C aˆ1 ,
aˆ2 ,
, aˆ -1 ,
2, 3,
,K
v (C ) v (unconstrained optimization problem)
arg max{J (v ) J (e[n]) : e[n] T x [n]}
v
C : P P
projection matrix
x [n] C x[ n]
Theorem 1: Let S be the set of all the extracted source signals up to
stage 1. With (A1), (A2), and the noise-free assumption, the
optimum e [n] T x [n] v T x[n] k sk [n] where k is an unknown nonzero constant and sk [n] S .
16
Signal Processing Procedure of NCMS-FKMA
(Initial Condition)
v ( 0) (1, 1, ,1) T / P
x[n ]
aˆ -1
C
Obtain
by
SVD of C and
x [n]
x [n ] C x[n ]
(F-a)
Estimate One
Source Signal
Using FKMA
v
sˆk [n ]
â
{
E{| sˆ
}
[n ] | }
E x[n ]sˆk [n ]
k
(F-b)
Estimate One
Source Signal
Using FKMA
Good Initial
Condition
Obtain
x[n ]
x[n ] v (0) v
e [ n]
â
2
17
Remarks:
The constrained source extraction filter v obtained in (F-a)
provides a suitable initial condition for the unconstrained source
extraction filter v in (F-b), which accordingly leads to one
distinct source estimate e [n] obtained at each stage neither
involving cancellation nor imposing any constraints on the source
extraction filter, as well as faster convergence than (F-a).
Therefore, unlike the MSC-FKMA, the NCMS-FKMA is free from
the error propagation effects at each stage.
As the MSC-TSEA performs better than the MSC-FKMA, the
NCMS-TSEA also performs better than the NCMS-FKMA at the
moderate expense of extra computational load for the temporal
processing of the TSEA.
18
Signal Processing Procedure of NCMS-TSEA
v (0) [n] v [n]
v (0) v x[n ]
(1, 1, ,1) T / P
(Initial Condition)
v ( 0)
x[n ]
aˆ -1
C
Obtain
by
SVD of C and
x [n]
x [n ] C x[n ]
(T-a)
Estimate One
Source Signal
Using TSEA
v [ n]
v
sˆk [n ] e [n ]
Good Initial
Condition
Obtain
x[n ]
â
{
E{| sˆ
E
}
[n ] | }
x[n ]sˆk [n ]
k
(T-b)
Estimate One
Source Signal
Using TSEA
e [ n]
â
2
19
5. Simulation Results --- Part 1
Parameters Used:
ui [n ] : zero-mean, independent binary sequence of {1} with equal probability
si [n ] : generated by filtering ui [n ] through the chosen FIR filters bi [n ]
w[n ] : real white Gaussian noise vector
SNR:
SNR
2
E { x[n ] w[n ] }
2
E { w[n ] }
50 independent runs
Output (extracted) signal to interference-plus-noise ratio
(Output SINR)
1
Output SINR
K
K
SINR
i 1
i
20
Part A: Performance of NCMS-FKMA and NCMS-TSEA
5
4 mixing matrix A (taken from Chang et al., 1998 [9]) (P=5, K=4)
0.2887
0.2380
0.3397 0.7494
A 0.6107
0.4959
0.2644
0.3558
0.5731 0.1983
bi [n ] exp(
n 1
)
10 i
0.7120
0.1157
0.2661
0.4216
0.4807
0.4914
0.2097
0.2504
0.6640
0.4593
n 0, 1, ..., 5
Four cases are considered as follows:
Case 1: Output SINR versus SNR for different data length N .
Case 2: Output SINR versus different data length N .
Case 3: Output SINR versus (si [n]) (or i ) for all i .
Case 4: (a) Output SINR versus L.
K
(b) (1 K ) J ( k [n ]) versus L.
k 1
21
i 0.5 (or
30
i (si [n]) 0.2368) for all i , L 5
NCMS-TSEA, N=1500
NCMS-TSEA, N=1000
NCMS-TSEA, N=500
OUTPUT SINR (dB)
25
NCMS-FKMA, N=1500
NCMS-FKMA, N=1000
NCMS-FKMA, N=500
20
15
10
5
0
5
10
15
20
25
30
SNR (dB)
Figure 1. Simulation results (Output SINR versus SNR) of Case 1.
22
i 1 (or
i (si [n]) 0.1856) for all i, L 5, and SNR=30 dB
32
30
OUTPUT SINR (dB)
28
26
24
22
20
18
NCMS-TSEA
NCMS-FKMA
16
14 3
10
10
4
N
10
5
10
6
Figure 2. Simulation results (Output SINR versus data length N ) of Case 2.
23
SNR=30 dB, N 2000, and L 5
32
0.745
0.345
0.240 0.183
0.144 0.115
0.091
0.068 0.01
30
OUTPUT SINR (dB)
28
26
24
22
20
18
16
14
NCMS-TSEA
NCMS-FKMA
12
10
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Figure 3. Simulation results (Output SINR versus ) of Case 3.
24
SNR=30 dB, N 2000
i 1 and 0.5 (i.e., (si [n]) 0.1856 and 0.2368 ) for all i
35
OUTPUT SINR (dB)
30
25
20
NCMS-TSEA, =0.5
(or =0.2368)
NCMS-TSEA, =1
(or =0.1856)
15
10
0
1
2
3
4
5
6
7
8
9
L
Figure 4a. Simulation results (Output SINR versus the order of
the temporal filter L ) of Case 4 (a).
25
SNR=30 dB, N 2000
i 1 and 0.5 (i.e., (si [n]) 0.1856 and 0.2368 ) for all i
2
1.8
[n])
1.6
k =1
(1/K)
K
J(
k
1.4
1.2
1
0.8
NCMS-TSEA, =0.5
(or =0.2368)
NCMS-TSEA, =1
(or =0.1856)
0.6
0.4
0
1
2
3
4
5
6
7
8
9
L
Figure 4b. Simulation results (Output SINR versus the order of
the temporal filter L ) of Case 4 (b).
26
Part B: Performance Comparison
The same 5 4 mixing matrix A in Part A and
bi [n ] exp (
n 1
), n 0, 1, ..., 5 (1 1 , 2 0.4, 3 0.3, 4 0.2)
10 i
(s1[n ]) 0.1856 , (s2 [n ]) 0.2706, (s3[n ]) 0.3335, (s4 [n ]) 0.4644
Data length N = 2000 and L = 5
Comparison with the MSC-FKMA, AMUSE ( 1) (Tong et al. 1990 [1]) and
SOBI algorithm ( i
i , i 1, 2,3)
(Belouchrani et al. 1997 [2])
Three cases are considered as follows:
Case A: Output SINR1 versus SNR for N 2000 and L 5.
Case B: Output SINR versus SNR for N 2000 and L 5.
Case C: Output SINR versus N for SNR = 20 dB and L 5.
27
35
NCMS-TSEA
MSC-TSEA
NCMS-FKMA
MSC-FKMA
FastICA
SOBI ALGORITHM
AMUSE
OUTPUT SINR1 (dB)
30
25
20
15
10
5
0
5
10
15
20
25
30
SNR (dB)
Figure 5. Simulation results (Output SINR1 versus SNR) of Case A.
28
NCMS-TSEA
MSC-TSEA
NCMS-FKMA
MSC-FKMA
FastICA
SOBI ALGORITHM
AMUSE
30
OUTPUT SINR (dB)
25
20
15
10
5
0
5
10
15
20
25
30
SNR (dB)
Figure 6. Simulation results (Output SINR versus SNR) of Case B.
29
22
20
OUTPUT SINR (dB)
18
16
14
12
NCMS-TSEA
MSC-TSEA
NCMS-FKMA
MSC-FKMA
FastICA
SOBI ALGORITHM
AMUSE
10
8
6
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
N
Figure 7. Simulation results (Output SINR versus data length N ) of Case C.
30
Case D: Output SINR versus
a 3x2 mixing matrix by removing the last two rows and columns of
the mixing matrix in Part A. (P=3, K=2)
A :
0.2887
0.2380
A 0.3397 0.7494
0.6107
0.4959
B1 (z) (1 0.5z 1 )(1 0.8z 1 )(1 4z 1 )
B2 (z) [1 (0.5 )z 1 ][1 (0.8 )z 1 ][1 (4 )z 1 ] 0.05 0.40
Data length N =1000, SNR=30 dB and L =3.
Comparison with the MSC-FKMA, AMUSE ( 1) and SOBI algorithm
( i i , i 1, 2, 3)
31
30
OUTPUT SINR (dB)
25
20
15
NCMS-TSEA
MSC-TSEA
NCMS-FKMA
MSC-FKMA
FastICA
SOBI ALGORITHM
AMUSE
10
5
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Figure 8. Simulation results (Output SINR versus ) of Case D.
32
6. Turbo Space-time Receiver for
CCI/ISI Reduction
Problem Statement: CCI and ISI Suppression in TDMA
Cellular Wireless Communications
w[n ] (Noise)
f5
f6
x [n ]
h [n ] (Multipath channel)
f4
f1
f7
u [n ]
f3
f2
CCI: Co-channel Interference
ISI: Intersymbol Interference (due to multipath)
GOAL
f1
CCI
Enhance data rate, link quality, capacity, and coverage.
Space-time processing using an antenna array has been used for
combating CCI and ISI in the receiver design [15-16].
33
Signal Model:
Consider the scenario where the base station is equipped with multiple
antennas, and the signal of interest and CCI are received from multiple
distinct directions of arrival (DOA), with a frequency-selective fading
channel for each DOA. (a general scenario)
(h11[n ], 11 )
u1[n ]
(h12 [n ],12 )
(h 21[n ], 21 )
u 2 [n ]
x1[n ]
x 2 [n ]
x P [n ]
(h 22[n ], 22 )
34
The received signal from the desired user and K 1 CCIs (users) can be
expressed as an instantaneous mixture of multiple sources
x[n ] As[n ] w[n ]
where
{A}P ( A1 , A2 ,..., AK )
K
Ak sk [n] w[n]
k 1
{A k }P pk (a( k1 ), a( k 2 ),..., a( kpk ))
s[n ] (s1T [n ], s T2 [n ],..., s TK [n ]) T sk [n ] (sk1[n ], sk 2 [n ],..., skpk [n ])T
skj [n ] hkj [n ] uk [n ] j 1, 2,..., pk
( pk is no. of DOAs of user k)
“ISI-distorted’’ signal (colored signal) from jth DOA of user k
a(kj ) : steering vector of jth DOA of user k
hkj [n ] : Lkj th-order channel impulse response of jth DOA of user k
K
pk (total no. of DOAs or ’’sources”)
k 1
35
Assumptions:
(A1) The unknown P
DOA matrix A is of full column rank and P
(A2) The data sequence u1[n ] of user 1 (the desired signal) is i. i. d. zero-mean
non-Gaussian with C4{u1[n ]} 0, and meanwhile statistically independent
of the other ( K 1 ) zero-mean i. i. d. data sequences uk [n ] (of CCI).
(A3) w[n ] is zero-mean Gaussian, and statistically independent of uk [n ] for all k .
K
pk (total no. of DOAs or ’’sources”)
k 1
sk [n ] (sk1[n ], sk 2 [n ],..., skpk [n ])T
pk correlated colored non-Gaussian sources
s[n ] (s1T [n ], s T2 [n ],..., s TK [n ]) T
K block mutually independent colored sources
36
Case I: Each user has a single DOA with multiple paths
(Venkataraman et al., 2003), i.e.,
pk 1,
A (a(11 ),
k 1, 2,..., K
, a(K 1 ))
sk1[n ] uk [n ] hk1[n ]
s[n ] (s11[n ],
, sK 1[n ]) T
Mutually independent
colored sources
Case II: Each user has multiple DOAs with disjoint domains of
support of multipath channel impulse responses, i.e.,
hki [n ] hkj [n ] 0 i j and 1 k K
2
E ski [n ]skj* [n ] hki [l ]hkj* [l ] E uk [n ] 0, i j
l
s[n ] (s1T [n ], s T2 [n ],..., s TK [n ]) T
K block mutually independent colored sources;
mutually independent random variables for each n
37
Conventional Cascade Space-Time Receiver (CSTR)
(Jelitto and Fettweis, 2002)
For Cases I and II, the conventional CSTR has been reported for
CCI and ISI suppression
Space-time Processor
x[n ]
Spatial Filter
v
e[n ] s1 j [n ] Temporal Filter
[n ] u1[n ]
v [n ]
In CAMSAP-06, we proposed two space-time receivers based on
kurtosis maximization for these two cases and a discussion of the
proposed space-time receivers for the general scenario.
Other existing structures: full-dimension (joint) ST processing,
reduced dimension ST processing (prewhitening followed by joint ST
processing).
38
Kurtosis Maximization (Ding ad Nguyen, 2000):
J ( v) J (e[n ])
C 4{e[n ]}
2
E { e[n ] }
2
magnutude of normalized
kutorsis of e[n ]
Maximization
(noise-free case)
e[n ] v T x[n ] kj skj [n ]
kj uk [n ] hkj [n ]
kj is an unknown complex scale factor
k {1, 2, , K } and j {1, 2,
, pk }
Closed-form solution for v : Not existent
Gradient-type iterative algorithms for finding a local optimum v :
Not very computationally efficient
Applicable not only for Case I but also for Case II (Peng et al.,
ICICS 2005)
39
Fast Kurtosis Maximization Algorithm (FKMA)
2001):
(Chi and Chen,
e (i ) [n ]
At the i th iteration
Yes
Compute
x[n ]
e (i 1) [n ]
v
(i )
J ( v (i ) ) J ( v (i 1) )
R 1d (i 1)
?
R 1d (i 1)
No
Update v through a
gradient type optimization
algorithm such that
(i )
To the
(i 1) th
iteration
e (i ) [n ]
J ( v (i ) ) J ( v (i 1) )
R E{x* [n ]xT [n ]}
(PxP matrix)
d ( i 1) cum {e ( i 1) [n ], e ( i 1) [n ], (e ( i 1) [n ]) * , x * [n ]}
40
Blind CSTR Using FKMA
Space-time Processor
Spatial Filter
x[n ]
e[n ] s1 j [n ]
v
Temporal Filter
v [n ]
[n ] u1[n ]
Spatial processing using FKMA for CCI suppression
e[n ] vT x[n ] s1 j [n ] u1[n ] h1 j [n ]
With a suitable initial condition for v , FKMA will converge at a
super-exponential rate with e[n ] s1 j [n ], j 1, 2, , p1 for high SNR.
Temporal processing using FKMA for ISI removal
[n ] v[n ] e[n ] v Te[n ] u1[n ]
where
v [v[0], v[1], , v[L ]]T
(L: order of the temporal filter)
e[n ] [e[n ], e[n 1], , e[n L ]]T
41
Usually, the ISI-distorted (desired) signal s1 j [n ], has higher power than all
2
the CCI, (i.e., E { s1 j [n ] }
E { sk i [n ] } k 1). So
2
2
1 j arg max E{ a ( )x[n ] },
H
(DOA estimate by delay-and-sum)
implying that a (1 j ) can be used as the initial condition for the spatial
filter v needed by the FKMA.
It can be easily shown that (Chi et al., 2003)
J (s1 j [n ]) (s1 j [n ]) J (u1[n ]),
where
L1 j
0 (s1 j [n ])
m 0
h1 j [m ]
4
2
h1 j [m ]
m 0
L1 j
2
1
The performance of the spatial filter v (to suppress CCI) using FKMA is worse
for smaller (s1 j [n ]), and worse for larger L1 j , leading to limited performance of
the temporal filter v [n ] of the blind CSTR.
42
Blind Turbo Space-Time Receiver (TSTR)
Space-Time Filter for Source Extraction
(Chi et al. 2003, 2006):
vTSTR [n ] vv[n ]
[n ] v TTSTR [n ] x[n ]
Design Criterion:
Optimum (noise-free case)
Maximization
J ( [n ])
[n ] vTTSTR [n ] x[n ] vT y[n ]
s1 j [n ] v[n ] 1u1[n 1 j ]
e[n ] s1 j [n ] vT x[n ] 1 j s1 j [n ]
y[n ] x[n ] v [n ]
43
Proposed Blind TSTR Using FKMA
Signal processing procedure at the i th cycle:
CSTR
x[n ]
e (i ) [n ] v T x[n ]
Spatial Filter
(i )
vv
J ( 1(i ) [n ])
s1 j [n ]
(S2)
Temporal Filter
2(i ) [n ] u1[n ]
FKMA
(i )
v (i ) [n ] J ( 2 [n ])
v (i )
i=i+1
1(i ) [n ]
Spatial Filter
FKMA
(S1)
y ( i ) [n ]
x[n ] v [n ]
Temporal Filter
x[n ]
v [n ] v (i 1) [n ]
CSTR
44
Why?
Performance of blind TSTR is superior to blind CSTR.
Interpretations:
1) Temporal
Processing:
2(i ) [n ] e (i ) [n ] v (i ) [n ] s1 j [n ] v (i ) [n ]
(1u1[n ] h1 j [n ]) v (i ) [n ] 1u1[n ] g1 j [n ]
g1 j [n ]
Increasing J ( 2(i ) [n ]) is equivalent to increasing
(
(i )
2
[n ])
m
m
2) Spatial
Processing:
g1 j [m ]
g1 j [m ]
4
2 2
(s1 j [n ])
~ [n ]
y[n ] x[n ] v (i 1) [n ] As[n ] w
[n ],..., s1j [n ] s1 j [n ] v (i 1) [n ] 2(i ) [n ],
s[n ] ( s11
( 2(i ) [n ]) (s1j [n ]) (s1 j [n ]) and
, sk 1[n ],
K [n ] )T
, sKp
(skl [n ]), k 1, l j
45
Remarks:
It can be proven that
J ( 2(i 1) [n ]) J (1(i ) [n ]) J ( 2(i ) [n ]) J (u1[n ])
for all i , implying the guaranteed convergence of the proposed blind
TSTR. Typically, the number of cycles spent by the TSTR before
convergence, is equal to 2 or 3. The computational load of the blind
TSTR is approximately 2 or 3 times that of the blind CSTR.
Because the design of v and that of v [n ] are coupled in a constructive
and boosting manner, the proposed blind TSTR outperforms the blind
CSTR for all L, and meanwhile their performance difference is larger
for larger L .
Compared with the blind CSTR, the proposed blind TSTR is insensitive to
the value of (s1 j [n ]) (i.e., robust against channel h1 j [n ] with multiple
paths or severe ISI).
46
Performance of the blind TSTR:
[n ] v TTSTR [n ] x[n ], where v TSTR [n ] vv[n ]
CASE I: CCI suppression by v
[n ] v T a(1 )h11[n ] v[n ] u1[n ] residual CCI and noise
1u1[n 1 ]
Multiple DOAs suppressed also by v
CASE II: CCI suppression by v
p1 T
[n ] v a(1i )h1i [n ] v [n ] u1[n ] v a(1 j )h1 j [n ] v [n ]* u1[n ]
j i
residual CCI and noise
1i u1[n 1i ]
T
GENERAL CASE: CCI suppression by v
p1 T
[n ] v a(1 j )h1 j [n ] v [n ] u1[n ] residual CCI and noise
j 1
the spatial filter v and the temporal filter v [n ] combine the signals
from all the DOAs in a constructive and boosting fashion
g1u1[n 1 ]
47
7. Simulation Results --- Part 2
Scenario of Case I
ui [n ] : zero-mean, independent binary sequence of {1} with equal probability
H1 (z) 0.6178 0.4325z 1 0.3707z 2 0.3089z 3 0.2471z 4 0.3707z 5
H 2 (z) 0.4056 0.2839z 1 0.3650z 2 0.2839z 3 0.1217z 4 0.1622z 5
H 3 (z) 0.3984 0.3187z 1 0.3586z 2 0.2390z 3 0.1992z 4 0.1195z 5
1 0 , 2 40, 3 60
P 10 (array size)
R S [0] E{s[n ](s[n ]) H } : Diagonal matrix
w[n ] : white Gaussian noise vector
SNR:
E{ a(1 )s1[n ] }
2
SNR
2
E{ w[n ] }
2
E{ s1[n ] }
w2
1
w2
50 independent runs
48
3 60
1 0
2 40
Order of the temporal filter L =20
SNR=20 dB
Data length N =2000
49
Blind CSTR
Proposed Blind TSTR
50
Data length
N =2000
Order of the temporal filter L =20
51
SNR=30 dB
Order of the temporal filter
L =20
52
SNR=30 dB
Data length
N =2000
53
Scenario of Case II
ui [n ] : zero-mean, independent binary sequence of {1} with equal probability
H11 (z) 0.5199 0.3639z 1 0.3119z 2
H 21 (z) 0.3562 0.3206z 1 0.1425z 2
H12 (z) 0.5754z 3 0.2466z 4 0.3288z 5 H 22 (z) 0.3776z 3 0.2098z 4 0.2518z 5
11 0 12 20 21 40 22 60
P 10 (array size)
w[n ] : real white Gaussian noise vector
SNR:
SNR
R S [0] E{s[n ](s[n ]) H }: Diagonal matrix
E { a(11 )s11[n ] a(12 )s12 [n ] }
2
2
E { w[n ] }
50 independent runs
54
22 60
12 20
11 0
21 40
Order of the temporal filter L =20
SNR=20 dB
Data length N =2000
55
Blind CSTR
Proposed Blind TSTR
56
Data length N =2000
Order of the temporal filter L =20
57
SNR=30 dB
Order of the temporal filter
L =20
58
SNR=30 dB
Data length N =2000
59
Scenario of the general case
ui [n ] : zero-mean, independent binary sequence of {1} with equal probability
H 21 (z) 1 0.9z 1 0.4z 2 0.3z 3
H11 (z) 1 0.7z 1 0.6z 2 0.5z 3
2
3
4
5
H12 (z) 0.8z 2 0.7z 3 0.3z 4 0.4z 5 H 22 (z) z 0.9z 0.5z 0.6z
11 0 12 20 21 40 22 60
(array size)
R S [0] E{s[n ](s[n ]) H }: Block-diagonal matrix
P 10
w[n ] : white Gaussian noise vector
SNR:
E{ a(11 )s11[n ] a(12 )s12[n ] }
2
SNR
2
E{ w[n ] }
50 independent runs
60
22 60
12 20
11 0
21 40
Order of the temporal filter L =20
SNR=30 dB
Data length N =2000
61
Blind CSTR
Proposed Blind TSTR
62
Data length N =2000
Order of the temporal filter L =20
63
SNR=30 dB
Order of the temporal filter L =20
64
SNR=30 dB
Data length N =2000
65
8. Conclusions
FKMA only involves spatial processing for extraction of one
non-Gaussian (i.i.d. or colored) source from source mixtures.
It performs well with super-exponential convergence rate, but its
performance depends on the parameter 0 (si [n]) 1 .
We have introduced a novel blind source extraction algorithm,
TSEA, which operates cyclically using the FKMA for both of the
temporal processing and spatial processing. The proposed TSEA
outperforms the FKMA for (si [n]) 1 in addition to sharing
convergence speed and computational efficiency of the later at
each cycle.
Because of performance degradation resultant from the error
propagation in the MSC procedure, we further introduced two
non-cancellation BSS algorithms, namely, NCMS-FKMA and
NCMS-TSEA, that can extract a distinct source at each stage
without error propagation.
66
The two BSS algorithms, NCMS-FKMA and NCMS-TSEA perform
better than the existing MSC-FKMA and the MSC-TSEA,
respectively, with moderately higher computational complexities.
FKMA and TSEA are under investigation for CCI and ISI in MIMO
wireless communications (e.g., OFDM and multi-rate CDMA) and
other applications such as 2-D MIMO systems in biomedical signal
processing (with certain constraints or partial correlation between
source signals).
Some works of Part 1/Part 2 will be published in
C.-Y. Chi and C.-H. Peng, “Turbo source extraction algorithm and noncancellation source separation algorithms by kurtosis maximization,” IEEE
Trans. Signal Processing, vol. 54, no. 8, pp. 2929-2942, Aug. 2006.
C.-H. Peng, C.-Y. Chi and C.-W. Chang, “Blind multiuser detection by kurtosis
maximization for asynchronous multi-rate DS/CDMA systems,” EURASIP
Journal on Applied Signal Processing, vol. 2006, Article ID 84930, 17 pages,
2006. doi:10.1155/ASP/2006/84930. (special issue: Multisenor Processing for
Signal Extraction and Applications)
67
Background materials of the talk can be found in the following
book:
C.-Y. Chi, C.-C.Feng, C.-H. Chen and C.-Y. Chen, Blind Equalization
and System Identification, London: Springer-Verlag, 2006.
Thank you very much
68
References
[1] L. Tong, R.-W. Liu, V. C. Soon, and Y.-F. Huang, ``Indeterminacy and identifiability of blind
identification,'' IEEE Trans. Circuits and Systems, vol. 38, pp. 499-509, May 1991.
[2] A. Belouchrani, K. Abed-Meraim, J. -F. Cardoso, and E. Moulines, ``A blind source separation
technique using second-order statistics,'' IEEE Trans. Signal Processing, vol. 45, pp. 434-444,
Feb. 1997.
[3] C.–Y. Chi, C.-J. Chen, F.-Y. Wang, C.-Y. Chen and C.-H. Peng, ``Turbo source separation algorithm
using HOS based inverse filter criteria,'' Proc. IEEE International Symposium on Signal
Processing and Information Technology, Darmstadt, Germany, Dec. 14-17, 2003.
[4] C.–Y. Chi and C.-Y. Chen, ``Blind beamforming and maximum ratio combining by kurtosis
maximization for source separation in multipath,'' Proc. IEEE Workshop on Signal Processing
Advances in Wireless Communications, Taoyuan, Taiwan, Mar. 20-23, 2001, pp. 243-246.
[5] C.-Y. Chi and C.-Y. Chen , C.-H. Chen and C.-C. Feng, ``Batch processing algorithms for blind
equalization using higher-order statistics,'' IEEE Signal Processing Magazine, vol. 20, pp. 25-49,
Jan. 2003.
69
[6] J. M. Mendel, ``Tutorial on higher-order statistic (spectra) in signal processing and system
theory: theoretical results and some applications,'' Proc. IEEE, vol. 79, pp. 278-305, Mar. 1991.
[7] C.-Y. Chi, C.-H. Chen and C.-Y. Chen, ``Blind MAI and ISI suppression for DS/CDMA systems
using HOS-based inverse filter criteria,'' IEEE Trans. Signal Processing, vol. 50, pp. 1368-1381,
June 2002.
[8] Z. Ding and T. Nguyen, ``Stationary points of a kurtosis maximization algorithm for blind signal
separation and antenna beamforming,'' IEEE Trans. Signal Processing, vol. 48, pp. 1587-1596,
June 2000.
[9] C. Chang, Z. Ding, S. F. Yau, and F. H. Y. Chan, ``A matrix-pencil approach to blind separation
of non-white sources in white noise,'' Proc. IEEE International Conference on Acoustics,
Speech, and Signal Processing, Seattle, WA, May 12-15, 1998, pp. 2485-2488.
[10] D. J. Moelker, A. Shah and Y. Bar-Ness, ``The generalized maximum SINR array processor
for personal communication systems in a multipath environment,'' Proc. IEEE International
Symposium on Personal, Indoor, and Mobil Radio Communications, vol. 2, Taipei, Oct. 15-18,
1996, pp. 531-534.
[11] V. Venkataraman, R. E. Cagley and J. J. Shynk, ``Adaptive beamforming for interference
rejection in an OFDM system,'' IEEE Conference Record of the Thirty-Seventh Asilomar
Conference on SSC, Nov. 9-12, 2003, vol. 1, pp. 507-511.
70
[12] J.-F. Cardoso, ``Source separation using higher order moments,'' Proc. IEEE International
Conference on Acoustics, Speech, and Signal Processing, Glasgow, UK, May 23-26, 1989, pp.
2109-2112.
[13] A. Hyvärinen, J. Karhunen, and E. Oja, Independent Component Analysis. New York: WileyInterscience, 2001.
[14] A. Hyvärinen and E. Oja, ``A fixed-point algorithm for independent component analysis,''
Neural Computation, vol. 9, pp. 1482-1492, 1997.
[15] J. Jelitto and G. Fettweis, ``Reduced dimension space-time processing for multi-antenna
wireless systems,'' IEEE Wireless Communications Mag., vol. 9, pp. 18-25, Dec. 2002.
[16] Jen-Wei Liang and A. J. Paulraj, ``Two stage CCI/ISI reduction with space-time processing in
TDMA cellular networks,'' Proc. 30th Asilomar Conference on Signals, Systems, and Computers,
vol. 1, Pacific Grove, CA, Nov. 3-6, 1996, pp. 607-611.
71