Professional Documents
Culture Documents
(1)
where r(t) is the received signal, K is the total number of users. A
k
, b
k
(t), and a
k
(t) are the
amplitude, bit sequence, and spreading chip sequence of user k respectively. The bit rate
R
b
is the inverse of the bit period T and the chip rate R
c
is the inverse of the chip period
T
c
. The spreading gain is
c
T T N / .
k
and
k
are the time delay and phase estimates of
user k, which are assumed to be accurate.
Furthermore, the bits and chips are assumed to be rectangular with i.i.d random values of
1 t that occur with probability 0.5. The
k
and
k
are assumed to be i.i.d. uniform
random variables in [0, T] and [0, 2].
SIC operates simply by subtracting off the contributions of MAI with the order of
subtractions given by the relative strength of the users. The Strongest signal is cancelled
out first followed by the second strongest, etc. This could be done by ranking the signal
strengths, which is obtained by separate channel estimates or directly from the
conventional detector. The successive cancellations are carried as follows [9]:
i) Recognize the strongest signal (one with maximum correlation value).
ii) Decode the strongest user.
iii) Estimate the amplitude of the decoded user from the output of the correlator.
iv) Regenerate the strongest users signal using its chip sequence and the estimate
of its amplitude.
v) Cancel the strongest user.
vi) Repeat (until all users are decoded or a permissible number of cancellations
are achieved).
Figure 3, below, shows the block diagram for the SIC receiver. Each users signal is
detected by using the conventional matched filter. Then the strongest user is determined
by using a selector. At the output of the low-pass filter (LPF) of the I-Channel, we get:
2
) (
2
) cos(
) ( ) (
)} cos( ) ( { ) (
1
t n
t b t a A
t w t r LPF t d
c
K
k
k
k k k k k
c
I
+
(2)
where n
c
(t) is the in-phase component of the low-pass filtered Gaussian noise n(t). d
Q
(t) is
obtained in a similar way. Also, the cross correlation term between user k and user i is
given by:
) ( ) cos( ) ( ) (
1
) , (
0
, , , , k k i k
T
i i k k i k i k i k
t b dt t a t a
T
I
]
]
]
(3)
7
Figure 3: SIC Receiver Block Diagram [13]
Hence we can express the (j+1)
st
decision variable in the I-Channel as:
+
+ + + + + + +
+ + + +
+ +
K
j k
j k j k j k k j j j
I
j
T
j j j
I I
j
I A b A n
dt t a t d Z
2
1 1 , 1 , 1
2
1 1 1
0
1 1 1 1
) cos( ) cos( ) (
2
1
) ( cos
2
1
2
1
) cos( ) ( ) (
(4)
where,
+ + + +
T
j j j c
I
j
dt t a t n
T
n
0
1 1 1 1
) cos( ) ( ) (
1
. Similarly, we
Q
j
Z
1
+
is obtained as:
+
+ + +
+ + + + +
+
+
K
j k
j k j k j k k
j j j
Q
j
Q
j
I A
b A n Z
2
1 1 , 1 ,
1
2
1 1 1 1
) sin( ) sin( ) (
2
1
) ( sin
2
1
2
1
(5)
After j cancellations, the decision variable for the (j+1)
st
user is given by:
1 1 1 1 1 1
2
1
2
1
+ + + + + +
+ +
j j j
Q
j
I
j j
C b A Z Z Z (6)
and C
j+1
is given by:
+ + + + +
+
+ + + +
+ +
j
i
i i i i i i i
Q
j
I
j
K
j k
j k j k j k k j
I C n n I A C
1
1 , 1 , 1 , 1 1
2
1 , 1 , 1 , 1
) , ( ) ( ) , ( (7)
In the above expression, the first term is the multiple access interference of the
uncancelled users, the second term is due to the Gaussian noise, and the third term is due
to the cumulative noise from imperfect cancellation. The variance of C
j+1
conditioned on
the ordered set of amplitudes A
k
is defined as follows:
[ ]
[ ]
( ) [ ]
+ + + + +
+
+ + +
+ +
+ +
j
i
i i i i i i i
Q
j
I
j
K
j k
j k j k j k k
k j j
I Var n n Var
I Var A
A C Var
1
1 , 1 , 1 , 1 1
2
1 , 1 , 1 ,
2
1 1
)] , ( [
) , (
|
(8)
8
Using results given in [] for asynchronous systems, the above expression yields:
+ +
+
+
j
i
i
K
j k
k j
N T
N
A
N
3
1
3
1
2
0 2
1
(9)
and the SNR conditioned on A
k
is then given by:
+ +
+
+
+
j
i
i
K
j k
k
j
j
N T
N
A
N
A
3
1
3
1
2
0 2
2
1
1
(10)
If we assume that noise C
j+1
is Gaussian with zero mean and variance
j+1
, the bit error
probability (BER) after the j
th
cancellation, conditioned on the amplitudes A
k
, is the given
by:
{ {
( )
1
1
1
1 1 1 1 1
1
| 0
+
+
+
+ + + +
+
,
`
.
|
< <
j
j
j
j j j j
j
e
Q
A
Q
A C P b Z P P
(11)
Since equations (9), (10), and (11), were conditioned on A
k
, which are the ordered set of
amplitudes of K users, the analysis of the BER performance of the SIC scheme were done
using order statistics as in []. The pdfs of the ordered A
k
is denoted by ) (x f
k
A
and is
obtained as follows:
( ) ( )
[ ] ) ( ) ( 1 ) (
! 1 !
!
) (
1
x f x F x F
k k K
K
x f
k k K
A
k
(12)
where,
2
2 ) (
x
xe x f
and
2
1 ) (
x
e x F
are the pdf and cdf of the Rayleigh fading
amplitudes A
k
, respectively.
The BER expression after the j
th
cancellation is then unconditioned using the pdf of the
(j+1)
st
strongest amplitude as follows:
[ ]
+
+ +
+
,
`
.
|
0 1
1 1
) (
1
dx x f
E
A
Q P
j
k
A
j A
j j
e
(13)
The average probability of error is then obtained as the average of the BER resulting
from all stages of cancellation.
PIC SCHEME
In this section we will analyze the second interference cancellation scheme known as
parallel interference cancellation (PIC). Unlike SIC, where interfering users are decoded
and cancelled successively, for a desired user, all the other interfering users are
collectively cancelled from the received signal. Figure 4 shows a block diagram of the
PIC receiver.
9
Figure 4: PIC Receiver Block Diagram [13]
As in the case of SIC, PIC uses the matched filter detector to detect all of the signals. The
decision variables Z
i
, as seen in figure 4, are the decision variables used for decoding by
the conventional receiver. These decision variables are then used to regenerate the user
signals and cancel it from the received signal to isolate the user of interest. The modified
received signals are once again fed through the matched filter of the user of interest and
another set of decision variables
i
Z
(14)
The Z
i
represents the decision variable for the i
th
user at the output of the conventional
detector and is given by:
10
i
K
i k
k
i k k i i
T
j
i i i
N I A b A
dt e t a t r
T
Z
i
i
i
2
1
2
1
2
1
) ( ) (
1
Re
1
,
+ +
'
'
(15)
These decision variables are then used to regenerate the user signals, which are cancelled
from the received signal to form a modified received signal. The modified received signal
becomes:
+
K
i m
m
j
m m i
K
i m
m
K
m k
k
j
m m m k k
z
j
i i i i i
K
i m
m
j
m m m i
mk
mk
i
mk
e t a N
e t a I A
t n e t b t a A
e t a Z t r t r
1
1 1
,
1
) 1 (
) (
2
1
) (
2
1
) (
2
1
) ( ) (
2
1
) ( ) ( ) (
(16)
The decision variable for the first stage for the i
th
user now becomes:
'
'
K
i m
m
m i i
K
i m
m
K
m k
k
m i m k k i i
T
j
i i i i
I N I I A b A
dt e t a t r
T
Z
i
i
i
1
,
1 1
, ,
) 1 (
2
1
2
1
2
1
) ( ) (
1
Re
(17)
This completes the first stage cancellation. To cascade one more stage of cancellation, the
new decision variables obtained above are used in the same manner as before to
regenerate a more accurate version of the user signals, which are then cancelled from the
received signal. This process can be repeated for s stages to obtain better results. Further
analysis of the decision variables become very difficult and very complicated to follow.
Hence, I will use the closed form of the average BER performance that was derived by A.
Kaul and B.D. Woerner in []. The BER can be expressed as:
( )
( ) ( )
( )
'
'
]
]
]
]
]
]
]
,
`
.
|
+
,
`
.
|
+
,
`
.
|
,
`
.
|
,
`
.
|
,
`
.
|
2 / 1
1
0
) (
1
1 1
3
1
3
1
1
3
1
1
2
1
s
k
K
k
k s s
s
s
b
s
b
P
P
K
K
N
N
K
N
K
N
E
Q P
k
k
(18)
11
COMPARISON RESULTS
BER/Capacity Performance Comparison
Using the average BER probabilities derived in the last two sections, I was able to do
matlab simulations to compare the two schemes. Figure 6 below shows the average BER
as a function of the number of active users in the system under ideal power control, i.e.
K
P P P ...
1 1
. In this case it was clear that PIC outperformed SIC especially in the
case of having two stages of parallel cancellations, while both SIC and PIC proved
superior to the conventional matched filter detector.
For the fading channel (Rayleigh fading), however, SIC had a better performance than
both single-stage and 2-stage PIC, as shown in figure 7. This results from the prominent
power variability among all users over the fading channel as opposed to the situation of
perfect power control. Therefore, selecting and decoding the user signals in descending
order, enables SIC to use the differing strengths of the signals to its advantage. Hence,
from the analytical results and simulation results we see that SIC has an overall
advantage with respect to BER performance over PIC because in wireless applications,
fading channels are more realistic than the ideal power control AWGN channel.
Figure 6: BER vs. Number of Users for AWGN Channel
(E
b
/N
0
= 8 and N = 31)
12
Complexity Comparison
With respect to complexity, SIC appears to be simpler, and requires less hardware than
PIC. It is worth noting that more stages of parallel cancellation improves the BER
performance where it almost achieves the same results as SIC (maybe better with more
stages), as seen in figure 7, but at the expense of exxessive hardware complexity and
processing delay.
Figure 7: BER vs. Number of Users for Fading Channel
(avg E
b
/N
0
= 8 and N = 31)
Processing Delay
Processing delay presents the biggest drawback to SIC, since only a single user bit is
decoded at each stage and thus, it takes at least K bit-times to decode all users for each
bit. In the case of PIC, it takes s bit-times, where, s is the number of stages, to decode all
users for each bit. This is a clear advantage for PIC since s is usually a lot less than K.
CONCLUSION
From the comparison results, it is clear that both successive and parallel interference
cancellation schemes have better performance than the conventional detector, and at the
same time, they are simpler and less complex than the optimum detector. While SIC
appears to be more resistant to fading than PIC, and achieves better results with regard to
BER and capacity performance, it suffers mightily from a high processing delay. This
proves to be major in applications that are less tolerant to high delays, such as cellular
13
systems. One way to deal with stringent delay requirement is to limit the number of
cancellations. A groupwise SIC (GSIC) has been proposed in [] to deal with delay
constraints. However, it is obvious that the performance will be degraded. Due to the
lack of time, I was not able to investigate GSIC and compare it to PIC and SIC. This
could be left for some future work.
REFERENCES
[1] J. G. Proakis, Digital Communications, 2 ed, McGraw-Hill, New York, 1989.
[2] K.S. Gilhousen, I. M. Jacobs, R. Padovani, A. J. Viterbi, L.A. Weaver, and C.E>
Wheatly III, On the Capacity of a cellular CDMA System, IEEE Transanctions
on Vehicular Technology, vol. 40, pp. 303-311, May 1991.
[3] J. Andrews, "Successive Interference Cancellation for Uplink CDMA," Ph.D.
Dissertation, Stanford University, 2002.
[4] R. Lupas, and S. Verdu, Near-far Resistance of Multi-user Detectors in
Asynchronous Channels, IEEE J. Select Areas Commun., vol. 38, no. 4, pp. 496-
508, 1990.
[5] A. Klien and B.W. Baier, Linear Unbiased Data Estimation in Mobile Radio
Systems Applying to CDMA, IEEE J. Select Areas Commun., vol. 11, no. 3-4,
pp. 1058-1066, 1999.
[6] Z. Zvonar, Multi-user Detection in Asynchronous CDMA Frequency Selective
Fading Channels, Wireless Personal Communications, Kluwer, vol. 3, no. 3-4,
pp. 373-392, 1996.
[7] Z. Xie, et al., A Family of Sub-optimum Detectors for Coherent Multi-user
Communications, IEEE J. Select Areas Commun., vol. 8, no. 4, pp. 683-690,
1990.
[8] A. Klien, et al., Zero Forcing and Minimum Mean-Square-Error Equalization for
Multi-user Detection in Code Division Multiple Access Channels, IEEE Tran.
Vehic. Tech., vol. 45, no. 2, pp. 276-287, 1996.
[9] J. M. Holtzman, Successive interference cancellation for direct sequence code
division multiple access, Military Communications Conference, vol. 3, pp. 997-
1001, 2-5 Oct. 1994.
[10] A. J. Viterbi, Very low rate convolutional codes for maximum theoretical
performance of spread-spectrum multiple-access channels, IEEE jour. On Sel.
Areas in Comm., vol. 8, pp. 641-9, May 1990.
14
[11] S. Verdu, Minimum probability of error for asynchronous Gaussian multiple-
access channels, IEEE Trans. Information Theory, vol. IT-32, pp. 85-96, Jan
1986.
[12] P. Patel and J. Holtzman, Analysis of a simple successive cancellation scheme in
a DS/CDMA system, IEEE jour. On Sel. Areas in Comm., vol. 12, no. 5, pp.
796-807, June 1994.
[13] P. Patel and J. Holtzman, Performance comparison of a DS/CDMA system using
successive interference cancellation (IC) scheme and a parallel IC scheme under
fading, Proceedings of IEEE ICC 94, pp. 510-514, 1994.
[14] S.H. Hwang, C.G. Kang, S.W. Kim, Performance analysis of interference
cancellation schemes for a DS/CDMA system under delay constraint, Seventh
IEEE International Symposium, vol. 2 , pp. 569 -573, 15-18 Oct. 1996.