Sie sind auf Seite 1von 4

A New Method to Design CDMA Spreading Sequences *

Bi Jianxin Wang Yingmin Tian Hongxin Yi Kechu


(National Key Lab. On ISN, Xidian University, Xi’an 710071, China)

Abstract: It is well-known that multi-value CDMA result suggests that the resulted codes behave better
spreading codes can be designed by means of a pair of than the conventional binary balance Gold codes for
mirror multi-rate filter banks based on some an asynchronous CDMA system.
optimizing criterion. This paper indicates that there
exists a theoretical bound in the performance of its . TWO-BAND MULTIRATE UNITARY
circulate correlation property, which is given by an FILTER-BANK THEORY(TBFMFB)
explicit expression. Based on this analysis, a criterion The coefficients of two-band unitary filter-banks
of maximizing entropy is proposed to design such only depend on h0 ( n) , which satisfies[3]
codes. Computer simulation result suggests that the ∑ h 0 (n )h 0 ( n + 2k ) = δ(k ) (1)
n
resulted codes outperform the conventional binary
balance Gold codes for an asynchronous CDMA From the viewpoint of the frequency domain, the
system. parameterization representation of the two-band
Key words: Multi-rate unitary filter banks; CDMA unitary is given as follows:
spreading code; Maximum entropy criteria.  H 0 ( z )  1 
 H (z )  = R (θ J ) ZR(θ J − 1 )Z L ZR(θ1 )  z −1  ( J = N / 2 ) (2)
 1   
 cos θ sin θ  1 0 
R( θ) =   Z= −2 
. INTRODUCTION  − sin θ cos θ 0 z 
The selection of spreading codes is essential for So, we can use N / 2 parameters {θ1 , θ 2 ,L , θ J } to
asynchronous CDMA systems. Recently, it is believed represent h0 ( n) .
to be promising that the CDMA spreading code is
designed based on multi-rate filter bank theory. For . The property of circulate correlation of the
this purpose, Akansu[1] has converted the design of codes based on TBFMFB
CDMA spreading code into an optimizing problem. From definition of circulate correlation, it is evident
Simulations suggested that the resulted multi-valued that
codes outperform Gold codes. Qinghua Shi[2], et al.,  + R (i ) (i = 0,2,4,L , N − 2)
1 R b (i ) =  a (3)
give another scheme, from frequency domain, for  − R a (i) (i = 1,3,5, L, N − 1)

spreading code generation based on multi-rate filter 2 R ab ( i) = 0 ( i = 0,2,4,L , N − 2) (4)


bank theory and a lattice structure. This paper where N-length normalized vector a and vector b
indicates that there exists a theoretical bound in the are the coefficients of the analysis filter bank, and
performance of spreading code’s circulate correlation R a (i ) denotes i-shifting circulate auto-correlation of

property, and gives an explicit expression for it. Based vector a and R ab (i ) denotes i-shifting circulate cross-
on this analysis, a criterion of maximizing entropy is correlation.
proposed to design such codes. Computer simulation Theorem 1 If N-length b (N= 2 u , u ∈ I ) is the

*
This paper is supported by National Natural Science Funds (N0. 69872027).
mirror of a normalized vector a , then, the following So we have
inequality is true: N−1 N −1
∑ R a (i ) + ∑ R ab ( i ) = 1 + 2R a (2 ) + 2 R a ( 4) L + 2 R a ( N − 2 ) ≥ 1
2 2 2 2 2
i= 1 i= 0
N −1 N −1
∑ R a (i )2 + ∑ R ab ( i) 2 ≥ 1 (0 ≤ i ≤ N − 1, i ∈ Z ) (5)
i =1 i =0 The equation is true if and only if
where the equality is valid if and only if R a (k ) = 0( k = 2,4, L, N − 2) .
R a (k ) = 0( k = 2,4, L, N − 2) .
It should be noticed that R a (i )2 (i ≠ 0) and R ab (i )2
Proof: Let A and B denote two circular matrices:

A = a0 [ a1 a2 L aN − 1 ] represent the normalized energies of all the possible


multi-path interfere and all the possible multi-access

B = b0 [ b1 b 2 L b N −1 ] interfere in the output of correlation receiver,


respectively. So the theorem implies :
where a i and bi represent two column vectors (1)It provides the theoretical basis for designing
obtained respectively by circular--shifting vector a CDMA spreading code based on unitary filter bank.
and b for i elements. Let When we design CDMA spreading code based on
N −1 N −1 TBFMFB, if we adopt the interfere energy minimum
y= ∑ R a (i ) 2 + ∑ R ab ( i) 2
i =0 i =0 criterion in the output of correlator, the optimization
solutions are the coefficients of unitary filter banks
= aA 2 + aB 2 = a ( AAT + BBT )a T (6)
(2)It gives us the lowest bound of the sum of all
Making use of the properties of circular matrix, we possible interfere energy in the output of correlator.
have [4] That is, if we use a pair of CDMA spreading codes
based on TBFMFB for user codes, the sum of all
A = a 1 I + a N P + a N −1 P 2 + L + a 2 P N −1
possible interfere energy in the output of correlator is,
on the condition of equal power transmission, greater
B = b1 I + b N P + b N −1 P 2 + L + b2 P N −1 (7)
than or equal to its “signal” energy. Moreover, the
0 1 0 L 0 0 equality is valid if and only if the codes are resulted
 
0 0 1 L 0 0
0 0 0 L 0 0 from a unitary filer bank.
where P=
M M M M M
 
0 0 0 L 0 1
1 0 0 
 0 0 L . a Proposed Method for Designing CDMA
Spreading Codes Based on Unitary Filter Bank
Substituting (7) into (6) and using ( Pi )T = P N − i , we
It is of significance that we adopt the minimum of
have the maximum of side lobes of absolute value of
y= a{(a 1I + a N P + a N−1P 2 + L + a 2 P N−1 )(a 1I + a 2 P + a 3 P 2 + L + a N P N−1 ) circulate auto-correlation and cross-correlation as an
+ ( b 1I + b N P + b N −1 P 2 + L + b 2 P N−1 )( b 1I + b 2 P + b 3 P 2 + L + b N P N−1 )}a T objection function. In another viewpoint, the flatter
(8) side lobes of absolute value of circulate auto-
Using the properties of matrix P ,we can simplify 8 correlation and cross-correlation are, the better the
into performance of the corresponding asynchronous
y=2+ R a (1) 2 + L + R a (N − 1) 2 + R a (1)R b (1) + L + R a ( N − 1) R b ( N − 1) CDMA systems will be. So,we use a criterion of
(9) maximizing entropy. The entropy of a vector
With Lemma1 we obtain p = [p1,L, p N ] =

 R (1) 2 , R ( 3) 2 , L, R ( N − 1) 2 , R (1) 2 , R ( 3) 2 , L, R ( N − 1) 2 
y=2+2[ R a (2 )2 + R a (4) 2 L + R a ( N − 2)2 ] (10)  a a a ab ab ab 
is defined as on the user code type presented above are investigated

[
entropy( p) = − p1 log 2 (p1 ) + L + pN log2 (p N ) ] (11)
in this section. Antipodal signaling and BPSK
modulation for CDMA are used in the simulations.
Furthermore, we also search the spreading code from The channel noise is modeled by additive white
the viewpoint of the frequency domain (Z-domain), gaussian noise (AWGN).
and get the same merits as reference [2]. It is obvious In the process of evaluating the BER performance,
that p is the function of θ1, θ 2, L , θ J and entropy(p ) it is assumed that for a user there are a multipath
is also the function of θ1, θ 2, L , θ J . So, the objection interfere and a multiuser interfere, whose delays are
function we use for optimizing is random variables evenly distributed in
( θˆ1 , θˆ2 , L, θˆJ ) = arg max [entropy(p{θ1 , θ 2 , L, θJ })] (12) [0 Tc L ( N − 1)Tc ] , where Tc denotes the duration
θ 1 , θ 2 ,L,θ J
of a chip. Because circulate correlation properties of
To make h0 ( n) suitable for DS-CDMA systems, we the optimized codes outperform those of the balanced
must spread h0 ( n) in both time and frequency as Gold code, it is expected that the BER performance of
evenly as possible. So, by using the entropy function the optimized codes are also better than the latter. The
defined above, we add two constrains as follows result showed in Fig.4 also substantiates it.

entropy(aa T ) ≥ c 1
. CONCLUDING REMARKS
2 Multi-rate unitary filter bank provides us powerful
entropy( FFT(a )FFT( a )T / FFT(a ) ) ≥ c 2
mathematical tool for designing CDMA spreading
where FFT(a ) denotes the Fourier transform of vector code. This paper provides the theoretical basis for
a . In experiment, these parameters designing CDMA spreading code based on unitary
are: c1 = c 2 = 4.46 , N = 32 . The experiment results are filter bank, and gives us a lower bound of the sum of
showed in fig.1,2, 3, and 4. Gold codes of length-31 all possible interfere energy in the output of correlator.
are used for comparison. The Gold codes result from Based on this analysis, a criterion of maximizing
summing (mod 2) a pair of m-sequences. The entropy is proposed to design such codes. It is worth
preferred pair of primitive polynomial is 45 and 67. while to further investigate how to extend this method
Fig.1 compares the spectra of a 32-length optimized to more generalized case where user number K is
code for the two-user case with a 31-length balanced more than 2.
Gold code. The circulate auto-correlation and
circulate cross-correlation of these codes are also REFERENCES
displayed in Fig.2 and 3. It is obvious from these [1] Ali N.Akansu, M.V.Tazebay and R.A.Haddad, “a new
figures that the circulate auto-correlation, circulate look at digital orthogonal transmultiplexers for CDMA
cross-correlation and frequency properties of the communications”,IEEE trans. On SP-45, No.1, January
multi-valued optimized code outperform the 1997.
comparable duration binary balanced Gold code. [2] Qinghua Shi and Shixin Cheng, “optimal spreading
sequence design based on PR-QMF theory”,
. PERFORMANCE OF THE PROPOSED ELECTRONICS LETTERS, 18th March 1999, Vol.35,
MULTIVALUED CDMA SPREADING CODES No.6.
By using Monte Carlo computer simulation method, [3] C.S.Burrus,R.A.Gopinath,H.Guo, “introduction to
the bit error rate (BER) performance of a two-user wavelets and wavelets transforms---a primer”, prentill
asynchronous CDMA communication system based hall,1998.
[4] 程云鹏等 矩阵论 第二版 西北工业大学出版社 [5] Verdu Sergio. Multiuser Detection. Cambridge University
1999 年 6 月 Press, 1998.
31-length balanced Gold code 31-length balanced Gold code
1

circulate auto-correlation function


2
magnitude function

1.5 0.5

1
0
0.5

0
0 0.5 1 1.5 2 2.5 3 -0.5
5 10 15 20 25 30
frequency(rad)
32-length optimized code time shift
32-length optimized code
1

circulate auto-correlation function


2
magnitude function

1.5
0.5

0.5 0

0
0 0.5 1 1.5 2 2.5 3
-0.5
frequency(rad) 5 10 15 20 25 30
time shift

Fig.1 Frequency spectra of the optimized 32-length Fig.2 Circulate auto-correlation functions of the
codes and 31-length balanced Gold codes. optimized 32-length codes and 31-length balanced
Gold codes.

31-length balanced Gold code 2-user asynchronous CDMA receiver


0
circulate cross-correlation function

0.4 10
single user
0.2 31-length balanced Gold code
32-length optimized code

0 -1
10

-0.2

-0.4
error rate

5 10 15 20 25 30
-2
time shift 10
32-length optimized code
circulate cross-correlation function

0.4

0.2
-3
10
0

-0.2

-4
-0.4 10
5 10 15 20 25 30 -2 -1 0 1 2 3 4 5 6 7 8
time shift SNR in dB

Fig.3 Circulate cross-correlation functions of Fig.4 BER performance of two-user asynchronous CDMA
the optimized 32-length codes and 31-length system for different user code types
balanced Gold codes.

Das könnte Ihnen auch gefallen