Performance Assessment of Polyphase Sequences Using Cyclic Algorithm

International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 5 Issue: 8 224 – 229
_______________________________________________________________________________________________
224
IJRITCC | August 2017, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
Performance Assessment of Polyphase Sequences Using Cyclic Algorithm
M. Rajasekhar
Research Scholar, Dept. of ECE
AU College of Engineering (A)
Visakhapatnam-530003, India
raja.sekhar441@yahoo.co.in
Dr. P. Rajesh Kumar
Professor & Chairman BoS, Dept. of ECE
AU College of Engineering (A)
Visakhapatnam-530003, India
rajeshauce@gmail.com
Abstract—Polyphase Sequences (known as P1, P2, Px, Frank) exist for a square integer length with good auto correlation properties are helpful
in the several applications. Unlike the Barker and Binary Sequences which exist for certain length and exhibits a maximum of two digit merit
factor. The Integrated Sidelobe level (ISL) is often used to define excellence of the autocorrelation properties of given Polyphase sequence. In
this paper, we present the application of Cyclic Algorithm named CA which minimizes the ISL (Integrated Sidelobe Level) related metric which
in turn improve the Merit factor to a greater extent is main thing in applications like RADAR, SONAR and communications. To illustrate the
performance of the P1, P2, Px, Frank sequences when cyclic Algorithm is applied. we presented a number of examples for integer lengths.
CA(Px) sequence exhibits the good Merit Factor among all the Polyphase sequences that are considered.
Keywords-Polyphase sequence, Cyclic Algorithm, Correlation level, Integrated sidelobe level, Merit Factor,
__________________________________________________*****_________________________________________________
I. INTRODUCTION
Radar waveform designs have examined Polyphase
sequences for a long time as a productive contracting option to
the diverse classes of Frequency-modulated signals [1][2]. In
radar & communication, sequences with good autocorrelation
properties are of main interest. The “goodness” depends on the
application we use. Frank sequence merit factors are analyzed
in [3]. P2 &Px sequences derived from linear-frequency-
modulated [4]. Frank & P1 are designed from step
approximation-to-linear frequency-modulation waveform [5].
Frank sequence merit factors are better than Chu. The two of
them significantly outperform binary sequences [6][7].
The sequences can be processed digitally even though they
are deriving from phase history of chirp or step-chirp Analog
signals. By and by, radar waveforms might be upgraded in a
first plan step by utilizing relationship measures and thus
connects with the uncertainty capacity to assess the effect of
phase shift changes on the execution. Prevalent execution as far
as the incorporated Sidelobe levels contrasted with the Frank
and P1 sequence is given by Px sequences that have been
presented by Rapajic and Kennedy [8].
The ideas driving Frank sequences have later been summed
up to encourage plans of Polyphase sequence of any length and
related work was combined in the plain Zad-off-Chu(FZC)
sequences or Chu Sequence[9][10]. A few execution parts of
the previously mentioned classes of Polyphase sequences have
been accounted for in writing fined years [11][12]. These
sequences were initially presented inside the specific
circumstance of utilizations for code division various to
(CDMA) frame works, while these conducts inside radar
situations have not been considered so for to the best of our
insight [13].
This paper is organized as follows. Section II characterizes
the measures used to encourage a quantitative execution
assessment of good correlation sequence. Section III presents
the essentials on classes of Polyphase sequences that are
utilized with radar applications. Section IV presents the basic
cyclic algorithm. On the basis of numerical results for
different are given in Section V. Section VI concludes the
paper.
II. PERFORMANCE MEASURES
Let N denotes the length of each Polyphase sequences
Ck=[Ck(0),Ck(1),….,Ck(N-1)…..] of a size P where
11  Nk
A. Correlation Function
The correlation function at a discrete shift k between a
Polyphase sequence is given by
1,......1,0,
**
1
 

 Nkcxxc kkn
N
kn
nk
(1)
Where
*
)( denotes the complex conjugate for scalar & the
conjugate transpose for vector & matrices.
B. Integrated Sidelobe Level (ISL)
TThe ISL for the Polyphase sequence
Ck=[Ck(0),Ck(1),….,Ck(N-1) can be defined as follows




1
1
2N
k
ckISL is the ISL metric (2)
The primary concentration of this paper is on calculation for
limiting the ISL metric or ISL related measurements over the
arrangement of Polyphase sequences. Note that minimization
of ISL metric is proportional to the improvement of the merit
factor defined as a performance metric in the below.
C. Merit Factor(MF)
The MF for the Polyphase sequence Ck=[Ck(0),Ck(1),….,Ck(N-
1) is defined as follows
ISL
k
MF N
C
C
N
ok
Nk
2
0
2
1
)1(
2
2





(3)
Polyphase sequences with good merit factor are desired in
many applications including range compression radar and
sonar and wireless communication.

Volume: 5 Issue: 8 224 – 229
_______________________________________________________________________________________________
225
_______________________________________________________________________________________
III. POLYPHASE SEQUENCES CLASSES
This section, we will describe the definitions of the
considered Polyphase sequence classes in terms of phases &
autocorrelation function. We adopted the sequence as it is used
in many radar-related publication work & communication
systems. In particular the P1, P2,Px, Frank sequences will be
described for radar applications. on behalf of that these
sequences beneficial properties and remarks also said.
A. Frank Sequence
Let the Polyphase sequence ),,.........,( 21 xxx N
X  of a
square integer length N=M2
(where M is a prime number). Due
to work on phase shift pulse codes in [14].The history of
complex-valued back as far as the 1950‟s.
The sequence elements are arranged as a M x M Matrix and
are given by Mth
roots of unity 



 
M
jw 2exp (5)
From the above, the actual Length „N‟ of Polyphase sequence
can be produced by matrix of roots of units row-by-row.
Polyphase sequence of perfect square length N=M2
are shown
in [15]. the related sequence are referred to as Frank sequence.
Frank Sequence: The elements xk(m,n) of kth
Frank sequence
is given as a matrix











 


mn
M
k
jnmx
MXM
MM
K k
X
2
exp)],([
-- (6)
The Phase components are mnk
Mk


2

Where 11  Mk , 1,0  Mnm and gcd(k,m)=1is
required. (m,n)th
element s of (8) can be point to the ith
element of a sequence length N in terms of the phase sequence
as follows
)()(),(: nmMinmnmMi kkk
  (7)
B. P1, P2, Px Sequence
This sequence can be considered for perfect square
length N=M2
only. In P1, P2, Px the phase components are
rearranged version of Frank phase components [16] by cluster
of zeros placed in the central part of the sequence.
)],(exp[),()()( nmjnmpnmMpip  (8)
here 1,0  Mnm and the phase components are
P1sequence:  nmMm
M
M
nm 








2
12
),(
(9)
P2 sequence:
















 n
M
m
M
M
nm
2
1
2
12
),(
(10)
Similarly for Px sequence Px:
evenMn
M
m
M
M
nm 
















2
1
2
12
),(
(11)
oddMn
M
m
M
M
nm 
















2
2
2
12
),(
(12)
Here 1,0  Mnm . Note that the phase elements of Px
are similar to that of P2 for M even.
IV. CYCLIC ALGORITHM
The approach in the following is much simpler &
computationally efficient than applying the optimization
technique for the Polyphase sequence [16] and [17]. This
makes feasible to work with quite large values of N (in some
radar and imaging applications we can choose 1000~N ). It
means we can choose Q first from practical consideration and
select QN  on computational as well as practical operation
accounts.
Let C
~
be the following block-Toeplitz matrix




























  







1
11
11
)()1(0
0)()1(
)()1(0
0)()1(
~
KL
NN
NN
L
L
L
L
C
CC
CC
CC
CC (13)
Note that C
~
is )1(  KLCNL . The auto & cross
correlation appeared in below are the elements of the positive
–semi finite matrix 
CC
~~
.
    







N
n
L
nnn
p
pp
N
n
L
LLL
prpr nnnn
1
1
~,1~
1
1
2
1
1
0,1
2
)()( ~ (14)
Where
,...2,1,0),()()()( ~~
1
~  


ppxplxlxpr nnn
L
pl
nnn
(15)
denote the (cross)- correlation of )(lxn
and )(~ lxn
at lag
.Consequently, a criterion related to above equation (15)
which has more compact form of the following
2
*
2
~~1
KICC
NLK

(16)
The above equation is the generalized correlation coefficient;
and we use this equation to evaluate the correlation quality of
a waveform. In such a case where *~~
CC is singular, it follows
that the maximum magnitude of its off-diagonal elements must
be of the order  1O or larger; consequently, the ratio
between )(max ~~,, pr nnnnp
and Krnn )0( is of the order
)/1( LO but not smaller.
In the following we assume that












1
1
1
1
1
N
K
Lposibly
N
K
LKLNL (17)
Under the above equation, if we relax any requirement on the
elements and the structure of C
~
, then the class matrix C
~
that
satisfies the equality KICC *~~
is given by
UKC 
~
(18)
Where U is an arbitrary semi-unitary matrix [18] i.e.
IUU *
usually the observation, we can reformulate (18) or
(20) in the following related(but bot equivalent) way:
2
)},({
~
min UKC
Uln


(19)
This is a non-convex problem, the following cyclic
minimization algorithm [19] [20], that is conceptually

Volume: 5 Issue: 8 224 – 229
_______________________________________________________________________________________________
226
_______________________________________________________________________________________
&computationally simple and also have good local
convergence properties.
A. BLOCK DIAGRAM OF THE CA
Step 1 Step 2
Cyclic Algorithm
Step 0:Initialize Uor possibly C
~
(in which case the sequence
of the next steps should be inverted), at some value suggested
by “prior knowledge”. i.e. initializing C with possibly good
existing sequence of Polyphase sequence (P1, P2, Px, Frank)
Step 1: Compute the semi-unitary matrix U& minimize the
(19) with respect to )}({ ln .
Step 2: with )}({ ln set to the most-recent values, minimize
the (19) w.r.to U.
Iteration: repeat step 1 and 2 until a practical convergence
criterion is satisfied.
The iteration can be terminated, when the relative difference
of the cost in (19) (i.e. the cost difference normalized by the
cost of the previous iteration) is less than or equal to the 10-3
value in the numerical example illustrated.
The minimization problem in step 1 has the following generic
form
























 

 

 j
L
p
L
p
eZpconstZpe j
1
2
1
Re2minmin
(20)



















 
L
p
Zp
1
argcosmax
(21)
Where  Zp are numbers given in (20) (21). The solution to
the above equation is given by








 
L
p
Zp
1
arg
(22)
In step 2 of the CA, the minimization problem solution can be
easily computed as. Let
*~~
UUCK  (23)
Above equation denotes the singular value decomposition
(SVD) of CK
~
, when U is NLNL , and U
~
is
NLKL  )1( then the said solution is given by [21][22].
*~
UU  (24)
Consider the simple illustration that with N=4 and L=22. Fig
1 shows the generalized correlation coefficient of the
waveform given by the CA, as a function of Sampler Number
L.(Note that for L<94, the GCC is too large to be of any
practical interest). As L increases, we can achieve the goal of
obtaining sequence with small value of auto & cross
correlation effectively which is sign of improving the Merit
Factor [23]. Additional simulation examples are shown in the
next section for different values of N.
Fig 1 Sample number Vs. GCC via Cyclic Algorithm
V RESULTS
We compared the merit factors of the Polyphase
sequence (P1, P2, Px, Frank), and that of CA Algorithm
initialized by sequence said above( denoted as CA(P1),
CA(P2), CA(Px), CA(Frank).
Note that the above sequences can be calculated for any value
of N of possible practical interest, with the only restriction that
N must be perfect square for Frank, P1, P2, Px sequences. We
computed the Merit factors of above eight type sequences (P1,
P2,Px, Frank) for the following length shown in Table I. the
results are shown in Fig (1 & 2). The correlation level is
defined as
1,...,1,log20
0
10  Nk
c
c
LevelnCorrelatio k
(25)
We calculated the Merit Factor of Polyphase
sequences for the lengths N=100 and N=256 and note that the
correlation levels of the CA(Px) and CA(Frank)sequence are
comparatively small from k close to zero and N-1.
The Merit Factor of Polyphase sequences using
conventional and cyclic algorithm for N=100 and N=256 are
shown in TABLE I. In conventional method the P2, Frank
exhibits the nearly same merit factor for length N=100 and P1,
P2 have the same merit factor for length N=256. The Px
exhibits the good merit factor among all the sequences when
cyclic algorithm is applied for both the lengths N=100 & 256.
Merit Factor of Polyphase sequences (p1, p2, Px,
frank) for integer values M=2 to 16 (i.e. N=4 to 256 where
N=M2
), the merit factor are exist lengths N=16 to N=256 are
shown in TABLE II. We notice that for integer values M=11 &
13 all the sequences such as CA(P1), CA(P2), CA(Px),
80 100 120 140 160 180 200 220 240 260 280
-55
-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
Sample Number
GCCindB
Input Data or good existing
Sequence of Polyphase
sequence
P1,P2,Px,
Frank
Compute the correlation
Function and normalize
result
Minimization
Problem
Compute the ISL
Metric
Computing the
Merit Factor
Plotting the result
i.e. Sequence length
Vs. Correlation
Level

Volume: 5 Issue: 8 224 – 229
_______________________________________________________________________________________________
227
_______________________________________________________________________________________
CA(Frank) exhibits the approximate value of the merit factor
for the co-integer values. Where the 11 and 13 are the prime
numbers. So, the P1, P2, Px, Frank can exhibit the good merit
factor for the sequences length N which are obtained from the
prime integer. Merit factor vs. sequence length are shown in
Fig. 3 and Fig. 4 for the conventional and when cyclic
algorithm applied.
TABLE I MERIT FACTOR OF POLYPHASE SEQUENCES
USING CONVENTIONAL AND CYCLIC ALGORITHM
FOR N=100 AND N=256.
Polyphase
Sequences
Conventional
Method
Merit Factor
Polyphase
Sequences
with
Cyclic
Algorithm
CA
Merit Factor
N=100 N=256 N=100 N=256
P1 22.452 36.103 CA(P1) 60.014 92.342
P2 23.121 36.014 CA(P2) 60.213 93.789
Px 25.012 40.012 CA(Px) 67.344 107.732
Frank 23.592 38.214
CA
(Frank)
61.414 94.355
TABLE II MERIT FACTOR OF POLYPHSE SEQUENCES
(P1, P2, Px, Frank) FOR LENGTHS N=4 TO N=256.
Sequence Length N=M2
M N
MF for
CA(P1)
MF for
CA(P2)
MF for
CA(Px)
MF for
CA(Frank)
2 4 --- ---- ---- ----
3 9 --- ---- ---- ----
4 16 8.614 8.081 12.808 10.671
5 25 13.459 14.0193 21.325 19.215
6 36 19.380 19.942 34.172 31.587
7 49 29.408 30.718 46.536 45.345
8 64 38.411 39.415 56.020 51.421
9 81 48.615 50.615 59.610 56.192
10 100 60.014 60.213 67.344 61.414
11 121 75.612 76.354 87.486 73.405
12 144 72.850 73.015 81.438 68.147
13 169 88.412 91.031 105.417 93.012
14 196 79.451 81.247 91.325 85.410
15 225 81.159 84.564 97.142 87.621
16 256 92.342 93.789 107.732 94.355
(a)
(b)
(c)
(d)
Fig 1. Correlation levels of the (a) CA(P1), (b) CA( P2), (c)
CA(Px) (d) CA(Frank) for the length N=100.
(a)
-100 -50 0 50 100
-250
-200
-150
-100
-50
0
k
CorrelationLevelin(dB)
CA(P1)
-100 -50 0 50 100
-250
-200
-150
-100
-50
0
k
CA(P2)
-100 -50 0 50 100
-250
-200
-150
-100
-50
0
k
CA(Px)
-100 -50 0 50 100
-250
-200
-150
-100
-50
0
k
CA(Frank)
-200 -100 0 100 200
-300
-200
-100
0
k
CA(P1)N=256

Volume: 5 Issue: 8 224 – 229
_______________________________________________________________________________________________
228
_______________________________________________________________________________________
(b)
(c)
(d)
Fig 2. Correlation levels of the (a) CA(P1), (b) CA( P2), (c)
CA(Px) (d) CA(Frank) for the length N=256.
Fig 3. Merit Factor of the P1, P2, Px, Frank for the length N=0
to 256.
Fig 4 Merit Factor of CA(P1), CA(P2), CA(Px), CA(Frank)
for the lengths N=0 to 256.
VI CONCLUSION
This paper presents the cyclic algorithm namely CA,
which can be applied to the Polyphase sequences such as P1,
P2, Px, Frank that have good correlation properties. The CA
algorithm makes use of SVD of matrix NLNL can be
computationally efficient upto the length of N=256. In
conventional method the best Merit Factor is obtained for Px
sequence only of 25.012& 40.012 for length N=100& 256
respectively. But when Cyclic Algorithm is appliedthe CA(Px)
and CA(Frank) sequences exhibits the better merit Factor
61.344 & 61.414 for N=100, 107.732 & 94.355 for N=256
respectively. The P1, P2, Px, Frank express good merit factor
for the M=11, 13 which are prime numbers (i.e. N=M2
,
N=121, 169). The merit factor comparison between P1, P2,
Px, Frank is Px>Frank>P2>P1. The minimum integer value in
N=M2
we can apply the CA is M=4 and maximum is 16.
V REFERENCES
[1] N. Levanon and E. Mozeson, “Radar Signals,” Chichester: John
Wiley & Sons, 2004.
[2] S. W. Golomb and G. Gong, “Signal Design for Good
Correlation for Wireless Communication, Cryptography, and
Radar,” Cambridge: Cambridge University Press, 2005.
[3] M. Antweller and L. Bomer, “Merit factor of Chu and Frank
sequences,” Electron. Lett., vol. 26, pp. 2068-2070, Dec. 1990.
[4] Lewis, B. L., and Kretschmer, F. F., Jr.(1982), “liner frequency
modulation derived Polyphase pulse compression codes” IEEE
Trans. Aerospace and Electronic Systems, AES-17, 5(September
1982), 637-641.
[5] Kretschmer, F.F.,Jr.,and Lewis, B.L.(1983). “Doppler properties
of Polyphase coded pulse compression waveforms.”IEEE Trans.
Aerospace and Electronic Systems. AES-19, 4(July 1983), 521-
531.
[6] J. Jedwab, “ A survey of the merit factor problem for binary
sequences,” in sequences and Their Applications – SETA 2004,
T. Helleseth, D. Sarwate, H. Y. Song, and K. Yang, Eds.
Heidelberg, Germany: Springer-Verlag, 2005, vol.3486, Lecture
Notes in Computer Science, pp. 30-55.
[7] M. J. E. Golay. “Sieves for low autocorrelation binary
sequences,” IEEE Trans. Inf. Theory, vol. IT-23, pp. 43-51, Jan.
1977.
[8] P. B. Rapajic and R. A. Kennedy, “Merit Factor Based
Comparison of New Polyphase Sequences,” IEEE Commun.
Letters, vol. 2, no. 10, pp. 269– 270, Oct. 1998
[9] R. L. Frank and S. A. Zadoff, “Phase Shift Pulse Codes with
Good Periodic Correlation Properties,” IEEE Trans. on Inf.
Theory, vol. 19, no. 1, pp. 115-120, Jan. 1975.
[10] D. C. Chu, “Polyphase Codes with Good Periodic Correlation
Properties,” IEEE Trans. on Inf. Theory, vol. 18, no. 4, pp. 531-
532, July 1972.
-200 -100 0 100 200
-300
-200
-100
0
k
CA(P2)N=256
-200 -100 0 100 200
-300
-200
-100
0
k
CA(Px)N=256
-200 -100 0 100 200
-300
-200
-100
0
k
CA(Frank)N=256
0 50 100 150 200 250 300
0
10
20
30
40
50
Sequence Length N
MeritFactor
P1
P2
Px
Frank
0 50 100 150 200 250 300
0
20
40
60
80
100
120
Sequence Length N
MeritFactor
CA(P1)
CA(P2)
CA(Px)
CA(Frank)

Volume: 5 Issue: 8 224 – 229
_______________________________________________________________________________________________
229
_______________________________________________________________________________________
[11] B. L. Lewis and F. F. Kretschmer, “Linear Frequency
Modulation Derived Polyphase Pulse Compression Codes,”
IEEE Trans. on Aerospace and Electronic Systems, vol. 18, no.
5, pp. 637–641, Sept. 1982.
[12] F. F. Kretschmer and B. L. Lewis, “Doppler Properties of
Polyphase Coded Pulse Compression Waveforms,” IEEE Trans.
on Aerospace and Electronic Systems, vol. 19, no. 4, pp. 521-
531, July 1983.
[13] B. L. Lewis and F. F. Kretschemer, “ A New Class of Polyphase
Pulse Compression Codes and Techniques, “IEEE Trans, on
Aerospace and Electronic Systems, vol. 17, no. 3, pp. 364-372,
May 1981.
[14] R. C. Heimiller, “Phase Shift Pulse Codes with Good Perodic
Correlatin Properties” IRE Trans. On Inf. Therory, Vol. 7, pp.
254-257, Oct. 1961.
[15] R. L. Frank and S. A. Zadoff, “Phase Shift Pulse Codes with
Good Periodic Correlation Properties,: IEEE. Trans. On Inf.
Theory, vol. 19, no. 1, pp. 115-120, Jan. 1975.
[16] H. Deng, “Polyphase code design for orthogonal netted radar
systems,” IEEE Trans. Signal Process., vol. 52, no. 1, pp. 3126-
3135, Nov. 2004.
[17] H. A. Khan, Y. Zhang, C. Ji, C. J. Stevens, D. J. Edwards, and
D. O‟Brien, “Optmizing Polyphase sequence for orthogonal
netted radar,” IEEE Trans. Signal Process., vol. 13, no. 10, pp.
584-592, Oct. 2006.
[18] P. Stoica, H. He, and J. Li, “New algorithms for desigining
unimodilar sequences with good correlation properties,” IEEE
Transactions on Signal Processing, vol. 57, no. 4, pp. 1415-
1425, April 2003.
[19] J. A. Tropp, I. S. Dhillon, R. W. Heath, and T. Strohmer,
“CDMA signature sequences with low peak-to-average-power
ratio via alternating projection,” in Proc. 37th
Asilomar Conf.
Signals, Systems, Computers, Pacific Grove, CA, Nov. 2013,
vol. 1, pp. 475-479.
[20] J. A. Tropp, I. S. Dhillon, R. W. Heath, and T. Strohmer, “
Designing structured tight frames via an alternating projection
method,” IEEE Trans. Inf. Theory, vol. 51, no. 1, pp. 188-209,
Jan. 2005.
[21] C. R. Rao, “Matrix approximations and reduction of
dimensionality in multivariate statistical analysis,” in
Multivariate Analysis, P. R. Krishnaiah, Ed. Amsterdam, The
Netherlands: North-Holland, 1980, Vol. 5, pp. 3-22.
[22] R. A. Horn and C. R. Johnson, Matrix Analysis. Cambridge,
U.K.: Cambridge Univ. Press, 1985.
[23] J. Li, P. Stoica, and J. Li, “Signal synthesis and receiver design
for MIMO radar imaging,” IEEE Trans. Signal Process., vol.56,
pp. 2593-2598.

Performance Assessment of Polyphase Sequences Using Cyclic Algorithm

More Related Content

What's hot (18)

Similar to Performance Assessment of Polyphase Sequences Using Cyclic Algorithm (20)

More from rahulmonikasharma (20)

Recently uploaded (20)

Performance Assessment of Polyphase Sequences Using Cyclic Algorithm