Construction of Equiangular Signatures for
Synchronous CDMA Systems
Robert W. Heath Jr.
Dept. of Elect. and Comp. Engr.
The University of Texas at Austin
1 University Station C0803
Austin, TX 78712-1084 USA
rheath@ece.utexas.edu
Joel A. Tropp
Inst. for Comp. Engr. and Sci. (ICES)
The University of Texas at Austin
1 University Station C0200
Austin, TX 78712
jtropp@ices.utexas.edu
Inderjit. S. Dhillon
Dept. of Comp. Sci.
The University of Texas at Austin
1 University Station C0500
Austin, TX 78712
inderjit@cs.utexas.edu
Thomas Strohmer
Dept. of Mathematics
University of California, Davis
Davis, CA 95616 USA
strohmer@math.ucdavis.edu
Abstract— Welch bound equality (WBE) signature sequences
maximize the uplink sum capacity in direct-spread synchronous
code division multiple access (CDMA) systems. WBE sequences
have a nice interference invariance property that typically holds
only when the system is fully loaded and the signature set must be
redesigned and reassigned as the number of active users changes
to maintain this property. An additional equiangular constraint
on the signature set, however, maintains interference invariance.
Finding such signatures requires imposing equiangular side con-
straints on an inverse eigenvalue problem. This paper presents an
alternating projection algorithm that can design WBE sequences
that satisfy equiangular side constraints. The proposed algorithm
can be used to find Grassmannian frames as well as equiangular
tight frames. Though one projection is onto a closed but non
convex set, it is shown that this algorithm converges to a fixed
point, and these fixed points are partially characterized.
I. INTRODUCTION
Signature sequences that maximize the sum capacity in the
uplink of direct-spread synchronous code division multiple
access (CDMA) systems in Gaussian noise are known to
satisfy Welch’s bound on the total squared correlation with
equality [1]. These sequences, known as Welch bound equality
(WBE) signature sequences, are determined by the number
of users and the dimensionality of the signature space. They
have the interesting interference invariance property in that
each signature sees exactly the same interference power. Thus
the interference experienced by a user is independent of
the signature assigned to that user. Unfortunately, when the
number of active users changes, the signatures must generally
be recomputed and reassigned to maintain the interference
invariance [2].
Recently a class of signatures, known as Grassmannian
signatures, were introduced that satisfy interference invariance
even when subsets of the available users are active [3]. This
signature construction is intimately related to the problem of
sphere packing in the Grassmann manifold, in this case one-
dimensional subspaces (lines), and more specifically to the
construction of Grassmannian tight frames [4]. The interfer-
ence invariance properties comes from the fact that because
Grassmannian signatures satisfy Welch’s lower bound on the
maximum correlation with equality, they are equiangular (the
correlation is the same for all distinct signature pairs) and
maximally spaced with the smallest possible inner product.
The equiangular property provides interference invariance.
Unfortunately, signatures that are both equiangular and max-
imally spaced are quite rare. Some explicit constructions are
available in the articles [3], [4], [5]. Signatures derived from
good packings in the Grassman manifold, even the best line
packings tabulated by Sloane [6], do not generally satisfy the
WBE property when they are not equiangular. In cases where
such signatures do not exist, we would be satisfied with a
WBE whose constituent signatures are close to equiangular.
Recently proposed numerical algorithms for finding WBEs
(e.g., [7], [8], [9], [10]), however, do not easily incorporate
equiangular side constraints.
In this paper, we present an algorithm for finding Welch
bound equality signature sequences that are exactly (or nearly)
equiangular. Our approach builds on our recently proposed
iterative algorithm for constructing CDMA signature se-
quences [11], which has also been used to find signatures
satisfying peak-to-average ratio constraints [12]. The idea is
to alternately solve two matrix nearness problems, one that
finds the closest signature set satisfying Welch’s bound with
equality and the other that finds the nearest set of equiangular
signatures. This algorithm is related to a method used by
Chu for solving an inverse eigenvalue problem [13]. Our
algorithm can also be used to find Grassmannian frames as
well as equiangular tight frames. We argue that our algorithm
converges to a fixed point, and we claim that the class of fixed
points contains the desired sequences. Detailed proofs of these
results are deferred due to space constraints [14].
II. S
IGNATURE DESIGN PRELIMINARIES
Consider the uplink of a single cell, short code, synchronous
CDMA system with N total signatures and a processing gain
d. Let x
k
denote the d×1 signature, code, or sequence, of user
k, normalized as x
k
=1for k =1,...,N. We assume that
the maximum number of active users allowed in the system is
N ≥ d>1.
If the signatures x
k
form an orthogonal set, the length d
determines the allowable number of users. It has been shown
that nonorthogonal signature sets where N>musers may be
necessary to achieve the full sum-capacity of the synchronous
single-cell CDMA channel [1]. These sequences are called
Welch bound equality sequences [15] since they satisfy the
Welch bound on the total squared correlation with equality.
ISSSTA2004, Sydney, Australia, 30 Aug. - 2 Sep. 2004
N-NNNN-NNNN-N/04/$17.00 © 2004 IEEE
1
WBE signature sequences have a number of nice properties
as summarized in [15], [16]. Perhaps the most interesting
property is that, using WBE sequences, the interference is
uniform across all users [16]. The sum total interference in
the system is given by
k
l=k
|x
k
, x
l
|
2
− N
2
which for
WBE sequences is simply
N
2
d
− N . Using WBE sequences,
the total interference power experienced by user k is
I(k)=
N
l=1
|x
k
, x
l
|
2
−1=
N − d
d
for k =1, 2,...,N
(1)
and is the same for every user. Thus the SINR performance
for any user k is simply
SINR
k
=
σ
2
s
σ
2
v
−1
+
d
N − d
−1
−1
(2)
and the performance only depends on N and d. Unfortunately,
interference invariance only occurs when the system is fully
loaded [2], [15], i.e. N users are active. The reason is that
a WBE set for N>dusers almost always ceases to be a
WBE set if any M<dsequences are removed from or added
to the set [3]. Thus if
¯
N<Nusers are active, the whole
signature set will need to be recomputed for (d,
¯
N) and the
signatures reassigned or additional power control will have to
compensate for interference inequality.
III. E
QUIANGULAR SIGNATURES FOR CDMA SYSTEMS
An interesting subclass of WBE signature sequences, known
as Grassmannian signatures, retains the interference invari-
ance property even when a subset of signatures are active
[3]. Grassmannian signatures are constructed from optimal
packings of lines on the Grassmann manifold. These signature
sequences satisfy two important properties:
1) They are equiangular, i.e.,
|x
k
, x
l
| = c for all k, l with k = l (3)
for some constant c ≥ 0.
2) They are maximally spaced, i.e. c in (3) is as small as
possible.
The equiangular property means that every signature is equally
“far” from every other signature. This is the origin of the
interference invariance property. For example, if N is the set
that indexes the active signatures, then the total interference
experienced by any user k =1, 2,...,N is
I(k)=
l∈N /k
|x
k
, x
l
|
2
= c (|N| − 1) (4)
which only depends on the cardinality of N.
The maximally spaced property implies that the signature
sequence minimizes the maximum angle between the lines
generated by x
k
and x
l
. Let
ρ(N,m):= max
k,l,k=l
|x
k
, x
l
|
denote the maximum correlation. Grassmannian signatures
achieve the lower bound on the maximum correlation for a
line packing given by (see [17] for example)
ρ(N,d) ≥
N − d
d(N − 1)
. (5)
Further, if equality holds in (5), then the signatures are
equiangular, maximally spaced, and form a WBE signature
sequence set [4].
In general, it is torturous to find signatures that satisfy (5)
with equality. Most of the current research has approached
the design problem with algebraic tools. A notable triumph of
this type is the construction of Kerdock codes over Z
2
and Z
4
due to Calderbank et al. [5]. Other explicit constructions are
discussed in the articles [4], [3]. In the numerical realm, Sloane
has used his Gosset software to produce and study sphere
packings in real Grassmannian spaces [6]. Sloane’s algorithms
have been extended to complex Grassmannian spaces in [18].
We are not aware of any other numerical methods.
Some examples of signatures that achieve the bound in (5)
are available in [4] but generally they are hard to find. The
reason is that while good line packings have been tabulated for
various d and N, these packings do not necessarily maintain
the equiangular property. On the other hand, some equiangular
signature sets do not achieve the maximally spaced property,
e.g., it is possible to find five equiangular vectors in R
3
but
they are not maximally spaced. In both cases, the resulting
packing may not satisfy the WBE property enjoyed when
equality is satisfied and thus may no longer be capacity-
optimal.
When signature sequences are not available that satisfy (5)
with equality, it is not possible to simultaneously obtain a
signature sequence that is equiangular, maximally spaced, and
satisfies Welch’s bound on the total squared correlation with
equality. Since the equiangular property provides interference
invariance, it may be of practical interest to sacrifice the
maximally spaced requirement but yet maintain the constraint
that the signature sequence forms a WBE sequence to ensure
sum-capacity maximization. The objective of this paper is
to present an algorithm for finding WBEs that are nearly
equiangular.
Let X =[x
1
, x
2
,...,x
N
] be the signature matrix con-
structed from the signature set. It can be shown that a neces-
sary and sufficient condition for a signature sequence to satisfy
the Welch bound with equality is that the d positive singular
values of S are identical. A matrix with this property is called
a tight frame. Our goal, then, is to construct a signature matrix
X with the following properties.
i. The matrix is a tight frame: XX
∗
= α I
d
.
ii. Each column has the correct norm: x
n
=1.
iii. The columns are equiangular: |x
k
, x
m
| = c for all
k = m and some c.
In this paper we present an algorithm that tries to calculate
such sequences that we call equiangular tight frames. In the
sequel, we summarize the method and its theoretical behavior.
2
IV. ALTERNATING PROJECTION PRELIMINARIES
Our technique is based on an alternating projection between
Property (i) and Properties (ii)–(iii). The algorithm attempts to
compute a nearby matrix (in terms of the Frobenius norm) that
satisfies Properties (i)–(iii).
Since the Gram matrix X
∗
X displays all of the inner
products, it is more natural to construct the Gram matrix of an
equiangular tight frame than to construct the signature matrix
directly. Therefore, our algorithm will alternate between the
collection of Hermitian matrices that have the correct spectrum
and the collection of Hermitian matrices that have sufficiently
small off-diagonal entries.
Define a collection that contains the Gram matrices of all
d × Nα-tight frames:
G
α
def
= {G ∈ C
N×N
: G = G
∗
and
G has eigenvalues (α,...,α
d
, 0,...,0)}. (6)
The set G
α
is essentially the Grassmannian manifold that
consists of d-dimensional subspaces of C
N
[17]. One may also
identify the matrices in G
α
as rank-d orthogonal projectors,
scaled by α.
Theorem 1 shows how to find a matrix in G
α
nearest to an
arbitrary Hermitian matrix.
Theorem 1: Suppose that Z is an N ×N Hermitian matrix
with a unitary factorization UΛU
∗
, where the entries of Λ are
arranged in algebraically non-increasing order. Let U
d
be the
N × d matrix formed from the first d columns of U . Then
α U
d
U
d
∗
is a matrix in G
α
that is closest to Z with respect to
the Frobenius norm. This closest matrix is unique if and only
if λ
d
strictly exceeds λ
d+1
.
Proof: See [14] for details.
Let H be a closed collection of N ×N Hermitian matrices
that satisfy the structural constraint set motivated by the
equiangular property:
H
µ
def
= {H ∈ C
N×N
: H = H
∗
,
diag H = m1 and max
m=n
|h
mn
|≤µ}.
It may seem more natural to require that the off-diagonal
entries have modulus exactly equal to µ, but our experience
indicates that the present formulation works better, perhaps
because H
µ
is convex.
The following proposition shows how to produce the nearest
matrix in H
µ
.
Proposition 2: Let Z be an arbitrary matrix. With respect
to Frobenius norm, the unique matrix in H
µ
closest to Z has
a unit diagonal and off-diagonal entries that satisfy
h
mn
=
z
mn
if |z
mn
|≤µ and
µ e
iargz
mn
otherwise.
We use i to denote the imaginary unit.
Proof: The argument is straightforward.
The objective of alternating minimization is to find a
solution to the following question.
Problem 1: Find a matrix in G
α
that is minimally distant
from H with respect to a given norm.
If the two sets intersect, any solution to this problem will lie in
the intersection. Otherwise, the problem requests a tight frame
with unit norm columns whose Gram matrix is “most nearly
equiangular.” We do not mention the problem of producing a
matrix in H that is nearest to G
α
because it is not generally
possible to factor a matrix in H to obtain a frame with
dimensions d × N.
V. S
TATEMENT OF THE ALGORITHM
Practically, implementing to proposed alternating minimiz-
ing involves alternately enforcing the two aforementioned
constraint sets until reaching a suitable stopping criterion.
Convergence is an issue since the tight frame constraint set
is non-convex.
Algorithm 1 (Alternating Projection):
I
NPUT:
• An arbitrary matrix S
0
•
The number of iterations J
O
UTPUT:
• A signature matrix X
J
PROCEDURE:
1) Let j =1and H = S
∗
0
S
0
.
2) Find G
j
, the Gram matrix nearest to H
j−1
in Frobenius
norm that has Property (i).
3) Find H
j
, the nearest Gram matrix to G
j
in Frobenius
norm that has Properties (ii) and (iii).
4) Increment j. Repeat Steps 2–4 until j>J.
5) Solve for X
J
by factoring G
J
using a finite-step method
such as [19] for example.
VI. S
UMMARY OF CONVERGENCE RESULTS
The machinery of point-to-set maps is required to under-
stand the convergence of this algorithm, so we must refer the
reader to [14] for details. In this section we shall summarize
the convergence results.
A. Basic Convergence Results
It should be clear that alternating projection never increases
the distance between successive iterates. This does not mean
that it will locate a point of minimal distance between the
constraint sets. It can be shown, however, that it is globally
convergent in a weak sense.
Define the distance between a point M and a set Y via
dist(M, Y )= inf
Y ∈Y
Y − M
F
.
Theorem 3 (Global Convergence of Algorithm): Let Y
and Z be closed sets, one of which is bounded. Suppose
that alternating projection generates a sequence of iterates
{(Y
j
, Z
j
)}. This sequence has at least one accumulation
point.
• Every accumulation point lies in Y × Z .
3
• Every accumulation point (Y , Z ) satisfies
Y − Z
F
= lim
j→∞
Y
j
− Z
j
F
.
• Every accumulation point (Y , Z ) satisfies
Y − Z
F
=dist(Y , Z )=dist(Z , Y ).
For a proof of Theorem 3, see the Appendix in [14].
The convergence of the proposed algorithm is geometric at
best [20], [21], [22], [23]. This is the major shortfall of
alternating projection methods.
Note that the sequence of iterates may have many accu-
mulation points. Moreover, the last condition does not imply
that the accumulation point (
Y , Z ) is a fixed point of the
alternating projection. So what are the potential accumulation
points of a sequence of iterates? Since the algorithm never
increases the distance between successive iterates, the set of
accumulation points includes every pair of matrices in Y ×Z
that lie at minimal distance from each other.
B. Convergence Results
Besides the general convergence result, Theorem 3, we also
obtain a local convergence result.
Theorem 4: Assume that the alternating projection between
G
α
and H
µ
generates a sequence of iterates {(G
j
, H
j
)},
and suppose that there is an iteration J during which
G
J
− H
J
F
<N/(d
√
2). The sequence of iterates possesses
at least one accumulation point, say (
G , H).
• The accumulation point lies in G
α
× H
µ
.
• The pair (G , H ) is a fixed point of the alternating pro-
jection. In other words, if we applied the algorithm to
G
or to
H, every iterate would equal (G , H ).
• The accumulation point satisfies
G − H
F
= lim
j→∞
G
j
− H
j
F
.
• The component sequences are asymptotically regular, i.e.
G
j+1
− G
j
F
→ 0 and H
j+1
− H
j
F
→ 0.
• Either the component sequences both converge in norm,
G
j
− G
F
→ 0 and
H
j
− H
F
→ 0,
or the set of accumulation points forms a continuum.
Proof: See the Appendix in [14].
VII. NUMERICAL EXPERIMENTS
A. Example Construction
First, let us illustrate just how significant a difference there
is between vanilla signature matrices and equiangular signature
matrices. Here is the Gram matrix of a six-vector, unit-norm
tight frame for R
3
:
1.0000 0.2414 −0.6303 0.5402 −0.3564 −0.3543
0.2414 1.0000 −0.5575 −0.4578 0.5807 −0.2902
−0.6303 −0.5575 1.0000 0.2947 0.3521 −0.2847
0.5402 −0.4578 0.2947 1.0000 −0.2392 −0.5954
−0.3564 0.5807 0.3521 −0.2392 1.0000 −0.5955
−0.3543 −0.2902 −0.2847 −0.5954 −0.5955 1.0000
.
d
N 23456
3 RR .. .. ..
4 CRR .. ..
5 .. . RR ..
6 .. R . RR
7 .. CC . R
8 .. . C ..
9 .. C ..C
10 .. .. . R .
11 .. .. . CC
12 .. .. . . C
13 .. .. C ..
14 .. .. . . .
15 .. .. . . .
16 .. .. C . R
17 .. .. .. . .
18 .. .. .. . .
19 .. .. .. . .
d
N 234 5 6
20 .. .. .. . .
21 .. .. .. C .
22 .. .. .. . .
23 .. .. .. . .
24 .. .. .. . .
25 .. .. .. C .
26 .. .. .. .. .
27 .. .. .. .. .
28 .. .. .. .. .
29 .. .. .. .. .
30 .. .. .. .. .
31 .. .. .. .. C
32 .. .. .. .. .
33 .. .. .. .. .
34 .. .. .. .. .
35 .. .. .. .. .
36 .. .. .. .. C
TABLE I
EQUIANGULAR WBE SIGNATURE SETS
The notations R and C respectively indicate that alternating projection was
able to compute a real, or complex, equiangular tight frame. Note that every
real, equiangular tight frame is automatically a complex, equiangular tight
frame. One period (.) means that no real, equiangular tight frame exists, and
two periods (..) mean that no equiangular tight frame exists at all.
Notice that the inner-products between vectors are quite dis-
parate, ranging in magnitude between 0.2392 and 0.6303.
These inner products correspond to acute angles of 76.2
◦
and
50.9
◦
. In fact, this tight frame is pretty tame; the largest inner
products in a unit-norm tight frame can be arbitrarily close
to one
1
. The Gram matrix of a six-vector, equiangular tight
frame for R
3
looks quite different:
1.0000 0.4472 −0.4472 0.4472 −0.4472 −0.4472
0.4472 1.0000 −0.4472 −0.4472 0.4472 −0.4472
−0.4472 −0.4472 1.0000 0.4472 0.4472 −0.4472
0.4472 −0.4472 0.4472 1.0000 −0.4472 −0.4472
−0.4472 0.4472 0.4472 −0.4472 1.0000 −0.4472
−0.4472 −0.4472 −0.4472 −0.4472 −0.4472 1.0000
.
Every pair of vectors meets at an acute angle of 63.4
◦
. The
vectors in this frame can be interpreted as the diagonals of an
icosahedron [17].
B. Summary of Basic Constructions
We have used alternating projection to compute equiangu-
lar tight frames, both real and complex, in dimensions two
through six. The algorithm performed poorly when initial-
ized with random vectors, which led us to adopt a more
sophisticated approach. We begin with many random vectors
and winnow this collection down by repeatedly removing
whatever vector has the largest inner product against another
vector. It is fast and easy to design starting points in this
manner, yet the results are impressive. These calculations are
summarized in Table I. Alternating projection can locate every
real, equiangular tight frame signature matrix in dimensions
two through six; algebraic considerations eliminate all the
1
To see this, consider a tight frame that contains two copies of an
orthonormal basis, where one copy is rotated away from the other by an
arbitrarily small angle.
4
d 4 8 16 32 64
N 5 9 18 36 70
Minimum Cor. 0.2500 0.1250 0.0021 0.0006 0.0000
Average Cor. 0.2500 0.1250 0.0765 0.0516 0.0326
Std. Dev. Cor. 0.0000 0.0000 0.0429 0.0301 0.0212
Max Cor. 0.2500 0.1250 0.1250 0.0966 0.0607
Max Cor. Packing 0.2500 0.1250 0.0911 0.0674 0.0427
Max Cor. Bound 0.2500 0.1250 0.0857 0.0598 0.0369
TABLE II
NEAR-EQUIANGULAR WBE SIGNATURE SETS
Summary of the correlation behavior of specific WBE sequences resulting
from the proposed algorithm. The last three lines compare the maximum cor-
relation of the candidate near-equiangular WBE with the maximum correlation
of the best line packing found for (d, N ) without the tight frame constraint
and the lower bound on the maximum correlation (5).
remaining values of N [4]. In the complex case, the algorithm
was able to compute every equiangular tight frame that we
know of. Unfortunately, no one has yet developed necessary
conditions on the existence of complex, equiangular tight
frames aside from the upper bound, N ≤ d
2
, and so we have
been unable to rule out the existence of other ensembles.
C. Overloaded System Example
We have also constructed some WBEs in dimensions of
d =2
k
for k =2, 3,...,6 and an overload factor of ten
percent. The results of this construction are illustrated in Table
II. Constructions (4, 5) and (8, 9) are exact equiangular tight
frames (corresponding to the simplex). In the other cases, the
WBEs are only nearly equiangular. Because of the tight frame
constraint, the maximum correlation is somewhat higher than
that of the best line packing for those combinations (without
the tight frame constraint), and is larger than the lower
bound. The standard deviation of the correlation between
two signatures provides a measure of “equiangularity.” Lower
values indicate the signatures are more equiangular. In the
proposed examples, there is some variability especially for
larger dimensions. This is because N is not much bigger
than d thus there are fewer degrees of freedom to enforce
the equiangular property. For d =64and N = 128, though, a
construction exists that is equiangular and maximally spaced
[3].
VIII. C
ONCLUSION
We have proposed an alternating minimization that is capa-
ble of finding optimal CDMA signature sequences that satisfy
equiangular side constraints and discussed convergence of the
algorithm. This algorithm can also be used to solve for uncon-
strained optimal CDMA signature sequences, sequences with
peak-to-average power ratio side constraints [12], and spec-
trum constraints. A major issue with the proposed algorithm
is that the resulting sequences are generally complex valued
and this may lead to implementation challenges. Incorporating
binary or finite alphabet constraints on the signatures is an
interesting topic for future research.
R
EFERENCES
[1] M. Rupf and J. L. Massey, “Optimum sequence multisets for syn-
chronous code-division multiple-access channels,” IEEE Trans. Inform.
Theory, vol. 40, no. 4, pp. 1261–1266, July 1994.
[2] V. Kravcenko, H. Boche, F. Fitzek, and A. Wolisz, “No need for
signaling: Investigation of capacity and quality of service for multi-code
CDMA systems usign the wbe ++ approach,” in Proc. 4th Int. Workshop
on Mob. and Wireless Commun. Networks, 2002, pp. 110–114.
[3] R. W. Heath Jr., T. Strohmer, and A. J. Paulraj, “Grassmanian signatures
for CDMA systems,” in Proc. of IEEE Global Telecommunications
Conf., San Francisco, CA, Dec. 2003.
[4] T. Strohmer and R. W. Heath Jr., “Grassmannian frames with appli-
cations to coding and communication,” Appl. Comp. Harmonic Anal.,
vol. 14, no. 3, pp. 257–275, May 2003.
[5] A. R. Calderbank, P. J. Cameron, W. M. Kantor, and J. J. Seidel, “Z
4
-
Kerdock codes, orthogonal spreads and extremal Euclidean line sets,”
Proc. London Math. Soc., vol. 75, no. 2, pp. 436–480, 1997.
[6] N. J. A. Sloane, “Packing planes in four dimensions and other myster-
ies,” in Proceedings of the Conference on Algebraic Combinatorics and
Related Topics, Yamagata, Japan, Nov. 1997.
[7] C. Rose, “CDMA codeword optimization: interference avoidance and
class warfare,” IEEE Trans. Inform. Theory, vol. 47, no. 6, pp. 2368–
2382, Sept. 2001.
[8] S. Ulukus and R. D. Yates, “Iterative construction of optimum signature
sequence sets in synchronous CDMA systems,” IEEE Trans. Inform.
Theory, vol. 47, no. 5, pp. 1989–1998, 2001.
[9] C. Rose, S. Ulukus, and R. D. Yates, “Wireless systems and interference
avoidance,” IEEE Trans. Wireless Comm., vol. 1, no. 3, pp. 415–428,
Jul. 2002.
[10] P. Anigstein and V. Anantharam, “Ensuring convergence of the MMSE
iteration for interference avoidance to the global optimum,” IEEE Trans.
Inform. Th., vol. 49, no. 4, pp. 873–885, Apr. 2003.
[11] J. A. Tropp, R. W. Heath Jr., and T. Strohmer, “Optimal CDMA signature
sequences, inverse eigenvalue problems and alternating projection,” in
Proceedings of the 2003 IEEE International Symposium on Information
Theory, Yokohama, July 2003, p. 407.
[12] J. A. Tropp, I. S. Dhillon, R. W. Heath Jr., and T. Strohmer, “CDMA
signature sequences with low peak-to-average-power ratio via alternating
projection,” in Proceedings of the 37th Annual Asilomar Conference on
Signals, Systems and Computers, Monterrey, Nov. 2003.
[13] M. T. Chu, “Constructing a Hermitian matrix from its diagonal entries
and eigenvalues,” SIAM J. Matrix Anal. Appl., vol. 16, no. 1, pp. 207–
217, Jan. 1995.
[14] J. A. Tropp, I. Dhillon, R. W. Heath Jr., and T. Strohmer, “Designing
structured tight frames via an alternating projection method,” Dec. 2003,
the University of Texas at Austin, ICES Report 03-50, December 2003,
also submitted to the IEEE Trans. Inform. Theory.
[15] D. V. Sarwate, “Meeting the Welch Bound with equality,” in Sequences
and their Applications. London: Springer, 1998, pp. 79–102.
[16] J. L. Massey and T. Mittelholzer, “Welch’s bound and sequence sets
for code division multiple-access systems,” in Sequences II: Methods in
Communication, Security, and Computer Science. New York: Springer-
Verlag, 1993, pp. 63–78.
[17] J. H. Conway, R. H. Hardin, and N. J. A. Sloane, “Packing lines, planes,
etc.: Packings in Grassmannian spaces,” Experimental Math., vol. 5,
no. 2, pp. 139–159, 1996.
[18] D. Agrawal, T. J. Richardson, and R. L. Urbanke, “Multiple-antenna
signal constellations for fading channels,” IEEE Trans. Inform. Theory,
vol. 47, no. 6, pp. 2618–2626, Sept. 2001.
[19] N. N. Chan and K.-H. Li, “Diagonal elements and eigenvalues of a real
symmetric matrix,” J. Math. Anal. Appl., vol. 91, pp. 562–566, 1983.
[20] N. Aronszajn, Introduction to the Theory of Hilbert Spaces, Vol. I.
Stillwater, OK: Research Foundation of Oklahoma A & M College,
1950.
[21] S. Kayalar and H. Weinert, “Error bounds for the method of alternating
projections,” Math. Control Signal Systems, pp. 43–59, 1988.
[22] J. de Leeuw, Information Systems and Data Analysis. Berlin: Springer,
1994, ch. Block relaxation algorithms in statistics, pp. 308–325.
[23] F. Deutsch, Best Approximation in Inner-Product Spaces. New York:
Springer-Verlag, 2001.
5
