Structured Lattice Codes for 2 2 2 MIMO

Interference Channel
Song-Nam Hong
Dep. of Electrical Engineering
University of Southern California
CA, USA
Email: [email protected]
Giuseppe Caire
Dep. of Electrical Engineering
University of Southern California
CA, USA
Email: [email protected]
AbstractWe consider the 2 2 2 multiple-input multiple-
output interference channel where two source-destination pairs
wish to communicate with the aid of two intermediate relays.
In this paper, we propose a novel lattice strategy called Aligned
Precoded Compute-and-Forward (PCoF). This scheme consists
of two phases: 1) Using the CoF framework based on signal
alignment we transform the Gaussian network into a deterministic
nite eld network. 2) Using linear precoding (over nite eld) we
eliminate the end-to-end interference in the nite eld domain.
Further, we exploit the algebraic structure of lattices to enhance
the performance at nite SNR, such that beyond a degree of
freedom result (also achievable by other means). We can also
show that Aligned PCoF outperforms time-sharing in a range of
reasonably moderate SNR, with increasing gain as SNR increases.
I. INTRODUCTION
In recent years, signicant progress has been made on the
understanding of the theoretical limits of wireless commu-
nication networks. In [1], the capacity of multiple multicast
network (where every destination desires all messages) is
approximated within a constant gap independent of SNR and
of the realization of the channel coefcients. Also, for multiple
ows over a single hop, new capacity approximations were
obtained in the form of degrees of freedom (DoF), generalized
degrees of freedom (GDoF), and O(1) approximations [2][4].
Yet, the study of multiple ows over multiple hops remains
largely unsolved. The 222 Gaussian interference channel
(IC) has received much attention recently, being one of the
fundamental building blocks to characterize the DoFs of two-
ows networks [5] One natural approach is to consider this
model as a cascade of two ICs. In [6], the authors apply
the Han-Kobayashi scheme [7] for the rst hop to split each
message into private and common parts. Relays can cooperate
using the shared information (i.e., common messages) for
the second hop, in order to improve the data rates. This
approach is known to be highly suboptimal at high signal-
to-noise ratios (SNRs), since two-user IC can only achieve
1 DoF. In [8], Cadembe and Jafar show that
4
3
DoF is
achievable by viewing each hop as an X-channel. This is
accomplished using the interference alignment scheme for
each hop. Recently, the optimal DoF was obtained in [9] using
aligned interference neutralization, which appropriately com-
bines interference alignment and interference neutralization.
S
1
S
2
R
1
R
2
D
1
D
2
Fig. 1. 2 2 2 MIMO Gaussian interference channel.
Also, there was the recent extension to the K K K
Gaussian IC in [10], achieving the optimal K DoF using
aligned network diagonalization.
We consider the 2 2 2 multiple-input multiple-output
(MIMO) IC as shown in Fig. 1 consisting of two sources,
two relays, and two destinations. All nodes have M multiple
antennas. Each source k has a message for its intended
destination k, for k = 1, 2. In the rst hop, a block of
n channel uses of the discrete-time complex MIMO IC is
described by
_
Y
R
1
Y
R
2
_
=
_
F
11
F
12
F
21
F
22
_ _
X
1
X
2
_
+
_
Z
R
1
Z
R
2
_
(1)
where the matrices X
k
1
and Y
R
k
contain, arranged by rows,
the k-th source channel input sequences x
k,
C
1n
for =
1, . . . , M, the k-th relay channel output sequences y
R
k,

C
1n
, and where F
jk
C
MM
denotes the channel matrix
from source k to j. In the second hop, a block of n channel
uses of the discrete-time complex MIMO IC is described by
_
Y
1
Y
2
_
=
_
G
11
G
12
G
21
G
22
_ _
X
R
1
X
R
2
_
+
_
Z
1
Z
2
_
(2)
where the matrices X
R
k
and Y
k
contain the k-th relay channel
input sequences x
R
k,
C
1n
, the k-th destination channel
output sequences y
k,
C
1n
, and where G
kj
C
MM
denotes the channel matrix between relay j and destination k.
The matrix Z
k
(or Z
R
k
) contains i.i.d. Gaussian noise samples
1
We use underline to denote the matrices whose horizontal dimension
(column index) denotes time and vertical dimension (row index) runs across
the antennas. For any matrix X, we denote x

by the -th row of X.

a
r
X
i
v
:
1
3
0
1
.
6
4
5
3
v
1


[
c
s
.
I
T
]


2
8

J
a
n

2
0
1
3
CN(0, 1). We assume that the elements of F
jk
and G
kj
are drawn i.i.d. according to a continuous distribution (i.e.,
Gaussian distribution). The channel matrices are assumed to
be constant over the whole block of length n and known to
all nodes. Also, we consider a sum-power constraint equal to
MSNR at each transmitter.
In this paper, we propose a novel lattice strategy named
Aligned Precoded Compute-and-Forward (PCoF). CoF makes
use of lattice codes so that each receiver can reliably decode a
linear combination with integer coefcients of the interfering
codewords. Thanks to the fact that lattices are modules over
the ring of integers, the linear combinations translates directly
into a linear combination of messages over a suitable nite
eld. In this way, each hop is transformed into a deterministic
noiseless nite eld IC. The end-to-end interferences in the
nite eld domain are eliminated by distributed precoding
(over nite eld) at relays. Using this framework, we char-
acterize a symmetric sum rate and prove that 2M 1 DoF
is achievable by lattice coding (this DoF result is proven in
[9] without resorting to CoF and lattice coding). Further, we
use the lattice codes algebraic structure in order to obtain
also good performance at nite SNRs. We use integer-forcing
receiver (IFR) of [11] in order to minimize the impact of noise
boosting at the receivers, and integer-forcing beamforming
(IFB), proposed by the authors in [12], in order to minimize the
power penalty at the transmitters. We provide numerical results
showing that Aligned PCoF outperforms time-sharing even at
reasonably moderate SNR, with increasing performance gain
as SNR increases.
II. PRELIMINARIES
In this section we provide some basic denitions and results
that will be extensively used in the sequel.
A. Nested Lattice Codes
Let Z[j] be the ring of Gaussian integers and p be a prime.
Let denote the addition over F
q
with q = p
2
, and let g :
F
q
C be the natural mapping of F
q
onto {a + jb : a, b
Z
p
} C. We recall the nested lattice code construction given
in [13]. Let = { = zT : z Z
n
[j]} be a lattice in C
n
,
with full-rank generator matrix T C
nn
. Let C = {c =
wG : w F
r
q
} denote a linear code over F
q
with block length
n and dimension r, with generator matrix G. The lattice
1
is dened through construction A (see [14] and references
therein) as

1
= p
1
g(C)T+ , (3)
where g(C) is the image of C under the mapping g (applied
component-wise). It follows that
1
p
1
is a chain
of nested lattices, such that |
1
/| = p
2r
and |p
1
/
1
| =
p
2(nr)
.
For a lattice and r C
n
, we dene the lattice quantizer
Q

(r) = argmin

r
2
, the Voronoi region V

=
{r C
n
: Q

(r) = 0} and [r] mod = r Q

(r). For
and
1
given above, we dene the lattice code L =
1
V

with rate R =
1
n
log |L| =
r
n
log q. Construction A provides
a natural labeling of the codewords of L by the information
messages w F
r
q
. Notice that the set p
1
g(C)T is a system
of coset representatives of the cosets of in
1
. Hence, the
natural labeling function f : F
r
q
L is dened by f(w) =
p
1
g(wG)T mod .
B. Compute-and-Forward
We recall here the CoF scheme of [13]. Consider a 2-user
Gaussian multiple access channel (MAC) with M antennas at
each transmitter and at the receiver, represented by
Y = HCX+Z (4)
where H C
MM
, C Z[j]
M2M
, and Z contains i.i.d.
Gaussian noise samples CN(0, 1). This particular form of
channel matrix HC will be widely considered in this paper,
as a consequence of signal alignment. All users make use
of the same nested lattice codebook L =
1
V

, where
has second moment
2

1
nVol(V

)
_
V

r
2
dr = SNR
eff
.
Each user k encodes M information messages w
k,
F
r
q
into
the corresponding codeword t
k,
= f(w
k,
) and produces its
channel input according to
x
k,
= [t
k,
+d
k,
] mod , (5)
for = 1, . . . , M, where the dithering sequences d
k,
s
mutually independent across the users and the messages,
uniformly distributed over V

, and known to the receiver.

Notice that x
k,
is the -th row of X
k
, k = 1, 2. The
decoders goal is to recover an integer linear combination
s = [b
H
CT] mod , with some integer vector b Z[j]
M1
.
Since
1
is a Z[j]-module (closed under liner combinations
with Gaussian integer coefcients), then s L. Letting s
be decoded codeword, we say that a computation rate R is
achievable for this setting if there exists sequences of lattice
codes L of rate R and increasing block length n, such that the
decoding error probability satises lim
n
P(s = s) = 0.
In the scheme of [13], the receiver computes
y =
_

H
Yb
H
CD

mod
= [s +z
eff
(HC, b, )] mod (6)
where C
M1
and z
eff
(HC, b, ) = (
H
H b
H
)CU+

H
Z denotes the effective noise, including the non-integer
self-interference (due to the fact that
H
H / Z[j]
1M
in
general) and the additive Gaussian noise term. The scaling,
dither and modulo- operation in (6) is referred to as the
CoF receiver mapping. Choosing
H
= b
H
H
1
, the variance
of the effective noise is given by
2
exact
= (H
1
)
H
b
2
. This
scheme is known as exact IFR [11]. In this way, the non-
integer penalty of CoF is completely eliminated. More in
general, the performance can be improved especially at low
SNR by minimizing the variance of z
eff
(H, b, ) with respect
to . In this case, we obtain:

2
(HC, b) = b
H
C(SNR
1
eff
I +C
H
H
H
HC)
1
C
H
b.(7)
Since is uniquely determined by HC and b, it will be
omitted in the following, for the sake of notation simplicity.
Source 1
Source 2 "not used"
"not used"
Precoding
Relay 2
Relay1 Destination1
Destination 2
Fig. 2. A deterministic noise-free 222 nite eld interference channel.
From [13], we know that by applying lattice decoding to y
given in (6) the following computation rate is achievable:
R(H, b, SNR
eff
) = log
+
(SNR
eff
/
2
(HC, b)) (8)
where log
+
(x) max{log(x), 0}. Also, the receiver can
reliably decode M linear combinations S = [BCT] mod
with integer coefcient vectors {b
H

: = 1, . . . , M} (i.e., the
-th row of B) if
R min

{R(H, b

, SNR
eff
)} R(H, B, SNR
eff
). (9)
Using the lattice encoding linearity, the corresponding M
linear combinations over F
q
for the messages are obtained
as
U = g
1
([B] mod pZ[j])g
1
([C] mod pZ[j])W
(a)
= [B]
q
[C]
q
W, (10)
where we use the notation [B]
q
g
1
([B] mod pZ[j]).
III. ALIGNED PCOF
In this section, we propose a novel lattice strategy called
Aligned Precoded CoF (PCoF). This consists of two phases:
1) The CoF framework transforms a Gaussian network into
a nite eld network. 2) A linear precoding scheme is used
over nite eld to eliminate the end-to-end interferences (see
Fig. 2). While using the CoF framework, the main perfor-
mance bottleneck consists of the non-integer penalty, which
ultimately limits the performance of CoF scheme in the high
SNR regime [15]. To overcome this bottleneck, we employ
signal alignment in order to create an aligned channel matrix
for which exact integer forcing is possible, as seen in Section
II-B. Namely, we use alignment precoding matrices V
1
and
V
2
at the two sources such that
_
F
k1
V
1
F
k2
V
2

= H
R
k
C
R
k
, (11)
where H
R
k
C
MM
and C
R
k
Z[j]
M2M
. However, the
precoding over Cmay produce a power penalty due to the non-
unitary nature of the alignment matrices, and this can degrade
the performance at nite SNR. In order to counter this effect,
we use the concept of IFB (see [12]). The main idea is that
V
k
can be pre-multiplied (from the oft) by some appropriately
chosen full-rank integer matrix A
k
since its effect can be
undone by precoding over F
q
, using [A
k
]
q
. Then, we can
optimize the integer matrix in order to minimize the power
penalty. The detailed procedures of Aligned PCoF are given
in the following sections.
A. CoF framework based on signal alignment
In this section we show how to turn any 2-user MIMO IC
into a noiseless nite eld IC using the CoF framework. We
focus on the rst hop of our 222 network since the same
scheme is straightforwardly used for the second hop. Consider
the MIMO IC in (1). Let {w
1,
F
r
q
: = 1, . . . , M} denote
the messages of source 1 and {w
2,
F
r
q
: = 1, . . . , M1}
denote the messages of source 2. All transmitters make use of
the same nested lattice codebook L =
1
V

, where
has the second moment
2

= SNR
eff
. Also, we let V
1
=
[v
1,1
v
1,M
] C
MM
and V
2
= [v
2,1
v
2,M1
]
C
MM1
denote the precoding matrices used at sources 1
and 2, respectively. They are chosen to satisfy the alignment
conditions, given by
F
11
v
1,k+1
= F
12
v
2,k
(12)
F
21
v
1,k
= F
22
v
2,k
(13)
for k = 1, . . . , M 1. The feasibility of the above conditions
was shown in [9] for any M.
Let A
1
Z[j]
MM
and A
2
Z[j]
M1M1
denote the
full rank integer matrices. They will be optimized to minimize
the power penalty in Section IV. Each source k precodes its
messages over F
q
as
W

k
= [A
k
]
1
q
W
k
, k = 1, 2. (14)
Then, the precoded messages are encoded using the nested
lattice codes. Finally, the channel input sequences are given
by the rows of:
X
k
= V
k
A
k
T
k
= V
k
T

k
, (15)
where t
k,
= [f(w

k,
) + d
k,
] mod , and where T

k
=
A
k
T
k
. Due to the power constraint, each source k must satisfy
SNR
eff
tr(V
k
A
k
A
H
k
V
H
k
) MSNR. (16)
Since sources use the same nested lattice codes, we can
choose:
SNR
eff
= min{SNR(V
k
, A
k
) : k = 1, 2} (17)
where SNR(V
k
, A
k
) = MSNR/tr(V
k
A
k
A
H
k
V
H
k
).
The decoding procedure is as follows. We rst consider the
aligned received signals. Relay 1 observes
Y
R
1
= F
11
X
1
+F
12
X
2
+Z
R
1
(a)
= F
11
V
1
. .
H
R
1
_

_
t

1,1
t

1,2
+t

2,1
.
.
.
t

1,M
+t

2,M1
_

_
+Z
R
1
(18)
= H
R
1
C
R
1
_
T
1
T
2
_
+Z
R
1
(19)
where (a) follows from the fact that the precoding vec-
tors satisfy the alignment conditions in (13) and C
R
1
=
[ A
1
C
12
A
2
] with
C
12
=
_
0
1M1
I
M1M1
_
. (20)
Similarly, relay 2 observes the aligned signals:
Y
R
2
= F
21
V
1
. .
H
R
2
_

_
t

1,1
+t

2,1
.
.
.
t

1,M1
+t

2,M1
t

1,M
_

_
+Z
R
2
(21)
= H
R
2
C
R
2
_
T
1
T
2
_
+Z
R
2
(22)
where C
R
2
= [ A
1
C
22
A
2
] with
C
22
=
_
I
M1M1
0
1M1
_
. (23)
The channel matrices in (19) and (22) follows the form
considered in Section II-B. Following the CoF framework in
(9) and (10), if R R(H
R
k
C
R
k
, B
R
k
, SNR
eff
), the relay k can
decode the M linear combinations with full-rank coefcients
matrix B
R
k
:
U
k
= [B
R
k
]
q
[C
R
k
]
q
_
W

1
W

2
_
= [B
R
k
]
q
_
[A
1
]
q
[C
k2
]
q
[A
2
]
q

_
W

1
W

2
_
(a)
= [B
R
k
]
q
_
I
MM
[C
k2
]
q

_
W
1
W
2
_
where (a) is due to the precoding over F
q
in (14). Let

W
1
=
[B
R
1
]
1
q
U
1
and

W
2
be the rst M 1 rows of [B
R
2
]
1
q
U
2
.
Then, we can dene the deterministic noiseless nite eld IC:
_

W
1

W
2
_
= Q
_
W
1
W
2
_
(24)
where the so-called system matrix is obtained using C
1
and
C
2
as
Q =
_
I
MM
Q
12
Q
21
I
M1M1
_
(25)
where
Q
12
=
_
0
1M1
I
M1M1
_
, Q
21
= [ I
M1M1
0
M11
].
B. Linear precoding over deterministic network
Eq. (24) denes the rst hop of the noiseless nite eld IC.
Next, we focus on the second hop. The relay k uses a precoded
version of decoded linear combinations W
R
k
= M
k

W
k
as its
messages. Operating in a similar way as for the rst hop, the
second hop noiseless nite eld IC is given by
_

W
R
1

W
R
2
_
= Q
_
W
R
1
W
R
2
_
. (26)
Concatenating (24) and (26), the end-to-end nite eld noise-
less network is described by
_

W
R
1

W
R
2
_
= Q
_
M
1
0
0 M
2
_
Q
_
W
1
W
2
_
. (27)
Lemma 1 shows that the linear combinations decoded at desti-
nation 1 are equal to its desired messages and are equal to the
messages with a change of sign (multiplication by 1 in the
nite eld) at destination 2 (see Fig. 2). Notice that the system
matrix is xed and independent of the channel matrices, since
it is determined only by the alignment conditions.
Lemma 1: Choosing precoding matrices M
1
and M
2
as
M
1
= (I
MM
(Q
12
Q
21
))
1
(28)
M
2
= (I
M1M1
(Q
21
Q
12
))
1
(29)
the end-to-end system matrix becomes a diagonal matrix:
Q
_
M
1
0
0 M
2
_
Q =
_
I
MM
0
0 I
M1M1
_
.
Proof: See the long version of this paper [16].
Based on the above, we proved the following:
Theorem 1: For the 2 2 2 MIMO IC dened in (1)
and (2), Aligned PCoF can achieve the symmetric sum rate of
(2M 1)R with common message rate
R = min
k=1,2
{R(H
R
k
C
R
k
, B
R
k
, SNR
eff
), R(H
k
C
k
, B
k
, SNR

eff
)}
for any full rank integer matrices A
k
, A
R
k
, B
k
, B
R
k
, and any
matrices V
k
, V
R
k
to satisfy the alignment conditions in (13),
where
H
R
k
= F
k1
V
1
, H
k
= G
k1
V
R
1
SNR
eff
= min{SNR(V
k
, A
k
) : k = 1, 2}
SNR

eff
= min{SNR(V
R
k
, A
R
k
) : k = 1, 2},
and C
R
k
= [A
1
C
k2
A
2
] and C
k
= [A
R
1
C
k2
A
R
2
] with
C
12
, C
22
in (20) and (23).
Showing that R grows as log SNR yields:
Corollary 1: Aligned PCoF achieves the 2M 1 DoF for
the 2 2 2 MIMO IC when all nodes have M multiple
antennas.
Proof: See the long version of this paper [16].
IV. OPTIMIZATION OF SYMMETRIC SUM RATES
Suppose that precoding matrices V
k
, V
R
k
are determined.
We need to optimize all integer matrices in Theorem 1 to
maximize the sum rates. First, the power-penalty optimization
problems take on the form:
argmin tr
_
VAA
H
V
H
_
=
M

=1
Va

2
subject to A is full rank (30)
where a

denotes the -th column of A. Also, the minimization

problem of variance of effective noise consists of nding an
integer matrix B solution of:
argmin max

{b
H

(SNR
1
eff
I +H
H
H)
1
b

}
subject to B is full rank (31)
where H denotes an aligned channel matrix and b
H

is the -th
row of B. By Cholesky decomposition, there exists a lower
triangular matrix L such that
b
H

(SNR
1
eff
I +H
H
H)
1
b

= L
H
b

2
. (32)
0 10 20 30 40 50 60 70
0
10
20
30
40
50
60
70
SNR [dB]
A
v
e
r
a
g
e

S
u
m

R
a
t
e
s

(
b
i
t
s

p
e
r

c
h
a
n
n
e
l

u
s
e
)
Time Sharing
Aligned PCoF (w/o opt)
Aligned PCoF 3 DoF
2 DoF
Fig. 3. Performance comparison of Aligned PCoF and time-sharing with
respect to ergodic symmetric sum rates.
We notice that problem (30) (or (31)) is equivalent to nding
a reduced basis for the lattice generated by V (or L
H
). In
particular, the reduced basis takes on the form VUwhere Uis
a unimodular matrix over Z[j]. Hence, choosing A = Uyields
the minimum power-penalty subject to the full rank condition
in (30). In practice we used the (complex) LLL algorithm,
with renement of the LLL reduced basis approximation by
Phost or Schnorr-Euchner lattice search (see the long version
for details [16]).
A. Numerical Results: Rayleigh fading
We evaluate the performance of Aligned PCoF in terms of
its average achievable sum rates. We computed the ergodic
sum rates by Monte Carlo averaging with respect to the
channel realizations with i.i.d. Rayleigh fading CN(0, 1).
For comparison, we considered the performance of time-
sharing where IFR is used for each M M MIMO IC.
We used the IFR since it is known to almost achieve the
performance of joint maximum likelihood receiver [11] and
has a similar complexity with Aligned PCoF. In this case, an
achievable symmetric sum rate is obtained as
R = min{R(F
kk
, B
1
, 2SNR), R(G
kk
, B
2
, 2SNR)} (33)
for any full-rank matrices B
1
and B
2
. The integer matrices
are optimized in the same manner of Aligned PCoF. Also,
in this case, we used the 2MSNR for power-constraint since
each transmitter is active on every odd (or even) time slot.
For Aligned PCoF, we need to nd precoding matrices for
satisfying the alignment condition in (13). For M = 2, the
conditions are given by
F
11
v
1,2
= F
12
v
2,1
and F
21
v
1,1
= F
22
v
2,1
. (34)
For the simulation, we used the following precoding matrices
to satisfy the above conditions:
V
1
=
_
F
1
21
F
12
1 F
1
11
F
22
1

and v
2,1
= 1.
Also, the same construction method is used for the second
hop. Since source 1 (or relay 1) transmits one more stream
than source 2 (or relay 2), the former always requires higher
transmission power. In order to efciently satisfy the average
power-constraint, the role of sources 1 and 2 (equivalently,
relays 1 and 2) is alternatively reversed in successive time
slots. In Fig. 3, we observe that Aligned PCoF can have the
SNR gain about 5 dB by optimizing the integer matrices for
IFR and IFB, comparing with simply using identity matrices.
Also, Aligned PCoF provides a higher sum rate than time-
sharing if SNR 15 dB, and its gain over time-sharing
increases with SNR, showing that in this case the DoF result
matters also at nite SNR.
REFERENCES
[1] S. Avestimehr, S. Diggavi, and D. Tse, Wireless network information
ow: A deterministic approach, IEEE Transactions on Information
Theory, vol. 57, pp. 1872-1905, Apr. 2011.
[2] V. Cadambe and S. Jafar, Interference alignment and the degrees of
freedom of the K user interference channel, IEEE Transactions on
Information Theory, vol. 54, pp. 3425-3441, Aug. 2008.
[3] T. Gou and S. A. Jafar, Capacity of a class of symmetric SIMO Gaussian
interference channels within O(1), in Proceedings of IEEE International
Symposium on Information Theory (ISIT), Seoul, Korea, Jun-Jul. 2009.
[4] S. A. Jafar and S. Vishwanath, Generalized Degrees of Freedom of the
Symmetric Gaussian K User Interference Channel, IEEE Transactions
on Information Theory, vol. 56, pp. 3297-3303, Jul. 2010.
[5] I. Shomorony and S. Avestimehr, Two-Unicast Wireless Networks: Char-
acterizing the Degrees-of-Freedom, To appear on IEEE Transactions on
Information Theory
[6] O. Simeone, O. Somekh, Y. Bar-Ness, H. V. Poor, and S. Shamai,
Capacity of Linear Two-Hop Mesh Networks with Rate Splitting,
Decode-and-Forward Relaying and Cooperation, in Proceedings of 45th
Annual Allerton Conference on Communication, Control, and Computing,
Monticello, Illinois, Sept. 26-28, 2007.
[7] T. Han and K. Kobayashi, A New Achievable Rate Region for the
Interference Channel, IEEE Transactions on Information Theory, vol.
IT-27, pp. 49-60, Jan. 1981.
[8] V. R. Cadambe and S. A. Jafar, Interference Alignment and Degrees of
Freedom of Wireless X Networks, IEEE Transactions on Information
Theory, vol. 55, pp. 3893-3908, Sept. 2009.
[9] T. Gou, S. A. Jafar, S.-W. Jeon, S.-Y. Chung, Interference Alignment
Neutralization and the Degrees of Freedom of the 2 2 2 Interference
Channel, IEEE Transactions on Information Theory, vol. 58, pp. 4381-
4395, July, 2012.
[10] I. Shomorony and S. Avestimehr, Degrees of Freedom of Two-Hop
Wireless Networks: Everyone Gets the Entire Cake, To appear in
proceedings of 2012 Allerton Conference.
[11] J. Zhan, B. Nzaer, U. Erez, and M. Gastpar, Integer-Forcing Linear
Receivers, submitted to IEEE Transactions on Information Theory.
[12] S. Hong and G. Caire, Lattice Strategy for Cooperative Distributed An-
tenna Systems, submitted to IEEE Transactions on Information Theory
2012.
[13] B. Nazer and M. Gastpar, Compute-and-Forward: Harnessing Inter-
ference through Structured Codes, IEEE Transactions on Information
Theory, vol. 57, pp. 6463-6486, Oct. 2011.
[14] U. Erez and R. Zamir, Achieving
1
2
log(1 + SNR) on the AWGN
channel with lattice encoding and decoding, IEEE Transactions on
Information Theory, vol. 50, pp. 2293-2314, Oct. 2004.
[15] U. Niesen and P. Whiting, The degrees-of-freedom of compute-and-
forward, IEEE Transactions on Information Theory, vol. 59, pp. 5214-
5232, Aug. 2012.
[16] S.-N. Hong and G. Caire, Structured Lattice Codes for Gaussian Relay
Networks, in preparation.