ON THE FEASIBILITY OF FULL-DUPLEX RELAYING

IN THE PRESENCE OF LOOP INTERFERENCE

Taneli Riihonen, Stefan Werner, Risto Wichman, and Eduardo Zacarias B.
SMARAD CoE and Department of Signal Processing and Acoustics
Helsinki University of Technology
P.O. Box 3000, FI-02015 TKK, Finland
ABSTRACT
We consider a two-hop infrastructure-based relay link with
three transmit power allocation methods. Both amplify-and-
forward and decode-and-forward protocols are studied. Fun-
damentally, a relay can operate either in a full-duplex or in a
half-duplex mode. The performance in full-duplex relaying is
degraded due to loop interference from the relay output to the
relay input. On the other hand, the half-duplex mode elim-
inates the loop interference, but this reduces the end-to-end
rate. In this paper, we study the feasibility of the full-duplex
mode in terms of end-to-end capacity. In particular, we evalu-
ate break-even loop interference levels, below which the full-
duplex mode outperforms the half-duplex mode. The analysis
shows that the full-duplex mode is feasible in the presence of
reasonably high loop interference power and can offer higher
capacity than the half-duplex mode.
1. INTRODUCTION
Wireless relays offer a viable, cost-effective solution for pro-
viding indoor or cell-edge coverage in cellular networks. In
this context, relays are infrastructure-based transceivers that
facilitate multihop communication by forwarding signals. Pro-
cessing in relays is usually classied into amplify-and-for-
ward (AF) and decode-and-forward (DF) protocols [1]. The
analysis presented in this paper covers both protocols.
Another classication divides the relay operation into full-
duplex (FD) and half-duplex (HD) relaying modes. In the HD
mode, the relay is restricted to receive and transmit on or-
thogonal (in time or in frequency) channels. The FD mode
requires only one channel for the end-to-end transmission. On
the other hand, the FD mode introduces loop interference due
to signal leakage between the relay output and input. Most
papers that consider mobile relays or cooperative communica-
tion (such as [1] and its successors) resort to the HDmode, be-
cause small portable devices operating in the FD mode would
suffer from overwhelming loop interference.
Contrary to mobile relays, xed infrastructure-based re-
lays can signicantly reduce the loop interference power by
utilizing two spatially-separated antennas (one for receiving
and the other for transmitting) [24]. In a typical scenario,
a highly directive backhaul antenna is placed on the roof of
a building and pointed towards the base station possibly with
a line-of-sight connection. The service antenna is placed on
the street-level or indoors with wider beam pattern towards
the covered area while pointing away from the backhaul an-
tenna. The interference power can then be maintained at a
much lower level than in mobile relays due to increased dis-
tance and physical obstacles between the antennas, antenna
polarization and directivity, or interference removal by using
various signal processing algorithms [57].
In order to study the feasibility of the FDmode in the pres-
ence of residual loop interference, we consider the follow-
ing trade-off: should loop interference be tolerated in the FD
mode, or should it be eliminated by resorting to the HD mode
at the expense of reduced end-to-end rate? This trade-off can
be investigated by comparing the capacities of the two modes.
To the best of our knowledge, such comparison has not been
previously considered in literature. In particular, we are in-
terested in the range of loop interference levels where the FD
mode outperforms the HD mode. This range can be deter-
mined by the break-even loop interference level for which the
two modes render the same capacity.
In the following, we rst derive instantaneous end-to-end
capacity expressions. Thereafter, we showthat transmit power
optimization can alleviate the effect of the loop interference.
We propose three power allocation schemes and solve for the
optimal transmit powers. For all power allocation schemes
and both AF and DF protocols, we provide closed-form ex-
pressions for the break-even loop interference level.
Our analysis shows that the FDmode is feasible also in the
presence of strong loop interference. Specically, below the
break-even level, the FD mode can offer signicant capacity
improvement over the HDmode. As can be expected, the loop
interference degrades the performance of the FD mode, but in
some scenarios the loss is insignicant when compared to the
end-to-end rate reduction of the HD mode. In general, the
presented study provides valuable information on the effect
of loop interference, which is useful for system design and
network planning.
2. SYSTEM MODEL
We consider a two-hop communication system in which a
source (S) transmits a continuous signal to a destination (D)
via a relay (R), as illustrated in Fig. 1. It is assumed that the
direct signal from the source to the destination is suppressed
due to strong attenuation and relatively low transmit power.
In the FD mode, the relay input signal is degraded due to loop
interference from the relay output which is prevented in the
HD mode by receiving and transmitting on separate channels.
S R D

SR

LI

RD
Fig. 1. Two-hop relay link with potential loop interference.
The normalized
1
transmit powers in the source and in the
relay are denoted by p
S
and p
R
, respectively. The normal-
ized
1
channel gain factors
SR
,
LI
, and
RD
represent the
respective frequency-at sourcerelay, residual loop interfer-
ence, and relaydestination channels. The signal model is
omitted here for brevity as the end-to-end capacities can be
determined from straightforward modications to the capac-
ity expressions available in literature for the HD mode.
The signal-to-interference and noise ratio of the rst hop
is
R
=
p
S

SR
p
R

LI
+1
and the signal-to-noise ratio of the second
hop is
D
= p
R

RD
. Thus, by exploiting [1, Eqs. 12, 13, and
15] and modifying the signal models thereof, the end-to-end
capacities in the full-duplex mode are expressed as
C
AF
FD
= log
2
_
1 +
p
S

SR
p
R

LI
+1
p
R

RD
p
S

SR
p
R

LI
+1
+ p
R

RD
+ 1
_
, (1)
C
DF
FD
= log
2
_
1 + min
_
p
S

SR
p
R

LI
+ 1
, p
R

RD
__
(2)
for amplify-and-forward (AF) and decode-and-forward (DF)
protocols, respectively.
Relaying in the HD mode is also described by the above
model, except that the loop interference is eliminated and
transmission consumes twice the channel resources of the full-
duplex mode. Thus, the end-to-end capacities in the half-
duplex mode can be determined from (1), (2) by setting
LI
=
0 and dividing the expressions by two:
C
AF
HD
=
1
2
log
2
_
1 +
p
S

SR
p
R

RD
p
S

SR
+ p
R

RD
+ 1
_
, (3)
C
DF
HD
=
1
2
log
2
(1 + min {p
S

SR
, p
R

RD
}) . (4)
1
Normalization means here that the channel gains are dened relatively
to receiver noise power, and that all signal amplication and attenuation is
embedded in the channel gains such that p
S
+ p
R
2.
3. FEASIBILITY OF THE FULL-DUPLEX MODE
In this section, we evaluate the range of loop interference
levels for which the full-duplex mode outperforms the half-
duplex mode in terms of capacity.
Trivially, C

FD
= 2C

HD
with protocol {AF, DF} if

LI
= 0. Furthermore, C

FD
/C

HD
is continuous and monoton-
ically decreasing in terms of
LI
, and lim

LI

FD
/C

HD
=
0. Thus, there exists a break-even loop interference level

LI
=

LI
for which C

FD
= C

HD
. For all loop interference
levels below the break-even level

LI
, the FD mode results in
better capacity than the HD mode.
Analytically, we dene the break-even loop interference
level

LI
with protocol {AF, DF} as the loop interference
channel gain that satises the statement
C

FD
C

HD
if and only if
LI

LI
. (5)
In the following, we propose three power allocation (PA)
methods and derive closed-form solutions for the correspond-
ing power optimization problems. Previously, similar power
allocation methods have been studied only for the HD mode,
for example, in [8,9]. With each PA method and both relaying
protocols, we determine closed-form expressions for

LI
.
3.1. Uniform power allocation
Let us rst consider transmission without transmit power opti-
mization. The system uses uniform power allocation given by
p
S
= p
R
= 1 with both modes and protocols in this section.
The advantage of this approach is that channel state informa-
tion is not needed.
The break-even loop interference level

LI
for protocol
{AF, DF} is determined by solving
LI
in terms of
SR
and
RD
from the inequality C

FD
C

HD
with (1)(4). Thus,
(5) is true when

AF
LI
=
_

SR
+ 1

RD
+ 1
(
SR
+
RD
+ 1) (6)
with the amplify-and-forward protocol and

DF
LI
=

SR
_
_
min{
SR
,
RD
} + 1 + 1
_
min{
SR
,
RD
}
1 (7)
with the decode-and-forward protocol. We note that

LI
1
for {AF, DF}. Therefore, C

FD
C

HD
for all
SR
and

RD
if
LI
1.
Furthermore, we notice with the decode-and-forward pro-
tocol that C
DF
FD
= 2C
DF
HD
if and only if
LI


SR

RD
1 or

LI
= 0. This means that performance is the same as without
loop interference when the loop interference power is reason-
ably low or when
SR

RD
. Consequently, if
SR
<
RD
then loop interference degrades performance with all
LI
> 0.
3.2. Power allocation with individual constraints
Transmit power optimization is motivated by the fact that the
end-to-end performance is basically limited by the weakest of
the two hops as seen from (1)(4). Thus, if the limiting factor
in the FD mode is the rst hop due to excessive loop interfer-
ence power, the end-to-end performance can be improved by
decreasing the relay transmit power. This reasoning leads to
optimization of p
S
and p
R
under individual constraints, which
is formulated as
(p

S
, p

R
) = arg max
(p
S
,p
R
)
C

subject to 0 p
S
1 and 0 p
R
1
(8)
for protocol {AF, DF} and mode {FD, HD}. In the
rest of the paper, we assume that necessary ideal channel state
information is available for determining p

S
and p

R
.
In the HD mode the solution to (8) with both protocols is
trivially given by p

S
= p

R
= 1, because C
AF
HD
and C
DF
HD
are
monotonically increasing in terms of both p
S
and p
R
. Simi-
larly p

S
= 1 in the FD mode since C
AF
FD
and C
DF
FD
are mono-
tonically increasing in terms of p
S
with any p
R
. Then, by
nding the maxima of the capacity expressions (1), (2) for
0 p
R
1, the optimal relay transmit power in the FD
mode becomes
p

R
=

min
_
1,
_

SR
+1

RD

LI
_
with AF,
min
_
1,
1
2
LI
__
4
SR

LI

RD
+ 1 1
__
with DF.
Especially, we note that the transmit power constraint is in-
voked (p

R
= 1) if
LI


SR
+1

RD
with the AF protocol or

LI


SR

RD
1 with the DF protocol.
To satisfy condition (5) for the AF protocol,
LI
is solved
from the inequality C
AF
FD
C
AF
HD
with the substitution of the
optimal transmit powers into (1), (3). Thereby, the break-even
loop interference level for the AF protocol is given by

AF
LI
=
SR

RD
_
2 +
1
A
+
1

SR
2
_
(1 +
1
A
)(1 +
1

SR
)
_
(9)
where A =
_
1 +
SR

RD
/(
SR
+
RD
+ 1) 1 if

(
SR
+ 1)(
SR
+
RD
+ 1)

RD
+ 1


SR
+ 1

RD
(10)
and otherwise
AF
LI
is given by (6) due to the transmit power
limitation (p

R
= 1). Now
AF
LI
1 for all
SR
and
RD
.
Likewise, the break-even loop interference level for the
DF protocol is determined by solving
LI
in terms of
SR
and

RD
from the inequality C
DF
FD
C
DF
HD
with (2), (4), yielding

DF
LI
=
_

SR

_
min{
SR
,
RD
} + 1 + 1
_

RD
_
_
min{
SR
,
RD
} + 1 1
_
2
. (11)
In particular, we note that
DF
LI
2 for all
SR
and
RD
.
3.3. Power allocation with a sum constraint
From the system perspective, it may be reasonable to impose
a constraint on the total transmit power of the system as some-
times assumed in related literature. By optimizing p
S
and p
R
with a sum constraint, the transmit power that is saved on one
hop can be used for improving the other hop. The optimiza-
tion problem is formulated as
(p

S
, p

R
) = arg max
(p
S
,p
R
)
C

subject to 0 p
S
, 0 p
R
, and p
S
+ p
R
2
(12)
for protocol {AF, DF} and mode {FD, HD}.
It is easy to verify that the optimal solution to (12) lies at
the boundary of the feasible region, i.e., p

S
+ p

R
= 2 with
all protocols and modes. Then, by nding the maxima of the
capacity expressions (1), (2) in terms of p
R
with substitution
p
S
= 2p
R
, the optimal relay transmit power in the FD mode
becomes
p

R
=

2
SR
+1

2
SR
+1+

(2
RD
+1)(2
LI
+1)
with AF,
4
SR

SR
+
RD
+

(
SR
+
RD
)
2
+8
SR

RD

LI
with DF.
The optimal transmit powers in the HD mode are obtained by
substituting
LI
= 0 into above expressions. Finally, with
both protocols and both modes, the optimal transmit power in
the source is given by p

S
= 2 p

R
.
The break-even loop interference level for the AF protocol
is determined by solving
LI
in terms of
SR
and
RD
from
the inequality C
AF
FD
C
AF
HD
with (1), (3) using the optimal
transmit powers, which yields

AF
LI
=
SR
+
RD
+ 2
SR

RD
(2 +
1
A
)
2
SR

RD
_
(1 +
1
A
)(2 +
1

SR
)(2 +
1

RD
),
(13)
where
A =

1 +
2
SR

RD
1 +
SR
+
RD
+
_
(2
SR
+ 1)(2
RD
+ 1)
1.
For the decode-and-forward protocol, similar calculations re-
sult in the break-even loop interference level given by

DF
LI
=

SR
+
RD
1
_

SR
+
RD

SR
+
RD
+2
SR

RD
. (14)
Especially, we see that
AF
LI
2(2

2) 0.69dB and

DF
LI
4 for all
SR
and
RD
.
3.4. Discussion
The calculated break-even loop interference levels, given by
(6), (7), (9), (11), (13), and (14), are illustrated Fig. 2. Fur-
thermore, the relative capacity improvement available by us-
ing the FD mode instead of the HD mode is exemplied in
Fig. 3 using (1)(4) in a scenario where
LI
= 6dB.
1
1
3
3
5
5
7
7
9
9
1
1
1
1
1
3
1
3
1
5
1
5
1
7
1
7
1
9
1
9
2
1
2
1
2
3
2
5
2
7
2
9
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(a) AF and uniform PA.
1
3
3
5
5
7
7
9
9
9
1
1
1
1
11
1
3
1
3
13
1
5
1
5
15
1
7
1
7
17
1
9
1
9
19
19
2
1
2
1
21
21
23
2
3
2
3
2
5
2
5
2
5
2
7
2
9
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(b) AF and PA with individual constraints. Maxi-
mum power is used (p

R
= 1) in the shaded region.
2
3
5
5
7
7
9
9
1
1
1
1
11
1
3
1
3
13
1
5
1
5
15
1
7
1
7
17
1
9
19
19
2
1
2
1
21
21
2
3
2
3
23
23
25
2
5
2
7
2
9
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(c) AF and PA with a sum constraint.
1
1
3
3
3
5
5
5
7
7
7
9
9
9
1
1
1
1
1
1
1
3
1
3
1
3
1
5
1
5
1
5
1
7
1
7
1
7
1
9
1
9
2
1
2
1
2
3
2
3
2
5
2
5
2
7
2
7
2
9
3
1
3
3
3
5
3
7
3
9
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(d) DF and uniform PA.
5
5
7
7
9
9
9
1
1
11
1
1
1
3
1
3
1
3
1
5
15
1
5
1
7
1
7
17
1
9
1
9
19
1
9
2
1
2
1
21
2
1
2
3
2
3
23
2
3
2
5
2
5
25
2
5
2
7
2
7
27
27
2
7
2
9
2
9
29
29
2
9
31
3
1
3
1
3
1
3
3
3
3
3
5
3
5
3
7
3
7
3
9
3
9
4
1
4
1
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(e) DF and PA with individual constraints.
7
9
9
1
1
11
1
3
13
1
5
1
5
1
5
1
7
17
1
7
1
9
1
9
1
9
2
1
21
2
1
2
3
2
3
23
2
3
2
5
2
5
25
2
5
2
7
2
7
27
2
7
2
9
2
9
29
2
9
31
3
1
3
1
31
33
3
3
3
5
3
5
3
7
3
7
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(f) DF and PA with a sum constraint.
Fig. 2. Contour plots for the break-even loop interference level
LI
[dB]: if
LI

LI
then C
FD
C
HD
.
The numerical results in Fig. 2 show that the FD mode
is feasible in the presence of reasonably strong loop interfer-
ence. The FD mode can also offer signicant capacity im-
provement below the break-even level as shown in Fig. 3. The
HD mode is better than the FD mode only when both hops are
weak or the loop interference power is very high.
The DF protocol is more robust to loop interference than
the AF protocol. This is because the DF protocol avoids ac-
cumulation of the loop signal due to signal regeneration. For
the same reason, with the DF protocol, the loop interference
does not degrade the performance at all when comparing to
the ideal case without loop interference if
LI


SR

RD
1.
The value of
SR
is critical with uniform PA. Especially,
when
RD

SR
, the break-even loop interference levels
are low due to the excessive transmit power usage. Hence,
the end-to-end capacity can be improved by applying the PA
methods. Especially, we see that (10) is satised for all
SR
if

RD
> (1+

5)/2 2.09dB. Thus, optimal PAwith individ-

ual constraints can increase the break-even loop interference
level in all practical cases, when compared to uniform PA.
The difference between PA with individual constraints and
with a sum constraint is not large, although both the FD mode
and the HD mode benet from PA with a sum constraint.
4. CONCLUSION
The decision between relaying in a full-duplex mode or in
a half-duplex mode is a fundamental capacity trade-off prob-
lem: the end-to-end transmission needs to tolerate either some
loop interference or consume extra channel resources to elim-
inate it completely. In this paper, we evaluated the break-
even loop interference level in a full-duplex relay link. The
full-duplex mode offers capacity improvement over the half-
duplex mode, if the loop interference is below the break-even
level. Finally, we showed with examples that the full-duplex
mode is a feasible alternative to the conventional half-duplex
mode in some setups, because the full-duplex mode can tol-
erate reasonably high loop interference power or offer higher
capacity than the half-duplex mode.
0
.
4
5
0
.
4
5
0
.
5
0
.
5
0
.
6
0
.
6
0
.
7
0
.
7
0
.
8
0
.
8
0
.
9
0
.
9
1
1
1
.
1
1
.
1
1
.
2
1
.
2
1
.
3
1
.
3
1
.
4
1
.
4
1
.
5
1
.
5
1
.
6
1
.
6
1
.
7
1
.
7
1
.
8
1
.
8
1
.
9
1
.
9
1
.
9
5
1
.
9
8
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(a) AF and uniform PA.
0
.
4
5
0
.
5 0
.
6
0
.
7
0.7
0
.
8
0.8
0
.
9
0.9
1
1
1
.
1
1.1
1
.
2
1
.2
1
.
3
1
.
3
1.3
1
.
4
1.4
1.4
1
.
5
1.5 1.5
1
.
6
1
.
6
1.6
1.6
1
.
7
1
.
7
1.7
1.7
1
.
8
1
.
8
1
.
9
1
.
9
1
.
9
5
1
.
9
8
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(b) AF and PA with individual constraints. Maxi-
mum power is used (p

R
= 1) in the shaded region.
0
.6
0
.7
0
.8
0
.
9
0.9
1
1
1
.
1
1.1
1
.
2
1.2
1
.
3
1.3
1
.
4
1
.4
1.4
1
.
5
1.5
1.5
1
.
6
1
.6
1.6
1
.
7
1
.
7
1.7
1.7
1.8
1.8
1
.
8
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(c) AF and PA with a sum constraint.
0
.
4
5
0
.
4
5
0
.
5
0
.
5
0
.6
0
.
6
0
.
7
0
.
7
0
.
8
0
.
8
0
.9
0
.
9
1
1
1
1
.1
1
.
1
1
.
1
1
.2
1
.
2
1
.
2
1
.3
1
.
3
1
.
3
1
.4
1
.
4
1
.
4
1
.5
1
.
5
1
.
5
1
.7
5
1
.7
5
1
.7
5
2
2
2
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(d) DF and uniform PA.
1
.9
1
.9
1
.9
5
1
.9
5
1
.9
8 1
.9
9 1
.9
9
5
0
.91 1
1
.
1
1
.1
1
.2
1.2
1
.3
1
.
3
1
.3
1
.4
1
.
4
1
.4
1
.5
1
.
5
1
.5
1
.5
1
.7
5
1
.
7
5
1
.
7
5
1
.7
5
1
.7
5
1
.7
5
2
2
2
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(e) DF and PA with individual constraints. Maxi-
mum power is used (p

R
= 1) in the shaded region.
1
.1
1
.
2
1
.2
1
.
3
1
.3
1
.
4
1.4
1
.
5
1.5
1
.5
1
.
6
1
.
6
1
.6
1
.
7
1
.
7
1.7
1
.7
1
.
8
1
.
8
1.8
1
.8
1
.8
1
.
9
1
.
9
1
.9
1
.9
1
.9
5
1
.9
5
1
.9
5
1
.
9
5
1
.9
8
1
.
9
8
10 5 0 5 10 15 20 25 30
10
5
0
5
10
15
20
25
30

SR
[dB]

R
D
[
d
B
]
(f) DF and PA with a sum constraint.
Fig. 3. Contour plots for the capacity ratio
C
FD
C
HD
when
LI
= 6dB in the full-duplex mode.
C
FD
C
HD
2 for all
SR
and
RD
.
5. REFERENCES
[1] J. N. Laneman, D. N. C. Tse, and G. W. Wornell, Cooperative
diversity in wireless networks: Efcient protocols and outage
behavior, IEEE Transactions on Information Theory, vol. 50,
no. 12, pp. 30623080, Dec. 2004.
[2] W. T. Slingsby and J. P. McGeehan, A high-gain cell en-
hancer, in Proc. IEEE 42nd Vehicular Technology Conference,
May 1992, pp. 756758.
[3] W. T. Slingsby and J. P. McGeehan, Antenna isolation mea-
surements for on-frequency radio repeaters, in Proc. 9th Inter-
national Conference on Antennas and Propagation, Apr. 1995,
pp. 239243.
[4] C. R. Anderson, S. Krishnamoorthy, C. G. Ranson, T. J. Lemon,
W. G. Newhall, T. Kummetz, and J. H. Reed, Antenna isola-
tion, wideband multipath propagation measurements, and inter-
ference mitigation for on-frequency repeaters, in Proc. IEEE
SoutheastCon, Mar. 2004, pp. 110114.
[5] H. Suzuki, K. Itoh, Y. Ebine, and M. Sato, A booster cong-
uration with adaptive reduction of transmitterreceiver antenna
coupling for pager systems, in Proc. IEEE 50th Vehicular Tech-
nology Conference, Sept. 1999, pp. 15161520.
[6] H. Hamazumi, K. Imamura, N. Iai, K. Shibuya, and M. Sasaki,
A study of a loop interference canceller for the relay stations
in an SFN for digital terrestrial broadcasting, in Proc. IEEE
Global Telecommunications Conference, Nov. 2000, pp. 167
171.
[7] K. M. Nasr, J. P. Cosmas, M. Bard, and J. Gledhill, Per-
formance of an echo canceller and channel estimator for on-
channel repeaters in DVB-T/H networks, IEEE Transactions
on Broadcasting, vol. 53, no. 3, pp. 609618, Sept. 2007.
[8] Q. Zhang, J. Zhang, C. Shao, Y. Wang, P. Zhang, and R. Hu,
Power allocation for regenerative relay channel with Rayleigh
fading, in Proc. IEEE 59th Vehicular Technology Conference,
May 2004, pp. 11671171.
[9] J. Zhang, Q. Zhang, C. Shao, Y. Wang, P. Zhang, and Z. Zhang,
Adaptive optimal transmit power allocation for two-hop non-
regenerative wireless relaying system, in Proc. IEEE 59th Ve-
hicular Technology Conference, May 2004, pp. 12131217.