Anda di halaman 1dari 6

IEEE ICC 2015 - Workshop on LTE in Unlicensed Bands: Potentials and Challenges

A Spectrum Etiquette Protocol and Interference


Coordination for LTE in Unlicensed Bands (LTE-U)
Hao Song

Xuming Fang

Key Lab of Information Coding & Transmission


Southwest Jiaotong University
Chengdu 610031, China
Email: songhao992013@gmail.com

Key Lab of Information Coding & Transmission


Southwest Jiaotong University
Chengdu 610031, China
Email: xmfang@swjtu.edu.cn

AbstractLTE in unlicensed bands (LTE-U) holds the promise


of alleviating the licensed spectra scarcity and enhancing capacity
by utilizing the unlicensed spectra. However, once LTE-U accesses
an unlicensed band, there is almost no any opportunity for
unlicensed systems (e.g. 802.11 WLAN) with inferior control
abilities and CSMA/CA, to access that any more. In order to
solve this unfair coexistence problem, we propose a spectrum
etiquette protocol that restricts the priority of LTE-U and
balances the extreme unfair competition. Besides, for engineering
implementation of the proposed spectrum etiquette, an LTE802.11 fusion protocol stack is designed for LTE-U. However,
the restrictions of the spectrum etiquette would degrade the
throughput of LTE-U on unlicensed spectra. Therefore, successive
interference cancellation (SIC) is adopted to perform interference
coordination for LTE-U. Moreover, we also study the problem of
deriving the optimal lower power and the time length of each
normal power transmission allocation strategy that maximizes
the ergodic capacity of LTE-U under the proposed spectrum
etiquette. The simulation study shows that the improved ergodic
throughput of LTE-U can be achieved by our proposed interference coordination and the trafc need of 802.11 WLAN can
be satised as much as possible under the proposed spectrum
etiquette.
Index TermsSpectrum Etiquette, LTE-U, 802.11 WLAN, SIC,
Throughput Maximization.

I. I NTRODUCTION
Over the past years, the wireless trafc demand has experienced the signicant growth with the introduction of smart
phones, tablet and other new mobile devices supporting a wide
range of multimedia applications. As a result, the current capacity of LTE networks will not be able to support the demand
in the future [1]. Spectrum extension is a straightforward way
to enhance capacity, however available licensed spectrum is
limited, scarce and costly, which needs to be authorized by
governments. Despite spectrum scarcity, spectrum extension
opportunities for LTE systems can be found from license-free
frequency bands, such as Industrial, Scientic and Medical
(ISM) and Unlicensed National Information Infrastructure (UNII) bands. At present, these bands are popularly used for the
small-scale coverage devices, such as cordless phone, Bluetooth, WLAN and so on. Generally, the public communication
systems, such as LTE, are not advocated to access unlicensed
spectrum. However, we should consider that future systems
need to support huge trafc demand. In addition, a report is
published by US Federal Communication Commission (FCC),

978-1-4673-6305-1/15/$31.00 2015 IEEE

which indicates that the current xed spectrum allocation


policy has resulted in severely low utilization on unlicensed
spectrum in both time and space, even within spectrum-scarce
urban areas [2]. Therefore, LTE is possible to make use
of unlicensed spectrum, namely LTE-U Systems [3][4], to
alleviate the licensed spectra scarcity and enhance capacity,
if it obeys the appropriate spectrum etiquette. To formulate
a proper spectrum etiquette, we should rstly investigate the
coexisting problem on unlicensed spectra between LTE-U
and the existing unlicensed systems. Because except 802.11
WLAN, other unlicensed systems, such as 802.16, Bluetooth,
Ad Hoc networks, which could use the unlicensed bands
distributed from 2GHz to 6GHz, have not been widely applied
in practice, we choose 802.11 WLAN as an example of
unlicensed systems to perform analysis.
A. Analysis of Coexistence Problem between LTE-U and
802.11 WLAN
The specic Carrier Sense Multiple Access with Collision
Avoidance (CSMA/CA) mechanism used in the 802.11 MAC
is referred to distributed coordination function (DCF) [5]. The
802.11 devices that wish to transmit should perform a clear
channel assessment (CCA) by sensing the medium for a xed
duration, namely the DCF inter-frame space (DIFS). To report
that a channel is busy, CCA has two methods, signal detection
and energy detection [6]. As for the interference from LTEU, only the energy detection could work, whose sensitivity
threshold is -62dBm for 20MHz channel. Despite no enforced
spectrum rules about the upper limit of power over unlicensed
bands stated by FCC or other organizations [7], all wireless
devices still strictly comply with a maximum power constraint,
generally 1W (30dBm). Assume that the total power of the
eNB in LTE-U is 300mw over a 20MHz channel, which
is close to the power of Pico but higher to overcome the
serious path loss of high unlicensed bands. The power with
only Reference Signal transmission, which can be viewed as
the minimum power to maintain the basic link connection, is
PrRS 20M Hz = 10 log10 (300mw 16 ) 17dBm, according
to the RS distribution density (inset one every six sub-carriers)
indicated by LTE specications [8]. Assume that the carrier
frequency is 5.8GHz, and the path loss and shadowing which
are relative to distance should be taken into consideration.

2338

IEEE ICC 2015 - Workshop on LTE in Unlicensed Bands: Potentials and Challenges

According to WINNER II channel model [9], the received


signal strength from eNB against distance of receiver is shown
as Fig.1. Because the coverage of eNB is less than 2km
commonly, it is obviously that the received signal strength
from eNB is far large than CCA sensitivity threshold of
802.11, even in the edge of cell covered by eNB. In view
of CSMA/CA mechanism used in 802.11, we can obtain the
conclusion that once the eNB accesses the unlicensed bands,
even congured with the minimum power, all 802.11 devices
within the cell can not utilize those bands any more.
Received downlink signal power (dBm)

20
10
0
10
20
30
40
50

0.5

1.5

2.5

Distance from eNB (km)

Fig. 1.

Received signal strength against distance with minimum power

B. The existing spectrum etiquette


Technically speaking, spectrum etiquette is a rule to regulate
the unlicensed spectrum utilization in terms of moral level and
not a compulsory standard in terms of legal level. Some spectrum etiquette standards and research work have emerged. US
FCC published a spectrum etiquette, ruling that the transmit
power of unlicensed devices operating in 902-928 MHz band
must be lower than 1W [7]. Unfortunately, with respect to
unlicensed operations in the 2.4GHz and 5.8 GHz bands, there
is no similar requirement in this publication. As for research
work, [10][11] studied the spectrum etiquette for Cognitive
Radio (CR) systems. [10] put forward a spectrum etiquette
scheme, in which the basic idea is to enable mutual information between neighboring unlicensed operations to be shared,
aiming at avoiding the classic tragedy of the commons
effect. [11] introduced a credit-token-based spectrum etiquette
framework that enables spectrum sharing among distributed
heterogeneous CR networks with equal priority. Additionally,
a spectrum etiquette for 802.11/802.16 systems proposed in
[12][13], whose basic idea is to standardize a simple common
protocol for announcement of radio and service parameters, is
called the common spectrum coordination channel (CSCC).
It is easy to see that the current research work focuses on
the spectrum etiquette to solve the coexisting problem among
the same systems (as [10][11]) or the systems with similar
interference coordination ability (as [12][13]). Thus, the aforementioned spectrum etiquettes are not suitable for solving
the coexisting problem between LTE-U and 802.11 WLAN.
Because 802.11 WLAN has poor interference coordination
ability, that has always been a listen-before-talk protocol to

access spectra. Oppositely, LTE-U, with the strong ability of


that, could overcome the strong interference by power control,
resource allocation, FEC (forward error correction), SIC, etc.
C. Our Contributions
Currently, unlicensed systems have been widely employed
and play important roles in public services, such as public entertainment, health care, and industrial production. Moreover,
consider that 802.11 WLAN is in the absolutely inferior position of the competition with LTE-U on the 2-6GHz unlicensed
spectrum, which is the only available medium for it. Hence,
we propose a spectrum etiquette that in the usage of unlicensed
spectra, LTE-U should regard 802.11 WLAN as primary
user with higher priority. Furthermore, LTE-U can access
an unlicensed band only if it is detected not being used by
802.11 devices, or no harmful interference is caused to 802.11
devices. To implement this spectrum etiquette, an LTE-802.11
fusion protocol stack is designed in eNB side to make LTE-U
and 802.11 WLAN identify the control information of each
other, based on which LTE-U adjusts its transmission model
to adapt to the condition of unlicensed spectra. Then SIC is
employed by eNB to perform interference coordination and
preserve the communication quality of LTE-U. Besides, we
also study the problem of maximizing the ergodic throughput
of LTE-U to obtain the optimal lower power and the time
length of each normal power transmission allocation under
our propose spectrum etiquette.
The rest of the paper is organized as follows. In section II,
we design the LTE-802.11 protocol fusion stack, and introduce
the implementation of our proposed spectrum etiquette in LTEU. In section III, we develop the analytical model and study
the ergodic throughput maximization problem of LTE-U under
our proposed spectrum etiquette. Simulation studies are carried
out in Section V. Finally Section VI concludes the paper.
II. S PECTRUM E TIQUETTE AND O PERATIONS IN LTE-U
The prerequisite for implementing the proposed spectrum
etiquette is to make LTE-U and 802.11 WLAN identify the
control information of each other. However, in spite of Orthogonal Frequency Division Multiplexing (OFDM) employed
by both of them, they are congured with different subcarrier bandwidth and symbol length, causing that they can
not decode the signal from each other. Therefore, to realize
the exchange of control information, a new protocol stack
need to be designed only for LTE-U, because 802.11 WLAN,
as a simple system with inferior technical ability, can not
support the complex system procedure and architecture. In the
following subsections, we describe how to solve this problem
by our designs.
A. LTE-802.11 fusion protocol stack for LTE-U
We design the LTE-802.11 fusion protocol stack presented
in Fig.2, in which the protocol in eNB side is divided into two
components, 802.11 part and LTE part. Each of them possesses
the independent Radio Frequency (RF) part. 802.11 part needs
to fulll two main functions, and one of them is to receive and

2339

IEEE ICC 2015 - Workshop on LTE in Unlicensed Bands: Potentials and Challenges

Fig. 2.

LTE-802.11 fusion protocol stack

decode the 802.11 PHY frame from 802.11 devices and obtain
the following control parameters: i) the time length of 802.11
PHY frame by decoding Non-HT Signal Field (L-SIG) of PHY
frame header and Duration/ID Field of MAC frame header;
ii) the PHY frame attributes, such as the channel bandwidth,
modulation and coding and so on, to meet the requirement of
SIC. Another function is to broadcast the control parameters
of LTE-U, namely the time length of the unlicensed spectra
occupied by LTE-U, in form of 802.11 PHY frame to enable
all 802.11 devices to identify and decode it. As for LTE part, it
has the responsibility for adjusting the transmit power model
to adapt to the radio condition of unlicensed spectra sensed
by 802.11 part. It is noticeable that information exchange
between 802.11 part and LTE part would happen normally
in this stack. Hence we redene two novel protocol layers,
namely e-MAC for 802.11 part and e-RRC for LTE part. In
addition to performing the original functions standardized by
802.11 and LTE specications, these two protocol layers also
need to carry out the following tasks: i) e-MAC identies the
control signaling delivered from e-RRC and send it by 802.11
RF part in form of 802.11 PHY frame; ii) e-RRC identies the
control signaling from e-MAC, depending on which it adjusts
the transmit power model over unlicensed spectra.

and lower power model, otherwise to protect 802.11 devices


from harmful interference. Different operations for LTE-U
would be executed in the cases of no or existing 802.11 devices
access, and the detailed operation for the former is illustrated
in Fig.4. Firstly, the e-MAC of 802.11 part reports the sensing
results (no 802.11 access) to the e-RRC of LTE part by
(2). Once LTE-U learns that, it would perform two tasks
simultaneously, informing 802.11 part how long the unlicensed
spectrum will be occupied by LTE-U in (3a) 802.11 Part
Tasks Command and reconguration of control parameters
for normal power model in (3b). According to (3a), 802.11
part broadcasts the LTE-U occupation time information for all
802.11 devices in the form of 802.11 PHY frame in (4), which
is carried by L-SIG of PHY frame header and Duration/ID
eld of MAC frame header. After nishing (4), 802.11 part
issues (5) Tasks Complete to notice LTE part. Then LTE part
conveys the relevant control information, generated by (3b),
to UEs which are ordered by (6) to transmit with normal
power model over unlicensed bands. (6) is the RRC (layer
3) signaling which should be transmitted over the licensed
spectra of LTE-U. After completing the normal power transmit
with the time length indicated by (3a), 802.11 part reruns the
spectrum sensing.

B. Operations of Spectrum Etiquette


Fig. 4.

Fig. 3.

Frame structure of the proposed spectrum etiquette

The proposed spectrum etiquette operates as Fig.3 shown.


At the beginning, an initial spectrum sensing is performed by
802.11 part within releasing time to determine the status of
the frequency band. The length of releasing time Trel should
be large enough, more than the summation of DCF inter-frame
space (DIFS) and the maximum length of contention window
(CW), Trel = TDIF S + TCW , to ensure that 802.11 devices
have the higher priority over unlicensed spectra. Based on the
decision of spectrum sensing, eNB communicates using the
normal power model if 802.11 devices are detected to be idle

Operations for no 802.11 devices access

As for the case of existing 802.11 devices access shown in


Fig.5, 802.11 part should decode the PHY frame from 802.11
device and obtain its length and attributes information in (2),
then send these information to LTE part by (3). Based on (3),
LTE part recongures the control parameters about unlicensed
spectrum transmission to lower power model and informs UEs
by signaling (5), which carries the attributes information of
802.11 PHY frame to assist performing SIC of UEs. After the
duration of 802.11 PHY frame, LTE part stops transmitting
with lower power model and 802.11 part reruns the spectrum
sensing.
C. 802.11 WLAN hidden node problem for LTE-U
For feasibility in practice, the 802.11 WLAN hidden node
problem for LTE-U should be addressed. When LTE-U uses

2340

IEEE ICC 2015 - Workshop on LTE in Unlicensed Bands: Potentials and Challenges

Fig. 5.

gains from eNB to UE, from AP i to STA i, from eNB to STA


i and from AP i to UE, respectively. Besides, we suppose that
one eNB and M 802.11 APs coexist over a 20MHz unlicensed
channel, each of which communicates with one receiver.
In the proposed spectrum etiquette, eNB adapts the transmit
power at the end of each releasing time based on the result
of spectrum sensing performed by 802.11 part. When the
unlicensed band is detected to be idle, eNB uses normal power
model with the transmit power PeN B , otherwise lower power
lower
model is activated with the lower power PeN
B . Following
the approach of [14][15], the instantaneous channel capacity
when eNB accesses the unlicensed band with normal power
B
is C1 =B log2 (1 + g0 PeN
), where B = 20M Hz. According
2
n
to CSMA/CA mechanism, only one 802.11 WLAN AP could
obtain the channel by competition. Thus, with taking lower
power model, the channel capacity can be given by B log2 (1+
lower
g0 PeN
B
2 +hi PAP i ) without SIC when AP i accesses this channel,

Operations for existing 802.11 devices access

the LTE-802.11 fusion protocol stack to listen for 802.11


WLAN signal, it can detect that only in a limited range due
to the lower power of 802.11 devices. If there is no 802.11
WLAN signal detected in this limited range centered at eNB,
the eNB then uses normal power model, which can possibly
interfere the 802.11 WLAN devices beyond the detection
range of eNB but within the coverage of normal transmission
power. Thus, we propose that LTE-U should deploy the dense
small cells with lower power adopted by eNB than that of
conventional LTE. And the coverage of eNB should be less
than or equal to its detection range to avoid the 802.11
WLAN hidden node problem. Moreover, the deployments of
low power and small cell also follow the maximum power
constraint and overcome the higher propagation loss on the
higher unlicensed spectrum.

g P lower

B
while it would be promoted to C2 = B log2 (1 + o eN
) if
2
n
the receiver adopts SIC, where PAP i represents the transmit
power of AP i. To avoid the harmful interference for 802.11
WLAN, we consider an interference power constraint that can
be formulated as follows:

lower
h0i PeN
B

eN B

volume of each 802.11 PHY frame is D. Thus, the average


transmission time of that, which denotes the duration from
initial sending by AP to this frame successfully received
by STA, and involves the time of initial transmission
and

retransmission, can be obtained by t802.11 = D C. If the total
arrival rate of 802.11 PHY frame of all the APs is and obeys
Poisson distribution, we can obtain the probability of existing
802.11 devices access an unlicensed band per releasing time
P (i 1) =


(Trel )i
i=1

Fig. 6.

(1)

where denotes the maximum interference power that is


tolerable by 802.11 receiver and M is the set of APs.
When lower power model is employed in LTE-U, the
channel capacity for STA i can be obtained by B log2 (1 +
gi PAP i
2 +h P lower ). According to CSMA/CA mechanism, we can
n
0i eN B
assume all 802.11 WLAN devices have the same probability to access a channel. Therefore, the average channel 
capacity for all STAs can be calculated by C =
gi PAP i
B
log2 (1 + 2 +h
lower ). Assume that the average data
M
0i P
iM

III. E RGODIC C APACITY UNDER O UR P ROPOSED


S PECTRUM E TIQUETTE

iM

Coexistence analysis model of LTE-U and 802.11 devices

In this section, we study the problem of deriving the optimal


lower power and time length of each normal power transmission allocation strategy that maximizes the ergodic capacity
of LTE-U systems under the proposed spectrum etiquette. For
analytical tractability, only downlink of LTE-U and 802.11
is considered in problem formulation with the system model
shown in Fig.6, in which g0 , gi , h0i and hi denote the channel

i!

eTrel

(2)

where i is the number of 802.11 PHY frames. Accordingly


the probabilities of normal and lower power model activated in
eNB are P1 =P (i 1) and P2 =P (i= 0)=1P1 , respectively.
Undoubtedly, the normal power transmission length Tocc is
also a decisive factor for the effectiveness of our proposed
spectrum etiquette. If it is too long, serious delay would be
caused for 802.11 WLAN, and conversely too short brings
heavy system burden and signaling overhead for LTE-U. We
pursue the optimal time length of each transmission with
normal power model, which is limited by time threshold T

2341

IEEE ICC 2015 - Workshop on LTE in Unlicensed Bands: Potentials and Challenges

to avoid too long and try the best to satisfy the load of 802.11
PHY frame over the unlicensed band. Thus the corresponding
constraint is given below:


(T )i
i=1

i!

eT i (t802.11 + T  ) T
(3)

where T  represents the overheads, such as sensing report,


task command and control parameters broadcast etc in Fig.4
to implement our proposed spectrum etiquette, and T  stands
for the average competition overhead of CSMA/CA. This conlower
straint reects a practical fact that higher PeN
B degrades the
channel capacity for 802.11 WLAN receiver C and lengthens
the average transmission time t802.11 . Furthermore, it shortens
the time of normal power transmission Tocc . Evidently, the
lower
relationship of the two variables, PeN
B and Tocc , of this
optimization problem is opposite.
Finally, the optimization problem that maximizes the ergodic throughput of LTE-U under the aforementioned constraint
derived from the proposed spectrum etiquette can be formulated as:
max imize

lower ,T
{PeN
occ }
B

C = E{P1
+P2

Tocc
Trel +T  +Tocc
t802.11
T  +t802.11

C1

C2 }

(4)

Subject to (1) (3)


Apparently, this joint optimization problem is convex and
could apply Karush-Kuhn-Tucker (KKT) conditions. Hence it
can be solved by Lagrangian method.
IV. S IMULATION R ESULTS
In this section, we present simulation results for the reachable ergodic capacity of LTE-U and 802.11 WLAN over the
unlicensed band under our proposed spectrum etiquette with
different threshold and T , and the 802.11 PHY frame arrival
rate , respectively. Moreover, the simulation results with SIC
is compared with that without SIC to verify the availability.
We assume that three APs exist within the coverage of eNB,
M = 3, and all of them have the same transmit power.
According to the specications of 802.11n [5], the sensing
time of 802.11n is from 34s to 9241s (from DIFS to
DIFS+CWmax), Trel should be larger than that, and equal
to 10ms in simulation. Besides, suppose that three APs have
the same probability to access a channel. The other parameters
in simulation are set as Table I.
TABLE I
SIMULATION PARAMETERS
Parameters Value Parameters Value Parameters Value
g0
0.3
h2
0.2
PeN B
300mw
g1
0.5
h3
0.1
PAP i
100mw
g2
0.4
h01
0.2
T
10ms
g3
0.3
h02
0.3
T 
5ms
h1
0.3
h03
0.25
D
1M bits

In Fig.7, the reachable ergodic capacity of LTE-U and


802.11 WLAN is illustrated versus the tolerable interference
threshold under the condition of the constant time threshold

x 10

LTEU W SIC
LTEU W/O SIC
802.11 WLAN

1.55

Ergodic throughput (bits/sec)

Trel + T  + Tocc +

1.6

1.5
1.45
1.4
1.35
1.3
1.25
1.2
1.15

10

20

30

40

50

60

70

80

90

100

110

Tolerable interference threshold (mw/20MHz)

Fig. 7. Ergodic throughput of LTE-U and 802.11 WLAN versus the tolerable
interference threshold

T = 1s and arrival rate = 4/s. It is apparent that the


ergodic capacity of LTE-U arises with the growth of at the
beginning, because higher leads to higher channel capacity
for LTE-U in the lower power model. Nevertheless, when
grows to a specic value, the ergodic capacity would begin
to decline with the reason that higher can also cause the
stronger interference for 802.11 devices and reduce the channel capacity of 802.11 devices (C). Furthermore, the decrease
of C leads to the increase of the average transmission time
of each 802.11 PHY frame (t802.11 ), and the length of normal
power model for LTE-U (Tocc ) is shortened correspondingly,
according to constraint (3). Thus, it is noteworthy that larger
may not bring the higher ergodic capacity. Furthermore,
by comparing red curve with blue curve, we can see that the
ergodic capacity of LTE-U with SIC, no matter how much
is, is much larger than that without SIC. It could fully verify
the availability of SIC under our proposed spectrum etiquette
and operation scheme. Additionally, from the green curve, we
can see that the interference from LTE-U has weak effect on
the ergodic capacity of 802.11 WLAN. It is because constraint
(3) can satisfy the trafc need of 802.11 WLAN effectively.
Fig.8 displays the ergodic throughput of LTE-U and 802.11
WLAN versus the time threshold T under the condition of
the constant threshold = 100mw/20MHz and arrival rate
= 4/s. It can be clearly seen that the ergodic throughput
of LTE-U increases as T becomes higher. This result can be
explained by the fact that as T grows, the Tocc grows either
according to constraint (3), and accordingly the overhead of
Trel and T  is mitigated, simultaneously. As a result, the
spectrum efciency of unlicensed band is promoted and the
ergodic throughput of LTE-U is enhanced. While it is also easy
to see that the curve of LTE-U ergodic throughput becomes
smooth gradually with the increase of T, which means that the
ergodic throughput almost stops enhancing. In consideration
of the serious delay for 802.11 devices caused by large T, it
should not be congured with a too large value. Additionally,
the ergodic capacity of LTE-U with SIC in this situation
signicantly exceeds that without SIC. As to the ergodic
throughput of 802.11 WLAN, there is nearly no change with

2342

IEEE ICC 2015 - Workshop on LTE in Unlicensed Bands: Potentials and Challenges

the growth of T. Because with the xed , the 802.11 PHY


frame need to transmit will proportionally increase with the
growth of T, and constraint (3) can guarantee the transmit
opportunity of each 802.11 PHY frame as well as possible.
1.8

x 10

Ergodic throughput (bits/sec)

1.7
1.6
1.5
1.4
1.3
LTEU W SIC
LTEU W/O SIC
802.11 WLAN

1.2
1.1
500

1000

1500

2000

2500

3000

Time threshold (ms)

Fig. 8. Ergodic throughput of LTE-U and 802.11 WLAN versus the time
threshold T

Fig.9 shows the ergodic throughput of LTE-U and 802.11


WLAN versus the 802.11 PHY frame arrival rate under the
condition of the constant threshold = 100mw/20MHz and
T = 1s. It can be easily observed that the ergodic throughput
of LTE-U is decreasing with the increase of , the reason of
which is that a great many 802.11 PHY frames occupy too
much time in the time threshold T according to constraint
(3) and suppress the time length of normal power model Tocc .
However, SIC in this situation can effectively boost the ergodic
throughput of LTE-U and alleviate the bad impact of the large
. Moreover, from the green curve, the ergodic throughput of
802.11 WLAN grows with the growth of due to constraint
(3). When grows to large enough, it would mostly stop to
increase since the channel has reached full load status.

Ergodic throughput (bits/sec)

2.5

x 10

1.5

0.5

LTEU W SIC
LTEU W/O SIC
802.11 WLAN
1

802.11 PHY frame arrival rate

Fig. 9. Ergodic throughput of LTE-U and 802.11 WLAN versus the 802.11
PHY frame arrival rate

V. C ONCLUSION
The analytical results of the coexisting problem between
LTE-U and 802.11 on unlicensed spectra show that 802.11

WLAN almost does not have any opportunity to access an


unlicensed band when it coexists with LTE-U. In order to
improve this unfair competition, we proposed a spectrum etiquette to restrict the priority of LTE-U and balance competitive
relationship. Moreover, to implement the spectrum etiquette,
we design an LTE-802.11 fusion protocol stack on eNB
side and describe particularly the operation of the spectrum
etiquette in LTE-U. With the spectrum etiquette restriction,
the throughput of LTE-U over unlicensed spectra may greatly
reduced. Therefore, SIC is congured in the receiver of LTE-U
to perform interference coordination. In addition, we formulate
the problem of deriving the optimal lower power and time
length of each normal power transmission allocation strategy
that maximizes the ergodic capacity of LTE-U. Finally, the
simulation results we provided indicate that our proposed
spectrum etiquette, the interference coordination scheme can
improve the ergodic throughput of LTE-U signicantly, and the
demand of 802.11 WLAN trafc could be satised as much
as possible.
VI. ACKNOWLEDGEMENT
The work was partially supported by the 973 Program
of China under Grant 2012CB316100, NSFC under Grant
61471303, and EU FP7 QUICK project under Grant PIRSESGA-2013-612652.
R EFERENCES
[1] CISCO Whitepaper, CISCO Visual Networks Index: Global mobile data
trafc forecast update 2012-2017, 2013.
[2] Federal Communications Commission, Spectrum Policy Task Force,
Rep. ET Docket, no.02-135, Nov. 2002.
[3] 3GPP RWS-140005, Scenarios, spectrum considerations and preliminary assessment results of U-LTE, 3GPP workshop on LTE in unlicensed spectrum, June. 2014.
[4] https://www.qualcomm.com/media/documents/les/extending-thebenets-of-lte-advanced-to-unlicensed-spectrum
[5] E. Perahia and R. Stacey, Next Generation Wireless LANs Throughput,
Robustness, and Reliability in 802.11n, Cambridge University Press,
2008.
[6] M. S. Gast, 802.11ac: A Survival Guide, Jul. 2013.
[7] Federal Communications Commission, Order and Second Memorandum Opinion and Order, Jun. 2014.
[8] 3GPP TS 36.211: Physical Channels and Modulation, V11.4.0, Sep.
2013.
[9] WINNERII Channel Models D1.1.2, V1.1.
[10] J. Chen, C.L. Liao and S.Q. Li, A Cooperative Spectrum Etiquette
for Cognitive Radio Network, 2006 IET International Conference on
Mobile and Multimedia Networks, 2006.
[11] B. Gao, Y. L. Yang and J. M. Park, A Credit-Token-Based Spectrum
Etiquette Framework for Coexistence of Heterogeneous Cognitive Radio
Networks, IEEE INFOCOM 2014, pp.2715-2723, May. 2014.
[12] D. Raychaudhuri and X. P. Jing, A Spectrum Etiquette Protocol for Efcient Coordination of Radio Devices in Unlicensed Bands, 14th IEEE
Proceedings on Personal, Indoor and Mobile Radio Communications
2003, vol.1, pp.172 - 176, Sept. 2003.
[13] X. P. Jing and D. Raychaudhuri, Spectrum co-existence of IEEE
802.11b and 802.16a networks using the CSCC etiquette protocol, First
IEEE International Symposium on New Frontiers in Dynamic Spectrum
Access Networks, pp.243 - 250, Nov. 2005.
[14] S. Stotas and A. Nallanathan, Enhancing the Capacity of Spectrum
Sharing Cognitive Radio Networks, IEEE Transactions on Vehicular
Technology, vol.60, no.8, Oct. 2011.
[15] L. Musavian and S. Aissa, Ergodic and Outage Capacities of SpectrumSharing Systems in Fading Channels, IEEE GLOBECOM, 2007.

2343

Anda mungkin juga menyukai