SUBMISSION TO IEEE TMC 1
Energy-efficient resource allocation for ultra-dense licensed and unlicensed
dual-access small cell networks
Adnan Shahid,Senior Member, IEEE, Vasilis Maglogiannis, Student Member, IEEE, Irfan Ahmed, Senior Member, IEEE, Kwang Soon Kim,Senior Member, IEEE, Eli De Poorter,
Ingrid Moerman
Abstract—In this study, an energy-efficient self-organized framework for sub-channel allocation and power allocation is presented for ultra-dense small cell networks, which can operate in both licensed and unlicensed bands. In order to protect legacy WiFi devices (operating in unlicensed bands), we consider the LTE operation in unlicensed bands based on CSAT, in which ’ON’ and ’OFF’ duty cycle approach is utilized. On the other hand, there are severe interference management problems among small cells (operating in licensed and unlicensed bands) and between macro cells and small cells (operating in licensed bands) due to co-channel and ultra-dense deployment of small cells. This article proposes a self-organized optimization framework for the allocation of sub-channels and power levels by exploiting a non-cooperative game with the objective to maximize the energy efficiency of dual-access small cells without creating harmful impact on coexisting network entities including macro cell users, small cell users, and legacy WiFi devices.
Simulation results show that the proposed scheme outperforms (6% and 11%) and (8% and 18%) the round-robin and the
spectrum-efficient schemes, respectively for two different small cell scenarios.In addition, it is shown that for less CSI estimation errors ς= 0.02, the maximum performance degradation of the proposed scheme is reasonably small (5.5%) as compared to the perfect CSI.
Index Terms—Self-organization, sub-channel allocation, power allocation, non-cooperative game, small cells, licensed band, unlicensed band.
F 1 INTRODUCTION
1
D
UEto the proliferation of real-time bandwidth hun-2
gry applications such as video streaming, online
3
gaming and mobile computing, there is a sharp increase
4
in the bandwidth demand. According to a recent report by
5
Cisco Global Mobile Data Traffic Forecast [1], there will be
6
around 11 billion mobile devices by2019which is above
7
the world’s expected population at that time (7.6 billion).
8
Also, it is reported by the industry that cellular traffic is
9
doubling every year; therefore, cellular network should
10
be able to deal with the 1000x increase in cellular traffic
11
by 2020 [2]. This increase in the traffic demand imposes
12
critical challenges in terms of meeting the quality of service
13
(QoS) requirements from the existing cellular network
14
architecture. The limiting factor here is the licensed spec-
15
trum, which is a real bottleneck in meeting the increasing
16
traffic demand. On the other hand, the continuous increase
17
in the cellular traffic is also a serious threat to the energy
18
consumption of future cellular networks and it has a direct
19
impact on the operating cost (OPEX) of cellular operators.
20
• A. Shahid (corresponding author), V. Maglogiannis, E. De Poorter and I. Moerman are with the IDLab, Department of Information Technology, Ghent University - imec, iGent Tower, Technologiepark - Zwijnaarde 15, B-9052 Ghent, Belgium. Email: [email protected]
• I. Ahmed is with the Department of Electrical Engineering, Higher Colleges of Technology, Ruwais Campus, Abu Dhabi, United Arab Emirates.
• K.S. Kim is with the Department of Electronics and Electrical Engineer- ing, Yonsei University, 50 Yonsei-ro, Seodaemun-gu, Seoul 120-749, Republic of Korea.
Manuscript received Oct 08, 2019; revised —.
Thereby, an energy-efficient design for the future cellular
21
networks is needed.
22
To cope with the ever increasing traffic demands, vari-
23
ous innovative technologies/methodologies have been de-
24
veloped such as non-orthogonal multiple access (NOMA),
25
multiple-input multiple-output (MIMO), carrier aggre-
26
gation (CA), small cells, context awareness, etc [3]–[6].
27
Among various options, small cells, which are deployed
28
as underlay within macro cells, boost the coverage and ca-
29
pacity by reusing the licensed spectrum.Qualcommstates
30
that the dense deployment of small cells is the foundation
31
of the 1000x vision of the fifth generation (5G) cellular
32
network [7]. This ultra-dense deployment of small cells
33
increases the coverage per square kilometer and the capac-
34
ity by reusing the spectrum. More specifically, this ultra-
35
dense deployment contributes a 56-fold in the 1000-fold
36
traffic demand [8]. Although this ultra-dense deployment
37
holds great promise for enhancing the spectrum efficiency
38
by exploiting both spatial and universal frequency reuse,
39
the co-location of small cells underlaid in macro cells
40
creates severe interference management problems. In this
41
context, recent scientific literature focuses on the interfer-
42
ence management for enhancing the spectrum efficiency
43
[9]–[11], while energy efficiency is almost neglected. The
44
increase in the spectrum efficiency corresponds to high
45
throughput but at the cost of large energy consumption
46
which is mostly not suitable for energy-aware networks or
47
constrained devices [12]. Recently, energy efficiency, which
48
is defined as information bits per unit of transmit energy
49
SUBMISSION TO IEEE TMC 2
(bits/J), has received significant attention from industry,
50
academia, and standardization bodies. Some studies have
51
revealed that energy efficiency will be an integral part of
52
the 5G cellular networks, especially for the ultra-dense
53
small cell environment for meeting the QoS requirements
54
[8]. Thus, we can say that both spectrum efficiency and
55
energy efficiency will be critical for interference manage-
56
ment in the ultra-dense small cell environment, which is
57
confirmed when looking at the 5G vision and requirements
58
[13]. Hu et al. [13] propose a cooperative mechanism for
59
finding a balance between energy efficiency and spectrum
60
efficiency, but it is not suitable for the ultra-dense en-
61
vironment because it incurs huge information exchange
62
overhead. On the contrary, this work aims to propose an
63
efficient and a less complex scheme which does not require
64
cooperation and thus, minimizes the information exchange
65
overhead.
66
Also, WiFi hotspot has been successfully utilized for
67
offloading the cellular traffic from licensed band to unli-
68
censed band [14]. Considering the benefits of small cells
69
and WiFi hotspots, one natural enhancement of small cells
70
is to operate in both licensed (e.g. LTE or third generation
71
(3G)) and unlicensed (e.g. WiFi) bands simultaneously. We
72
refer to these enhanced small cells as the dual-access small
73
cells since they can operate in both licensed and unlicensed
74
bands. The coexistence of LTE and WiFi in unlicensed
75
bands has recently gained significant attention and in
76
this regard, two prominent standardization efforts have
77
been made: LTE in LTE in unlicensed spectrum (LTE-U)
78
[15] and Licensed Assisted Access (LAA) [16]. LTE-U was
79
standardized by the LTE-U forum which is based on
80
Carrier Sense Adaptive Transmission (CSAT) for selecting
81
’ON’ and ’OFF’ duty cycles according to WiFi transmis-
82
sion occupancy. Whereas, LAA was standardized by the
83
3rd Generation Partnership Project (3GPP) in Release 13
84
and uses the listen-before-talk (LBT) technique (similar to
85
WiFi). In this work, we consider LTE-U based CSAT and
86
leave LAA for future investigations. The term dual access
87
small cell that we use in this work can be termed as LTE-U
88
based on CSAT. Later in this article, we will use the terms
89
dual access small cells and LTE-U interchangeably.
90
The rapid evolution of LTE-U raises critical concerns on
91
the coexistence of LTE and WiFi. Primarily, the coexistence
92
of LTE and WiFi is a challenging problem. The main chal-
93
lenge is to find a balance between the operation of LTE and
94
the proprietary technologies including WiFi that operate
95
in unlicensed bands. Traditionally, LTE was designed to
96
operate centrally as a non-contention technology and has
97
no concern with cross-tier interference i.e., interference
98
from other technologies. In addition, LTE is a streaming
99
technology which means that even in the absence of
100
data traffic, control and reference signals are transmitted
101
over-the-air. On the contrary, WiFi is a contention-based
102
technology and it only transmits when the channel is not
103
occupied. This makes WiFi an ideal candidate to operate
104
in unlicensed bands. Due to these fundamental differences,
105
the operation of traditional LTE in unlicensed bands could
106
have a significant impact on the performance of WiFi [17].
107
Even though due to the adoption of CSAT in LTE, its
108
operation can degrade the WiFi performance [18]. We also
109
observed the similar performance degradation trend by
110
experimentally validating the LTE impact on WiFi by using
111
open source LTE implementation [19]. This degradation
112
mainly because of two factors: (i) the incompatible LTE-U
113
duty cycle mechanism with WiFi devices and (ii) the lack
114
of an efficient coexistence mechanism when LTE and WiFi
115
devices are far apart from each other and the received
116
power is low [20]. To address such coexistence problems,
117
this paper considers dual-access small cells based on the
118
CSAT approach which provides a coexistence mechanism
119
by using ’ON’ and ’OFF’ duty cycles for LTE and other
120
unlicensed technologies including WiFi [21]. In the LTE-U
121
’ON’ period, each dual-access small cell base station (SBS)
122
utilizes sub-channels from both licensed and unlicensed
123
bands while the legacy WiFi devices backoff due to the
124
adoption of the carrier-sense multiple access with collision
125
avoidance (CSMA/CA) protocol. While in the ’OFF’ pe-
126
riod, the SBSs do not transmit and provide transmission
127
opportunities to the legacy WiFi devices. In this work,
128
we assume that ’ON’ and ’OFF’ periods are determined
129
by each dual-access SBS in a self-organizing way and
130
are based on the transmission occupancy of the legacy
131
WiFi devices. However, this creates another challenge of
132
interference management specially in the ’ON’ period due
133
to the co-channel operation of (i) small cells and macro
134
cells in licensed bands and (ii) small cells and the legacy
135
WiFi devices in unlicensed bands. In such a coexisting
136
environment, meeting the QoS requirements is a critical
137
challenge and for this, we aim to solve this by inves-
138
tigating sub-channel allocation and power allocation for
139
dual-access small cell networks. Generally, a centralized
140
architecture can be an option here for managing the inter-
141
ference optimally. However, such a centralized control is
142
not applicable in the considered dense environment due to
143
its lack of scalability and increased information exchange
144
overhead.
145
Motivated from the above-mentioned concerns, in this
146
work, a self-organized sub-channel allocation and power
147
allocation scheme for the downlink of dual-access small
148
cell networks is modeled as a non-cooperative game. Since
149
for an ultra-dense environment centralized management
150
approaches are not appropriate, we aim to propose a
151
self-organizing solution and for this game-theoretic ap-
152
proaches are most suited. Such a game-theoretic approach
153
provides a mathematical tool for modeling an optimiza-
154
tion problem [22], [23]. The main goal of the proposed
155
scheme is to enhance the energy efficiency of small cell
156
networks while meeting the throughput requirements of
157
small cell users, macro cell users, and the legacy WiFi
158
devices. Thus, energy efficiency is considered as the QoS
159
criterion of the proposed scheme. In the proposed non-
160
cooperative game, the dual-access SBSs are the players
161
of the game while the sub-channels and the power levels
162
are the strategies of the game. Since the goal is to design
163
an energy-efficient self-organized resource management
164
scheme, the problem is formulated as follows: each player
165
of the game (SBS) during the ’ON’ period selects the sub-
166
channels (from licensed and unlicensed bands) and the
167
power levels for enhancing its energy-efficiency without
168
creating harmful impact on the coexisting networking
169
entities including small cell users, macro cell users, and
170
the legacy WiFi devices. In addition, we propose that ’ON’
171
SUBMISSION TO IEEE TMC 3
and ’OFF’ periods are selected by each dual-access SBS in
172
a self-organizing way based on the legacy WiFi devices
173
occupancy. Concerning the complexity of the joint sub-
174
channel allocation and power allocation problem, the joint
175
problem is decomposed into two sub-problems: the sub-
176
channel allocation and the power allocation. In the sub-
177
channel allocation step, the allocation of sub-channels from
178
both licensed and unlicensed bands is carried out with the
179
concern of minimizing the impact of cross-tier and co-tier
180
interference components. Where the cross-tier interference
181
is the interference among different tiers i.e., small cells,
182
macro cell, and legacy WiFi devices while the co-tier is
183
between the same tier. In the power allocation step, the
184
selection of power levels on the sub-channels (selected
185
in the previous step) is carried out by exploiting dual-
186
decomposition Lagrangian multipliers.
187
1.1 Contributions
188
The main contributions of our work are summarized as
189
under:
190
1) A framework is proposed for solving the sub-channel
191
allocation and power allocation problem in the
192
downlink of dual-access small cells which operate in
193
both licensed and unlicensed bands. The framework
194
aims to maximize the energy efficiency of small cells
195
using a non-cooperative game without compromising
196
the QoS requirements of small cell users, macro cell
197
users, and the legacy WiFi devices.
198 199
2) Since the energy-efficient sub-channel allocation
200
and power allocation is in a fractional form and a
201
non-convex optimization problem, traditional convex
202
optimization techniques cannot be applied for finding
203
the optimal solution. We first transform the problem
204
into its equivalent subtractive form using the properties
205
of non-linear fractional integer programming and then,
206
propose an efficient and a less complex iterative
207
algorithm for solving the sub-channel allocation and
208
the power allocation individually. In addition, the
209
proposed iterative algorithm is shown to converge
210
to the optimal energy efficiency and to the Nash
211
Equilibrium point.
212 213
3) Simulations results are carried out in terms of energy
214
efficiency and spectrum efficiency to compare the pro-
215
posed scheme utilizing both licensed and unlicensed
216
bands with the comparative counterparts: a) the pro-
217
posed scheme utilizing only licensed bands, b) the
218
round robin non-cooperative power optimization game
219
(RR-NPOG) [24], and c) the spectrum efficient scheme.
220
The performance improvement of the proposed scheme
221
as compared to the counterparts (a,b, and c) increases
222
from 5%, 7.6%, and 10.9% to 7.73%, 11.04%, and 17.06%,
223
respectively, with the increase of the small cell density
224
from 20 to 40. In addition, the proposed scheme is eval-
225
uated for different levels of imperfection in estimating
226
CSI: perfect CSI and imperfect CSI. Which concludes
227
that for less CSI estimation errorsς = 0.02, the maxi-
228
mum performance degradation of the proposed scheme
229
is reasonable small (5.5%) as compared to the perfect
230
CSI.
231
2 RELATED WORK
232
The concept of aggregating radio resources is nowadays
233
an active research topic and in this context, the literature
234
includes several works that address the coexistence of LTE
235
and WiFi. On the other hand, an energy-efficient design
236
specially for a dense environment has recently gained
237
significant attention from the scientific community. Based
238
on this, we present the related work in three parts: i)
239
energy-efficient resource allocation in wireless networks,
240
ii) general approaches for LTE and WiFi coexistence, and
241
iii) game-theoretic approaches for LTE and WiFi coexis-
242
tence.
243
2.1 Energy-efficient resource allocation in wireless
244
networks
245
Li et al.[12] present a comprehensive survey on an energy-
246
efficient design of wireless networks and highlight key
247
areas such as information-theoretic analysis, MIMO tech-
248
niques, relaying, and resource allocation for meeting the
249
QoS requirements. Due to the explosion of high data rate
250
applications in future wireless networks, more energy will
251
be consumed for meeting such requirements. Therefore,
252
it is imperative to design energy-efficient mechanisms
253
while satisfying the requirements. On the other hand,
254
future wireless networks will be extremely dense and
255
in this regard, centralized management architectures are
256
not suitable due to their lack of scalability and increased
257
information exchange overhead.Meshkati et al.[25] present
258
a detailed overview of energy-efficient resource allocation
259
mechanisms in wireless networks using game theoretic
260
approaches. In a dense environment, game-theoretic ap-
261
proaches are most suited due to their ease of scalability
262
and reduced information exchange overhead. Xie et al.
263
[26] investigate and formulate an energy-efficient resource
264
allocation problem for heterogeneous femto cell cognitive
265
radio networks cells using a Stackelberg game. It only con-
266
siders co-tier interference and for cross-tier, it is assumed
267
that a spectrum resource borrowed from a primary net-
268
work can only be allocated to a single femto base station.
269
This assumption is not valid for a dense environment.
270
Arani et al. [27] present a self-organized mechanism for
271
joint channel and power allocation with the concern of
272
maximizing the energy efficiency of heterogeneous net-
273
works using a non-cooperative game. It only considers
274
co-channel interference but in a dense environment where
275
LTE and WiFi operate in similar unlicensed bands, cross-
276
channel interference also becomes important. Considering
277
the importance of cross-channel interference, we include
278
it in the problem formulation and by doing so better
279
spectrum decisions can be made. Samarakoon et al. [28]
280
propose a joint power control and user scheduling scheme
281
for enhancing the energy efficiency of ultra-dense small
282
cell networks using a dynamic stochastic game. It does not
283
address the coexistence problem of LTE and WiFi, which
284
is the prime focus of our work. Also, the scheme in [28],
285
cannot be directly applied to solve the formulated problem
286
in this work.
287
SUBMISSION TO IEEE TMC 4
2.2 General approaches for LTE and WiFi coexistence
288
Zakrzewska et al.[29] propose an idea of splitting the control
289
plane (C-plane) and the user data plane (U-plane) where
290
the C-plane of small cells is provided by a macro base
291
station (MBS) while the U-plane is provided by SBSs. This
292
splitting concept can be employed in dual-access small
293
cells which can operate in both licensed and unlicensed
294
bands. However, the authors do not consider interference
295
management, which is indeed of vital importance when
296
considering coexisting and dense environments.Khawer et
297
al. [30] propose an interference coordination scheme for
298
a CSAT gating mechanism in LTE-U and illustrate multi-
299
operator network performance.Lui et al.[31] propose a re-
300
source allocation scheme for balancing the traffic between
301
licensed and unlicensed bands while the WiFi throughput
302
is maintained by an optimization framework. He et al.
303
[32] present a novel proportional fair scheme for a fair
304
coexistence of LTE and WiFi is proposed. Despite of the
305
interesting results [29]–[32], all the focus is on spectrum
306
efficiency by managing or avoiding the interference while
307
energy efficiency is ignored and the solutions in these
308
propositions cannot be directly converted to energy ef-
309
ficiency problems. In addition, all the works focus on a
310
centralized control environment which is not suitable in a
311
ultra-dense environment as it poses formidable computa-
312
tional complexity.
313
2.3 Game-theoretic approaches for LTE and WiFi co-
314
existance
315
Etkin et al. [33] propose a sharing between LTE and
316
WiFi using a repeated non-cooperative game. A similar
317
scheme with an extended model is proposed in [34] using
318
a repeated non-cooperative game. The authors in [35],
319
[36] propose a spectrum allocation problem in unlicensed
320
bands using a non-cooperative game in which the base
321
stations are the players and the spectrum resources are the
322
strategies of the game. The goal in their proposition is to
323
optimize the spectrum efficiency for both downlink and
324
uplink. Thakur et al. in [37] present a resource allocation
325
and a cell selection mechanism for femto cell networks op-
326
erating in licensed and unlicensed bands. For the resource
327
allocation scheme, they propose a non-cooperative game
328
and a Q-learning framework for LTE operating in licensed
329
and unlicensed bands. Finally, they combine the resource
330
allocation scheme with the cell selection and show the
331
performance improvements in terms of throughput and
332
energy efficiency. However, they do not consider co-tier
333
and cross-tier interference which is a challenging problem
334
in a dense environment. In addition, they model the re-
335
source allocation problem while maximizing the through-
336
put instead of energy efficiency. Despite some efforts on
337
the exploitation of game-theoretic approaches for the LTE
338
and WiFi coexistence, all the propositions consider only
339
the spectrum efficiency as the performance objective and
340
the ultra-dense environment is not considered. In addition,
341
the propositions cannot be directly converted to energy
342
efficiency problems in order to achieve the required per-
343
formance.
344
According to the best of our knowledge, a self-
345
organized framework for an energy-efficient sub-channel
346
allocation and power allocation for ultra-dense small cell
347
networks which can satisfy the minimum QoS require-
348
ments of small cell users, macro cell users, and legacy
349
WiFi devices, has not been investigated so far. In addi-
350
tion, the extension of the results [26], [33]–[37] cannot
351
lead to a reduced complexity iterative sub-channel and
352
power allocation solution. We proposed a similar energy-
353
efficient self-organizing framework for a multi-tier device-
354
to-device communication environment [38]. In [38], the
355
original fractional form of energy efficiency was consid-
356
ered as the objective function. And because of its non-
357
convex nature, a sub-optimal solution for channel alloca-
358
tion and power allocation is proposed using an iterative
359
non-cooperative game approach. On the contrary, in this
360
work, we transform the fractional form of energy efficiency
361
into its subtractive which leverages to find a tractable
362
solution using dual-multiplier Lagrangian optimization.
363
The following points make our investigation novel and
364
worth investigating. First, most of the existing literature
365
on the coexistence of LTE and WiFi focus on spectrum
366
efficiency, while energy efficiency is largely ignored. In
367
this study, we aim to propose an energy-efficient sub-
368
channel allocation and power allocation scheme while
369
satisfying a) the minimum capacity requirement of dual-
370
access SBSs and b) the maximum tolerable interference by
371
macro cell users and the legacy WiFi devices.Second, the
372
interference management is also a critical issue in LTE-U
373
based on CSAT, especially during the ’ON’ period and this
374
is far more important in the considered dense small cell
375
environment. Third, a self-organized framework for dual-
376
access small cells is proposed without any involvement of
377
a centralized entity which is an attractive option for the
378
dense environment. Fourth, a novel less-complex iterative
379
algorithm for the sub-channel allocation and power alloca-
380
tion is proposed, which is shown to converge to the Nash
381
Equilibrium.
382
The rest of the paper is structured as follows: the
383
system model and the problem formulation are presented
384
in Section 3. The objective function transformation is pre-
385
sented in Section 4. The proposed self-organizing frame-
386
work for the sub-channel and the power allocation is given
387
in Section 5 and the simulation results are presented in
388
Section 6. Finally, Section 7 concludes the paper.
389
3 SYSTEM MODEL AND PROBLEM FORMULATION
390
3.1 Network model
391
A network model of the dual-access small cell network is
392
shown in Fig. 1. Considering the co-channel environment,
393
the two main interference components that are present in
394
the ’ON’ period of LTE-U based CSAT are:cross-tierandco-
395
tier. The former one is comprised of interference between
396
small cells and macro cells (operating in licensed bands),
397
and between small cells and the legacy WiFi devices (op-
398
erating in unlicensed bands). The later one is comprised of
399
interference among small cells (operating in licensed and
400
unlicensed bands). The ’ON’ period of LTE-U is based on a
401
number of WiFi characteristics including traffic load, node
402
density, etc., and is proposed to be determined by each
403
dual-access SBS in a self-organizing way. This overcomes
404
the scalability problem in the centralized management of
405
SUBMISSION TO IEEE TMC 5
small cell networks. Due to the determination of ’ON’
406
and ’OFF’ periods at the SBS level, it becomes extremely
407
important and challenging to manage the co-tier and cross-
408
tier interference components. Since the calculation of ’ON’
409
and ’OFF’ periods is based on the detection of WiFi signals,
410
it is possible that the received signal power of some legacy
411
WiFi devices is too low to be identified by the carrier
412
sense mechanism in CSAT [39]. Consequently, the trans-
413
mission of LTE-U in the ’ON’ period can interfere with the
414
legacy WiFi devices. In order to protect and respect the
415
QoS requirement of the legacy WiFi devices, we assume
416
according to [40] that a control channel is set up over
417
the wired backhaul between the neighbouring SBSs and
418
WiFi stations. If the QoS requirement of the legacy WiFi
419
devices is violated, the concerned SBS will get an alert
420
via the control channel which switches its period from
421
’ON’ to ’OFF’ accordingly. In addition, we assume that the
422
legacy WiFi devices operating in unlicensed bands follow
423
the LBT technique which is important for maintaining
424
a fair coexistence between LTE-U and other unlicensed
425
technologies. However, if a technology operates in unli-
426
censed bands and does not use LBT such as Bluetooth,
427
the following solutions can be utilized (in any order) for
428
addressing the problem and adjusting the duty cycle of
429
LTE-U SBSs: (i) coordination of dual-access SBSs and the
430
technology which does not employ LBT via a control
431
channel (ii) assessment of the activity of the technology,
432
and (iii) prediction of transmission of the technology using
433
machine learning techniques [41] [42] [43].
434
In the proposed self-organizing framework, each dual-
435
access SBS has to perform three tasks iteratively:sensing,
436
learning, andtuning. In thesensingtask, each SBS interacts
437
with the environment and acquire channel gains from both
438
licensed and unlicensed bands. For determining the LTE-
439
U ’ON’ or ’OFF’ period, each SBS detects the WiFi trans-
440
mission and adjust the duty cycle periods accordingly. In
441
thelearningtask, an iterative mechanism is taken into ac-
442
count for the allocation of sub-channels and power levels
443
with the concern of enhancing the energy efficiency while
444
guaranteeing the QoS requirements of small cell users,
445
macro cell users, and the legacy WiFi stations. In thetuning
446
task, the best learned strategies in task 2 are updated by
447
each dual-access SBS. The objective of the proposed self-
448
organizing framework is to enhance the energy efficiency
449
by exploiting sub-channels and power levels in ultra-dense
450
small cell networks without creating harmful impact on
451
the other network entities. In this study, we make the
452
following assumptions:
453
1) Channel state information (CSI) availability: It is assumed
454
that each dual-access SBS has the capability to ac-
455
quire CSI from macro cell users (operating in licensed
456
bands), small cell users (operating in both licensed
457
and unlicensed bands), and the legacy WiFi devices
458
(operating in unlicensed band). Uplink CSI can be
459
obtained at SBSs from the uplink pilots transmitted by
460
macro cell users, small cell users, and the legacy WiFi
461
devices. In case of time division duplexing (TDD),
462
downlink CSI can be estimated due to the channel
463
reciprocity between the uplink and the downlink [44].
464
While in frequency division duplexing (FDD), due to
465
the lack of the reciprocity, estimation of downlink CSI
466
is a difficult task and is achieved through separate
467
downlink sounding and feedback from all the tier
468
users. Although the CSI of the users far apart from
469
the SBSs are hardly obtained in practice, the impact
470
of such interferers is so small that the proposed algo-
471
rithm provides an upper bound on the performance of
472
the practical system. In order to evaluate the impact
473
of errors in estimating the actual CSI, an analysis is
474
also carried out with imperfect CSI.
475
2) Control channels for QoS satisfaction: In order to satisfy
476
the QoS requirements of macro cell users, small cell
477
users, and the legacy WiFi devices, we assume that
478
reliable control channels are available that connect
479
MBSs, SBSs and WiFi access points. In case the QoS
480
requirements of small cell users, macro cell users,
481
and/or the legacy WiFi devices are not satisfied, the
482
particular SBS will be alerted from the SBS, the MBS,
483
and/or the WiFi access point in order to change the
484
strategies.
485
3) Synchronization on sub-channels: A tight synchroniza-
486
tion is assumed among the transmitters so that the
487
interference only occurs whenever there is a transmis-
488
sion on similar sub-channels.
489
3.2 System model
490
In this study, we consider the downlink of dual-access
491
small cell networks with a frequency reuse-1 in an or-
492
thogonal frequency-division multiple access (OFDMA)-
493
based two-tier cellular environment, which comprises of
494
K macro cells each operated by a MBS at its center
495
overlaid withN dual-access small cells each operated by
496
a SBS. In addition, the frequency reuse-1 corresponds that
497
each dual-access SBS has complete access to licensed and
498
unlicensed bands. The total number ofY macro cell users
499
and X small cell users are randomly deployed within
500
coverage ofKmacro cells andN small cells, respectively.
501
Furthermore, we consider that there are E legacy WiFi
502
devices within coverage ofK macro cells and these are
503
randomly deployed around the vicinity ofN small cells.
504
Although we consider the duty cycle approach of CSAT for
505
the fair LTE and WiFi coexistence, still there is an impact of
506
interference of dual-access SBSs on the legacy WiFi devices
507
and vice versa (both operating in unlicensed band). This
508
interference is possible in two cases: (i) the transmission
509
of SBSs is not detected by the legacy WiFi devices due
510
to the hidden terminal problem and this happens because
511
of the incompatibility of CSAT and carrier-sense multiple
512
access (CSMA) and (ii) the transmission of the legacy WiFi
513
devices can interfere with SBSs due to lack of LBT in
514
LTE-U. In order to model this interference, we consider
515
that the number of legacy WiFi access points that interfere
516
with the dual-access SBSs is represented as V, where
517
V ⊂E.
518
The total number of G and H sub-channels are as-
519
sumed to be available from licensed and unlicensed bands,
520
respectively, and the number of sub-channels acquired
521
by each SBS is denoted as: Q and J, respectively, such
522
that Q ≤ G and J ≤ H, while the number of sub-
523
channels acquired by each MBS is denoted as G. The
524
SUBMISSION TO IEEE TMC 6
Proposed self-organizing framework for Dual-access
small cells
Initial assignment
Sense the environment and acquire channel gains
Sub channel allocation
Power allocation
Sensing
Learning
Update the profile
Tuning
1 2 X
...
Self-organized resource management
Self-organized resource management Self-organized
resource management
Self-organized resource management Self-organized
resource management
Self-organized resource management
Small cell 1 Small cell 2
Small cell 3 Small cell 4
Small cell 5 Small cell K
Macrocell 1
Macrocell 2 Macrocell L
Small cell user having dual connectivity Macrocell user
Cross-tier licensed interference
Co-tier licensed interference Co-tier unlicensed interference
Legacy AP 1
Legacy AP E
Legacy AP 2 Cross-tier unlicensed
interference
Fig. 1. Proposed self-organizing framework for ultra-dense dual-access small cell network.
total number ofRpower levels are available on the sub-
525
channels acquired by each SBS, while the power levels
526
on the sub-channels acquired by MBSs are assumed to be
527
fixed. Moreover, letGv,HvandRvdenote the sets of sub-
528
channels from licensed band, unlicensed band, and power
529
levels such that Gv = {1,2, . . . , G},Hv ={1,2, . . . , H},
530
andRv ={1,2, . . . , R}, respectively.
531
The transmission power of thekth MBS in the licensed
532
band sub-channels and of the nth dual-access SBS
533
in the licensed and unlicensed band sub-channels
534
are represented as Pk,gM ∈ h
Pk,1M, Pk,2M, . . . , Pk,GM i ,
535
Pn,gS,L ∈ h
Pn,1S,L, Pn,2S,L, . . . , Pn,GS,L i
and Pn,hS,U L ∈
536
h
Pn,1S,U L, Pn,2S,U L, . . . , Pn,HS,U Li
, respectively, wherePn,gS,L = 0
537
if the gth sub-channel from the licensed band is not
538
acquired by the nth SBS; otherwise, Pn,gS,L 6= 0.
539
The similar argument holds for Pk,gM and Pn,hS,U L.
540
The maximum power constraint on the acquired Q
541
and J sub-channels by MBSs and SBSs is given by
542
PM AXM and PM AXS such that PG
g=1Pk,gM ≤ PM AXM and
543
PG
g=1Pn,gS,L+PH
h=1Pn,hS,U L ≤PM AXS , respectively.
544
The performance of the proposed scheme is evaluated
545
in terms of energy efficiency (expressed in bits/J) of the
546
dual-access small cell networks. The signal-to-interference-
547
plus-noise ratio (SINR) on a macro cell user associated
548
with the kth MBS operating on the gth sub-channel is
549
given in (1) on the next page.
550
where GM→Mkk,g , GM→Mak,g , and GS→M,Lnk,g are the channel
551
gains on the similar sub-channels between the kth MBS
552
and the associated macro cell user, between thekth MBS
553
and theath MBS, and between the nth SBS and thekth
554
MBS, respectively; δkgM, δngS,L, δS,U Lnh , and δvhLeg−W,U L are
555
the binary indicators which correspond to whether the
556
gth sub-channel from the licensed band and thehth sub-
557
channel from the unlicensed band is used by the kth
558
MBS, the nth SBS and the vth legacy WiFi access point
559
respectively; and finally,σ2 is the noise power. Similarly,
560
the SINR on thegth sub-channel from the licensed band
561
and thehth sub-channel from the unlicensed band of the
562
nth SBS is given in (2) and (3) on the next page.
563
According to the additive modeling of error in estimat-
564
ing CSI as proposed in [45], here the imperfect CSI ˆ GS→S,Lnn,g
565
is modeled as the deviation from the actual CSIGS→S,Lnn,g
566
and is written as:
567
GS→S,Lnn,gˆ =GS→S,Lnn,g + ΩS→S,Lnn,g ,∀n. (4)
568
whereΩS→S,Lnn,g is the channel estimation error and is mod-
569
eled by a zero mean Gaussian random variable with vari-
570
anceςnn,gS→S,L. Additionally, the random variable ΩS→S,Lnn,g
571
is independent and identically distributed (i.i.d) on G
572
and H sub-channels from the licensed and unlicensed
573
bands, respectively, and on all the tiers including small
574
cells, macro cells, and the legacy WiFi devices. For the
575
sake of simplicity, we consider a minimum mean square
576
error (MMSE) channel estimator and because of that the
577
CSI estimation error and the actual CSI become mutu-
578
ally uncorrelated [46]. Similarly, the imperfect CSI of the
579
remaining channel gains in (1), (2), and (3) are modeled
580
by (4). For the sake of simplicity, we use the actual CSI
581
notations in the remainder of the paper. However, the
582
corresponding imperfect CSI is introduced according to
583
(4).
584
The average throughput of thekth MBS operating in
585
SUBMISSION TO IEEE TMC 7
SIN RMk,g = Pk,gMGMkk,g→M σ2+
K
X
a=1,a6=k
δagMPa,gMGMak,g→M
| {z }
macro cell−licensed−components
+
N
X
n=1
δngS,LPn,gS,LGS→M,Lnk,g
| {z }
small cell−licensed−components
,∀k, g, (1)
SIN RS,Ln,g = Pn,gS,LGS→S,Lnn,g σ2+
K
X
k=1
δMkgPk,gMGMkn,g→S
| {z }
macro cell−licensed−components
+
N
X
a=1,a6=n
δagS,LPn,gS,LGS→S,Lan,g
| {z }
small cell−licensed−components
,∀n, g, (2)
SIN RS,U Ln,h = Pn,hS,U LGS→S,U Lnn,h σ2+
N
X
a=1,a6=1
δahS,U LPa,hS,U LGS→S,U Lan,g
| {z }
small cell−unlicensed−components
+
V
X
v=1
δvhLeg−W,U LPv,hLeg−W,U LGLeg−Wvn,h →S,U L
| {z }
Legacy W iF i−unlicensed−components
,∀n, h. (3)
the licensed band and thenth SBS operating in the licensed
586
and unlicensed bands is given by,
587
ΨMk =W G
G
X
g=1
δkgMlog2
1 +SIN RMk,g
,∀k, (5)
ΨS,Ln = W G
G
X
g=1
δngS,Llog2
1 +SIN RS,Ln,g
,∀n, Q≤G,
(6)
ΨS,U Ln =W H
H
X
h=1
ρnδnhS,U Llog2
1 +SIN RS,U Ln,h
,∀n, J≤H, (7) whereW represents the system bandwidth,ρnis the ’ON’
588
period and 1−ρn is the ’OFF’ period of the LTE-U. In
589
the proposed dual-access framework,ρnis determined by
590
each SBS which is based on the determined WiFi transmis-
591
sion occupancy of dual-access SBSs and the legacy WiFi
592
devices in unlicensed bands. In order to determine this, the
593
dual-access SBSs are required to turn their radio frequency
594
(RF) transmission ’OFF’ which can guarantee the precise
595
estimation of WiFi traffic. In addition, the sensing time
596
needs to be sufficiently large to allow the SBSs to precisely
597
estimate the WiFi traffic. In order to fine tune the decisions
598
regarding the ’ON’ and ’OFF’ periods, the received WiFi
599
signal is averaged by each SBS [47].
600
The total average throughput of thenth SBS, operating
601
in both licensed and unlicensed sub-channels is given by,
602
ΨSn = ΨS,Ln + ΨS,U Ln ,∀n. (8) Also, the energy efficiency (expressed in bits/J) of the
603
nth SBS is given in (9) on the next page. Where pc is
604
the circuit power of the transmitter and is assumed to be
605
constant in this work.
606
Although we consider the downlink, the similar
607
scheme can be employed for the uplink as in [48] in which
608
CSI is estimated at SBSs from the orthogonal uplink pilots
609
from macro cell users, small cell users and the legacy WiFi
610
devices.
611
3.3 Notations
612
The notations and abbreviations used in Table 1 will be
613
used throughout the rest of the paper.
614
3.4 Problem formulation
615
Since the goal of the study is to enhance the energy effi-
616
ciency of the dual-access small cell network by exploiting
617
sub-channel allocation and power allocation from licensed
618
and unlicensed bands, the problemP1is formulated along
619
with the system constraints and is given in (10) and (11)
620
on the next page. Where GS→M,Lkn,g,z is the channel gain
621
between thenth SBS and thezth macro user of thekth MBS
622
operating on thegth sub-channel. The similar description
623
holds forGS→Leg−W,U L
vn,h,e , but between thenth SBS andnth
624
legacy WiFi device. The binary indicatorsδS,Lng andδnhS,U L
625
are the output optimization variables and are included in
626
P1in the form of ASn, whileδMkg and δvhLeg−W,U L are the
627
binary input indicators and are updated randomly.
628
The constraint C1 refers to the minimum capacity
629
requirement of dual-access SBSs and is represented as
630
RSM IN. The value is proposed to be selected by the op-
631
erator of SBSs. If the constraint is violated, then the SBS
632
gets an alert from the MBS to change the strategies so
633
that the minimum capacity requirement can be achieved.
634
The constraint C2 is used to protect macro cell users from
635
dual-access SBSs operating in licensed bands. For this, the
636
maximum tolerable limit is selected and is represented as
637
ζM,L which can be adjusted by the MBS operator based
638
on the location of users in the coverage area. In case the
639
constraint C2 is violated, the concerned SBS gets an alert
640
