JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 108

MOBILE CONVERGED NETWORKS

COOPERATIVE SMALL CELL NETWORKS:

HIGH CAPACITY FOR HOTSPOTS WITH
INTERFERENCE MITIGATION
MINGJIE FENG, TAO JIANG, DA CHEN, AND SHIWEN MAO

ABSTRACT ter and receiver is greatly reduced, resulting in

high signal-to-interference-plus-noise ratio
Due to the high potential to enhance spectral (SINR). The low power of each BS enables more
efficiency and spatial reuse, small cell networks, efficient spatial reuse of spectrum, which in turn
SCNs, have emerged as a promising solution to improves the capacity of the entire network.
improve the capacity of mobile communication Moreover, compared to a traditional macrocell
systems so as to satisfy the ever growing demand BS that brings high leasing and maintenance
for high data rate services. However, without fees, a SCN can greatly reduce the cost of wire-
proper planning, the dense deployment of SCNs less operators.
may cause severe interference, resulting in limit- However, the benefits of SCNs come at the
ed capacity. In hotspots with a large number of price of a more complicated network architec-
users, the small cell network is challenged by the ture. The following features pose great chal-
extremely high aggregated capacity requirement lenges to the design and operation of an SCN.
and may fail to guarantee the quality of service
of all users. To leverage the benefits of SCNs A Large Number of Small Cell BSs — The number of
and overcome the drawbacks, we propose a SCN BSs within a given area is expected to be
Mingjie Feng, Tao Jiang cooperative small cell network, CSCN, architec- much larger than that of macrocell BSs. With
and Da Chen are with ture that jointly utilizes several advanced tech- limited backhaul capacity, it would be difficult to
Huazhong University of niques to enhance the capacity of hotspots. In perform centralized control and coordination.
Science and Technology. this article, we first examine the existing solu-
tions for capacity enhancement and hotspots. Vulnerability to Interference — Due to the small cov-
Mingjie Feng and Shiwen We then present the basic concept of the pro- erage area, small cell BSs may be close to each
Mao are with Auburn posed CSCN architecture, and discuss the relat- other in a hotspot, which may easily cause strong
University. ed technical aspects. The high potential of a intercell interference. If different bands are
CSCN in terms of capacity improvement and assigned to neighboring cells, the available band-
The work presented in interference mitigation is demonstrated by a width for each cell is greatly reduced. Moreover,
this article was support- simulation study. Finally, we present several deploying a small cell tier over an existing cellu-
ed in part by the Major open problems for future research based on the lar network may also cause inter-tier interference
State Basic Research CSCN architecture. if the two tiers occupy the same spectrum band.
Development Program
of China (973 Program)
with Grant number
INTRODUCTION Irregular Coverage Area — Unlike traditional cellu-
lar networks with a hexagonal coverage area for
2013CB329006, Joint The recent wide development of user terminals each BS, the coverage area of a small cell BS is
Specialized Research (e.g., smartphones and tablets) has triggered a usually irregular. To mitigate interference as
Fund for the Doctoral drastically increasing demand for high data rate well as guarantee network connectivity, it is
Program of Higher Edu- services. With limited spectrum, such demand expected that the overlapping coverage areas of
cation (SRFDP) and necessitates more efficient use of the spectrum different small cell BSs are small, and there are
Research Grants Council resource to improve the capacity of wireless net- no coverage holes. These requirements make the
Earmarked Research works. To this end, small cells have been recog- deployment of small cell BSs more complicated.
Grants (RGC ERG) nized as an effective means of enhancing
with number network capacity [1, 2]. Limited Power Budget — Compared to a macrocell
20130142140002, and in Compared to a single macrocell base station BS, the power of a small cell BS is quite limited.
part by the U.S. National (BS) with high power and large coverage area, a It is thus critical to efficiently utilize the power
Science Foundation small cell network (SCN) consists of multiple resource. Advanced technologies are needed to
under Grants CNS- low-power and spatially separated small-scale deal with the case when a large number of users
1247955 and CNS- BSs with small coverage areas. With such an are served by a small cell BS under a stringent
1320664. architecture, the distance between the transmit- power constraint.

108 1536-1284/14/$25.00 © 2014 IEEE IEEE Wireless Communications • December 2014

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 109

With the increasing number of mobile users, connections and wireless routers as access points, The objective of the
these SCN problems would be aggravated in WiFi technology has been widely adopted due to
hotspots with a heavy traffic load. When a large its salient features of low cost and easy deploy- CSCN is to improve the
amount of users assemble in a certain area (e.g., ment. However, it operates on unlicensed spec-
a shopping mall or a football stadium), small cell trum bands and adopts contention-based
user QoS and the total
BSs could be overloaded and unable to serve medium access control (MAC) protocols, mak- network capacity in
each user with limited spectrum resources. ing it hard to guarantee the QoS requirements
Under this circumstance, traffic congestion and of a large number of users. hotspots. The key com-
call outage may frequently happen, resulting in The femtocell concept was proposed for ponents of the CSCN are
degradation of user quality of service (QoS). QoS provisioning and capacity enhancement.
Although the SCN is initially regarded as a solu- Femtocells are user deployed indoor low-power BS deployment, BS-coor-
tion to guarantee user QoS in hotspots, support- access points that operate on licensed spectrum
ing high data rate services for a large number of band with typical coverage range of 10–50 m.
dination-based dynamic
users now becomes a new challenge. The small propagation loss and low power resource management,
In this article, we focus on the problem of enable high data rates and spectrum reuse, ben-
exploiting cooperative small cells to provide efiting both indoor users and wireless opera- and interference coordi-
high data rate services for hotspots. Since the tors. Furthermore, the traffic burden of a nation.
SCN needs to provide high data rate services to cellular network is reduced since indoor users
a large number of users, it faces many new are served by femtocells. However, femtocell
challenges when applied in hotspots. Although technology only improves indoor coverage,
the proposed schemes in prior works [3–5] can while adding femtocells over an existing cellular
deal with the overloading problem in hotspots, network potentially causes interference to out-
they are based on the traditional cellular net- door cellular users.
work architecture, and the traffic loads mainly The drawbacks of femtocells are largely
consist of voice services with low requirements caused by the chaotic nature of their deployment
for data rates. Therefore, these schemes cannot by users. Thus, effective control and coordina-
be directly applied in the SCN to serve tion are necessary for overcoming such draw-
hotspots. Furthermore, the inherent features of backs. To this end, an operator-deployed
SCN mentioned before introduce new chal- distributed antenna system (DAS) offers an
lenges in the design of SCNs. In particular, the effective solution, and has been applied to
large number of SCN BSs makes it a great chal- improve both indoor and outdoor capacity. The
lenge to adopt centralized network control basic idea of a DAS is to deploy spatially dis-
strategies. Thus, decentralized strategies that tributed antenna units that are close to users,
allow local cooperation among the small cell and these antenna units are connected to BSs
BSs are highly appealing. To overcome these with dedicated wireline connections. As a femto-
problems and satisfy the high capacity require- cell, a DAS brings about benefits in improved
ment at hotspots, we propose a cooperative link quality and spatial diversity. However, a
small cell network (CSCN) architecture in this potential problem for a DAS may be the limited
article. Based on the SCN architecture, the pro- capability of antenna units. Since the antenna
posed CSCN leverages several existing tech- units can only act as transceivers, different
nologies with flexible designs to improve the antenna units cannot distributively coordinate/
spectrum utilization and network capacity to cooperate with each other. Thus, a BS needs to
provide enhanced QoS to users in hotspots. perform centralized control over all the connect-
The goal of this article is to provide an insight- ed antenna units, which brings new challenges to
ful look into the architecture and achievable the backhaul and processing units in the BS,
performance enhancements of the CSCN. An especially when the number of antenna units is
illustrative example and a simulation study are large. Besides the DAS, relay was proposed as
presented to demonstrate the high potential of another approach to improve coverage and
the proposed CSCN approach. capacity. In areas with poor coverage, relay
The remainder of this article is organized as nodes are deployed to receive, decode or ampli-
follows. We first review several existing tech- fy, and forward signals for users, resulting in
nologies for capacity improvement and serving improved SINR and network capacity [8]. How-
hotspots. The CSCN concept is presented next, ever, since the relay nodes only serve as
followed by detailed discussions on various tech- transceivers without coordination ability, a BS
nical aspects of the proposed architecture. An needs to perform centralized control for the
illustrative example is provided along with a sim- resource allocations of all the relay nodes [9].
ulation study to demonstrate the performance of Thus, relay enhanced cellular networks bear sim-
the CSCN. We then discuss open problems and ilar limitations as DASs.
conclude this article. The essence of the above capacity improve-
ment techniques is to improve SINR and spec-
OVERVIEW OF EXISTING TECHNOLOGIES tral efficiency. According to Shannon’s formula,
increasing the total bandwidth is the most effec-
CAPACITY ENHANCEMENT tive way to improve network capacity. However,
given the limited spectrum resource, the only
With the idea of reducing transmission distance, choice is to create spectrum reuse opportunities
a major approach to improve capacity can be to improve spectrum utilization. From the spa-
classified as deploying access points that are tial domain, directional antennas, multiple-input
close to users. The main techniques include multiple-output (MIMO), and smart antennas all
WiFi access points, femtocells [6], and distribut- exploit spatial diversity to achieve spatial multi-
ed antenna systems [7]. With high-speed wireline plexing gain. With proper design of antenna

IEEE Wireless Communications • December 2014 109

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 110

parameters, the application of directional anten- SERVING HOTSPOTS

nas can reduce undesired signal leakage, thus
creating more spatial reuse opportunities in To satisfy the high capacity requirements at
wireless networks [10]. hotspots, several techniques have been proposed
For a MIMO system, through adjusting based on the cellular network architecture,
beamforming parameters according to the chan- including cell splitting [12], cell sectoring [3],
nel conditions of different antennas, capacity channel borrowing [4], and load balancing [5].
can also be effectively improved. For a smart The cell splitting method is essentially similar to
antenna system, a BS first estimates the location the small cell approach, where the original cells
of a user based on the arrival direction of uplink are split into smaller cells so that the average
signals. It then controls the downlink transmis- number of available channels within a given area
sion beam pattern so that the desired user is is increased. The cell sectoring technique
served while other users do not experience employs directional antennas at each BS, and
interference. This way, it is unnecessary to each BS serves multiple sectors with different
assign different channels or time slots to users spectrum bands, thus reducing the co-channel
served by the same BS; each user can be allocat- interference among different cells and improving
ed more resource, resulting in higher network the frequency reuse efficiency.
capacity. Through channel borrowing, a cell borrows
Furthermore, cognitive radio (CR) technolo- channels from adjacent cells to serve the users in
gy [11] allows secondary users to opportunistical- a hotspot. Last but not least, load balancing pro-
ly access the channels that are currently not vides another approach to deal with the heavy
occupied by primary users. With this approach, traffic load in hotspots. When a cell is overload-
spectrum opportunities can be utilized by sec- ed due to the hotspots in its coverage area, it
ondary users; thus, higher spectrum efficiency is can hand over some of its users to adjacent cells.
achievable. This way, the traffic burden in the hotspot cell
Nevertheless, due to limited spectrum sensing can be mitigated, and the QoS of its users can
capability, the CR system faces several problems be guaranteed.
such as misdetection, false alarms, and hidden
terminals. Consequently, existing CR systems,
such as IEEE 802.22 wireless regional area net- THE CONCEPT OF
works (WRANs), largely depend on a spectrum COOPERATIVE SMALL CELL NETWORKS
database to acquire channel state information.
The CSCN is an extension of the SCN for
enabling local cooperation among neighboring
small cell BSs. It consists of multiple small BSs
CSCN BS equipped with directional antenna connected to
each other via a wireline backhaul through the
X2 interface. The BS deployment pattern and
Cellular BS coordination are carefully designed. The objec-
tive of the CSCN is to improve user QoS and
the total network capacity in hotspots. The key
components of the CSCN are BS deployment,
BS-coordination-based dynamic resource man-
agement, and interference coordination.
Hotspot
Base Station Deployment — In the CSCN, the BSs
are deployed according to the environment and
CSCN BS average demand (with a certain amount of
redundancy). Then the coverage area of each BS
is adjusted to an appropriate shape to improve
spatial reuse and reduce interference, which can
be realized by applying directional antennas with
CSCN BS adjustable parameters. With the development of
hardware technology, BSs can be produced with
much smaller size than before, and the coverage
area of each BS can be flexibly controlled [13].
Therefore, flexible coverage areas and deploy-
Hotspot ment patterns of antennas are feasible in the
CSCN. When CSCN BSs are deployed in a
CSCN BS
hotspot, the environment information, such as
the architectural layout, is utilized to determine
the optimal shapes of the coverage areas.
CSCN BS
Dynamic Resource Management — In hotspots, the
user distribution and traffic load may change
quickly over time. Therefore, the CSCN should
dynamically allocate spectrum resources by joint-
ly considering the instantaneous requirements
and interference mitigation in order to improve
Figure 1. An example of a CSCN in a metropolitan area. capacity and accommodate the load. We pro-

110 IEEE Wireless Communications • December 2014

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 111

pose a decentralized BS cooperation scheme

that enables channel borrowing and user han- Coverage area 3
Coverage area 1
dover among neighboring BSs.

Interference Coordination — In the CSCN, BSs can

sense the spectrum occupation condition and
exchange information with neighboring BSs or
cellular BSs. They then coordinate with each
other to mitigate interference as well as effi- BS 1
ciently utilize spectrum resources. Since the
transmission beam of a mobile device (MD) is
omnidirectional, MDs served by neighboring
CSCN BSs may cause uplink interference to
each other. Moreover, the transmission beam of
a BS cannot be perfectly controlled, which poten- BS 3
tially causes downlink interference. Thus, coordi- Interference
nation between the BSs is necessary to mitigate BS 4
interference. We propose a cooperative and
decentralized interference management scheme
that employs spectrum sensing and inter-BS
BS 2
coordination to control the interference in real
time.
Coverage area 2
To better understand the principles of the
CSCN, we present an intuitive scenario in Fig. 1,
where the CSCN is employed in a metropolitan Coverage area 4
area to enhance the QoS of users in hotspots.
The CSCN BSs provide coverage to the desired : Interfering area of each BS : Interfered area
area while the interference caused to other areas
is controlled. Neighboring BSs coordinate with Figure 2. Application of a directional antenna for BS deployment in the
each other to allocate spectrum resources CSCN.
according to the instantaneous traffic load and
reduce interference. From the perspective of
both time and space, the CSCN improves the mission pattern illustrated in the figure. The spe-
spectrum utilization, resulting in the increased cific coverage areas of the two BSs can be preset
network capacity. or distributively coordinated between the two
Compared to an SCN, the spectrum utiliza- BSs, depending on the technology available. This
tion of a CSCN is improved due to the adoption way, we can first divide a hotspot into several
of directional antennas and dynamic resource areas, then employ multiple BSs to serve each
allocation through BS coordination. As shown in area with the above pattern. BSs that serve the
Fig. 2, two close BSs with directional antennas same area can coordinate with each other to
(i.e., BS 3 and BS 4) can simultaneously utilize optimize the network performance and improve
the same spectrum band since there is no over- the QoS of users in the area.
lap between their coverage areas. For BS 1 and Next, within one area, we propose a prelimi-
BS 2 with overlapped coverage, they can sense nary architecture of BS deployment to mitigate
the spectrum environment and instantaneous the downlink interference. The proposed archi-
traffic conditions, and exchange information with tecture bears a decentralized feature that can
each other via backhaul signaling. After that, BS reduce the control overhead. Take a rectangular
1 and BS 2 can coordinate with each other to area (which may be a square or a lobby in a real-
avoid mutual interference and traffic congestion. world scenario) shown in Fig. 4 as an example.
Suppose that with a transmission range of d m,
TECHNICAL ASPECTS the received signal strength from a CSCN BS is
sufficiently small that it causes negligible inter-
BASE STATION DEPLOYMENT ference. Then we divide the rectangular area
into three kinds of regions with parameter d. To
The goal of CSCN BS deployment is to improve mitigate the interference to users outside the
spatial reuse as well as mitigate downlink inter- rectangular area, we add constraints on antenna
ference through the application of directional directions for BSs located in regions 1 and 2.
antennas. Through careful design on the location For a BS in region 1, the antenna azimuth is
and coverage area of each BS, spectrum utiliza- restricted within a right angle, and the radiation
tion can be improved to satisfy the high capacity direction is limited to the interior zone of the
requirement of a large number of users. rectangular area in order to guarantee that the
The deployment of antennas in a CSCN can BS does not cause interference to users outside
take advantage of the specific architectural lay- the rectangular area. Similarly, antennas of BSs
out. Note that the coverage area of a BS can be in region 2 can only direct to the inner side to
controlled by using a tilted antenna with adjust- avoid interference with the outside. For BSs in
ed tilt angle [14], as illustrated by the example region 3, due to the relatively longer distance to
given in Fig. 3. Suppose the zone between the users outside the rectangular area, there is no
two buildings is a hotspot, and the BSs at the restriction on the directions of antennas. With
top of buildings A and B provide service to users such constraints, the downlink interference only
in the hotspot, with the vertical antenna trans- occurs within users and BSs in the given area, so

IEEE Wireless Communications • December 2014 111

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 112

allocation, we propose a cooperative and decen-

CSCN BS tralized BS coordination mechanism to improve
the performance of a CSCN in the hotspot.
When the instantaneous traffic load is changing,
CSCN BSs can dynamically adjust the occupied
spectrum resource by the antennas through
backhaul signaling, and can coordinate with each
other to meet the traffic demand.
CSCN BS
Initial Spectrum Allocation — In the previous section,
we divide a hotspot into several areas and con-
sider the BS deployment pattern for each area.
Within such an area, given the coverage areas of
all the BSs, we propose a spectrum allocation
strategy that considers both spatial reuse and
interference mitigation. The objective is to maxi-
mize the total number of channels of all BSs,
under constraints on interference mitigation and
guaranteeing fairness among the BSs. We
assume that there are M CSCN BSs and N chan-
Hotspot
nels in this area. As shown in Fig. 2, each CSCN
Building A Building B BS has a coverage area and an interfering area.
Define interference list as an M × M binary
Figure 3. Application of tilted antennas to cover a hotspot in the CSCN. matrix with element Fij, where Fij = 1 indicates
that BSs i and j interfere with each other due to
overlapped coverage areas, and Fij = 0 indicates
local coordination among neighboring BSs is that BSs i and j do not interfere with each other,
feasible. so they can utilize the same channel.
For an area with a general shape, we first To avoid the situation that some BSs are allo-
find the reference line that is d m away from the cated with all the channels while some other BSs
boundary of the area. Then, according to the have no available channel, the number of channels
reference line, we divide the area into three allocated to each BS m should have an upper
kinds of regions: corner, border, and center. A bound, denoted by Am. Therefore, the spectrum
corner region is between the edges and angles allocation problem can be formulated as
on the boundary and the reference line. A bor-
N M
der region is between the edge of the boundary
and the reference line. A center region is inside
max ∑ ∑ δ nm ,
{δ m } n =1 m =1
n

the reference line. These three kinds of regions

correspond to regions 1, 2, and 3 in the rectan- s.t. δ in + δ nj ≤ 2 − Fij , ∀i, j ∈{1, 2, …, M )} ,
gular example, respectively. Similar to the rect- ∀n ∈{1, 2, …, N } (1)
angle area example, we add constraints on the
N
∀m ∈{1, 2, …, M )} ,
antenna directions of BSs to mitigate the down-
link inter-area interference. In the corner region, ∑ δ nm ≤ Am ,
n =1
the antenna azimuth of a BS is restricted by the
angle based on the shape of the region, and the where dm
n
is an indicator and defined as
radiation direction is limited to the interior zone
of the area. In the border region, the antenna of ⎧1, channel n is assigned to BS m,
δ nm = ⎨
a BS can only be directed to the inner side of ⎩0, otherwise. (2)
the area. In the center region, there is no restric-
tion on the antenna direction. Obviously, Eq. 1 is a 0–1 integer program-
The proposed BS deployment architecture ming problem with linear constraints. Finding
not only controls the interference, but also the optimal solution of Eq. 1 through exhaustive
enables spectrum reuse. As the mutual downlink searching incurs NP-hard complexity, since the
interference between different areas is con- number of all possible solutions is 2MN. Never-
trolled, each area can utilize all the available theless, the solution to the optimization problem
downlink spectrum bands, which in turn is an initial spectrum allocation result that does
improves the network capacity. not consider the traffic load. Therefore, a subop-
timal solution is still useful. To obtain a subopti-
DYNAMIC RESOURCE MANAGEMENT mal solution with low complexity, we relax the
Although the BS deployment could mitigate constraint d m n
Π{0,1} to d m
n
Π[0,1]. Hence, the
inter-area interference, intra-area interference original problem is transformed into a linear
remains a problem. Since providing seamless programming problem (LP) and can be solved
coverage is the primary target, the coverage with existing methods such as the simplex algo-
areas of different CSCN BSs in the same area rithm. The solutions will be rounded up to 1 or
may overlap, resulting in intra-area interference. down to 0 while satisfying the constraints in Eq.
Here, we propose spectrum allocation strategies 1 to obtain a feasible suboptimal solution.
to deal with this problem. We first propose an Obviously, A m directly affects the spectrum
initial spectrum allocation strategy to mitigate allocation result. Note that neighboring BSs can-
the downlink interference among BSs in the not simultaneously utilize the same channel,
same area. Then, based on the initial spectrum while the objective function of the optimization

112 IEEE Wireless Communications • December 2014

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 113

problem is the total number of channels of all

BSs. Therefore, a BS with more neighboring BSs d
will be allocated fewer channels to leverage spa-
tial reuse and achieve a large objective value.
With the constraint of Am, even the BS with the 1 2 1 d
most neighboring BSs is expected to be allocated
a certain amount of channels. There is a trade-
off between fairness and overall network perfor-
mance. When Am is so small that it approaches
N/2, the total number of channels allocated to
all BSs is small. When A m is large, the fairness 2 3 2
among the BSs will become poor.
The derivation of the channel allocation con-
straint is based on the downlink scenario, since it
is based on the analysis of BS coverage areas. Due
to the mobility of users and the omnidirectional
transmission pattern of uplink signals, the uplink
interference cannot be perfectly controlled. To 1 2 1
this end, a cooperative interference avoidance
scheme will be proposed to address this problem.
The goal of adding the channel allocation con-
straints is to reduce the expected overhead for 1: 2: 3:
interference management in the operating pro-
cess, since neighboring BSs have already been
allocated exclusive channels, and the ratio of users
that require interference coordination is small.

Load Balancing through Base Station Coordination — In Figure 4. Area division for BS deployment in the CSCN.
this part, we consider the instantaneous traffic
for spectrum allocation. To deal with the poten-
tial overloading problem in a hotspot, we pro- transmissions, uplink interference, and inter-
pose two cooperative and decentralized BS ference between the cellular users and the
coordination strategies. CSCN, should also be managed. Furthermore,
although the directional antennas can be used
Channel Borrowing — When a CSCN BS is over- to control interference, the potential interfer-
loaded, it sends a request for more channels to ence caused by undesired sidelobes should also
all neighboring BSs. Upon receiving the request, be mitigated.
the neighboring BSs with idle channels feed back We develop a cooperative interference avoid-
the idle channel information. Then the overload- ance scheme to mitigate the uplink interference
ed BS utilizes the channels all its neighboring between CSCN BSs. This scheme can also be
BSs reported as idle. These additional channels applied to control the interference between the
can assist the overloaded BS in dealing with CSCN and the cellular network with minor mod-
heavy traffic. With this mechanism, BSs with ifications. Through spectrum sensing and chan-
heavy traffic could borrow channels from neigh- nel scheduling, interference can be effectively
boring BSs, and neighboring BSs utilize the controlled. In particular, the following proce-
same channel at different time instants. dures are adopted by CSCNs to achieve uplink
interference avoidance.
User Handover — Although the channel borrowing
mechanism can offer additional channels to an Spectrum Sensing — We assume that CSCN BSs
overloaded BS, another problem emerges when periodically sense the spectrum occupation of
the traffic load of a BS keeps increasing. Since a the radio environment. If a CSCN BS (denoted
CSCN BS needs to allocate limited power among BS A) detects the uplink signal of a user that is
all the channels, with the limited power budget, served by another BS, and the signal of this user
the power allocated to each channel may not be causes interference to the uplink transmission of
enough to guarantee the QoS of users. User BS A, it records the time-frequency usage pat-
handover could be an effective solution to this terns of this user, and sends the information in a
problem. When a CSCN BS detects that the control message to the neighboring BSs. The
number of users within its coverage area is too control message also contains the channel avail-
high and a handover is necessary, it sends a ability information, which indicates the number
request for handover to all neighboring BSs. of remaining channels at the BS.
Then the neighboring BSs with available power
and spectrum resource respond to the request. Confirming the ID of Interfering Users — Once a
With the feedback information, the overloaded neighboring BS (denoted BS B) receives the
BS reduces the pilot signal strength, and some of time-frequency usage patterns from BS A, it
the users will switch to some neighboring BSs. compares this information with its uplink
scheduling information. If the channel usage pat-
INTERFERENCE COORDINATION tern coincides with the scheduling information of
The proposed deployment and spectrum allo- a user, it indicates that this user interferes with
cation methods can only deal with downlink BS A. This way, for each CSCN BS, nearby inter-
interference. Due to omnidirectional uplink fering users can be identified.

IEEE Wireless Communications • December 2014 113

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 114

fading, where R is the transmission distance in

1000 meters.
SCN-DA+CB
SCN-DA In the first scenario, we consider the original
900 SCN-CB SCN without any modifications. Each area is
SCN
served by one BS located at the center, and the
800
Sum capacity of all users (Mb/s)

BS adopts an omnidirectional antenna. The

neighboring BSs use different sets of channels to
700
mitigate interference, and the spectrum usage of
600
each BS is static. In the second scenario, BSs are
equipped with directional antennas (DAs), and
500
one area is served by multiple BSs with the pro-
posed deployment architecture. For simplicity,
400 we consider that in each area served by two BSs
with directional antennas, a user is served by the
300 BS with the strongest received signal strength. In
the third scenario, we enable channel borrowing
200 (CB) between neighboring BSs, and the BSs
employ omnidirectional antennas. In the last
100 scenario, both DA and CB are employed. We
0 50 100 150 200
assume that when the number of users is lower
Average number of users in each area
than the number of channels allocated to the
serving BS, each user is allocated one channel.
Figure 5. Network capacity vs. average number of users (m) under different When the number of users is greater than the
scenarios. number of channels, one channel is shared by
multiple users in a time-division multiple access
(TDMA) pattern. In detail, suppose the number
Interference Scheduling — With the channel avail- of users is U and the number of channels is N.
ability information contained in the control mes- When N < U < 2N, the BS randomly selects U –
sage, BS B compares the numbers of remaining T user pairs and assigns one channel for each
channels of BS A and BS B. If BS A has more user pair. When 2N < U < 3N, the BS randomly
remaining channels, BS B informs BS A not to selects U – 2N groups of users, each group con-
schedule the channel used by the interfering user sists of 3 users, and these 3 users share one
for uplink transmission to avoid interference. If channel in different time slots. This process can
BS B has more remaining channels, BS B will be repeated as U further increases.
reschedule the uplink channel for the user that Figure 5 shows the aggregated capacity of all
causes interference to BS A. users in the hotspot as a function of different
To avoid the downlink interference caused by average numbers of users (m). The incorporation
BS A to the interfering user, BS B sends the of both deployment designs with DA and CB
downlink channel scheduling information of the between neighboring BSs improves the capacity
user to BS A. Thus, BS A avoids allocating the compared to the original SCN. The deployment
channel used by this user for downlink transmis- of BSs with DAs creates more spatial spectrum
sion. reuse opportunities; thus, the number of chan-
nels allocated to each user is increased, which in
turn improves the sum capacity. CB offers some
ILLUSTRATIVE EXAMPLE AND capacity gain when the average number of users
SIMULATION RESULTS is larger than the number of pre-allocated chan-
nels. This is because the borrowed channels
To demonstrate the high potential of the pro- from neighboring BSs could assist an overloaded
posed approach for capacity improvement and BS to provide better service to users and improve
interference mitigation, we present selected sim- the sum capacity. When the average number of
ulation results with different schemes applied. users keeps on increasing, the capacity gain
Consider a rectangular hotspot with size 200 m offered by CB becomes limited, since the neigh-
× 400 m, and divide this hotspot into 8 areas boring BSs are less likely to have any unused
with size 100 m × 100 m. In each area, the num- channels. When both CB and deployment design
ber of users is uniformly distributed with mean with DA are applied, the network capacity is fur-
value m. The network capacity and outage proba- ther improved. Therefore, we conclude that the
bility are two criteria for performance evalua- enhancements adopted in CSCN over SCN can
tion, and we assume that an outage event effectively improve network capacity.
happens when the capacity of a user falls below Figure 6 shows the outage probability vs. dif-
a predefined threshold. In this simulation, we ferent average number of users (m). Due to the
focus on the effects of applying deployment pat- capability for capacity improvement, the deploy-
tern and channel borrowing. Downlink transmis- ment design and CB also contribute to outage
sion is considered since most high data rate reduction. Note that the performance gain of CB
services are provided via the downlink. In the decreases when the average number of users is
simulations, the BS transmission power is 20 large, since neighboring BSs are also likely to be
dBm and the noise power density is –174 overloaded, and there is no channel to borrow.
dBm/Hz; the channel bandwidth is 200 kHz, and From the perspectives of space and time, both
there are 100 channels; the outage threshold is schemes create spectrum opportunities and
set to 100 kb/s; the channel has path loss 15.3 + improve spectrum utilization, resulting in less
37.6log10(R) in dB [15] and experiences Rayleigh traffic congestion. As expected, the combination

114 IEEE Wireless Communications • December 2014

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 115

of the two schemes could achieve the lowest out-

age probability. Similarly, we can conclude that 0.06
SCN-DA+CB
the CSCN is effective in serving hotspots with SCN-DA
proper design of the technical aspects, as dis- SCN-CB
0.05 SCN
cussed.

FUTURE RESEARCH DIRECTIONS

Average outage probability

0.04
Based on the CSCN architecture, the following
problems are open issues that should be further
investigated to fully harvest the high potential of 0.03
the CSCN.

OPTIMAL BASE STATION DEPLOYMENT 0.02

In the proposed BS deployment scheme, the
downlink interference between different areas is
controlled through adding constraints on anten- 0.01
na directions. Each area can thus reuse more
spectrum resource, leading to improved capacity.
0
However, one open yet challenging problem is 10 20 30 40 50 60 70 80 90 100
how to optimize the placement of BSs, as well as Average number of users in each area
the azimuth and range of antennas, so that the
sum capacity can be maximized while the QoS Figure 6. User outage probability vs. average number of users (m) under dif-
requirements of all users in the hotspot are satis- ferent scenarios.
fied. We can employ cooperation between BSs
to optimize performance, but the technical
details require further research. approach has distributed operation, which is
amenable to practical systems. The high poten-
OPTIMAL HANDOVER STRATEGY tial of the proposed architecture is demonstrated
Due to the relatively small coverage area of a with an illustrative example and simulation
CSCN BS, multiple neighboring BSs can be can- results. We conclude this article with a discus-
didates for load balancing when a CSCN BS is sion of open problems for future research.
overloaded. The handover decisions determine
the capacities of users and interference patterns. REFERENCES
Furthermore, the available power and spectrum [1] C.-L. I, L. J. Greenstein, and R. D. Gitlin, “A Macrocell/
Microcell Cellular Architecture for Low- and High-
resources and traffic loads of these neighboring Mobility Wireless Users,” IEEE JSAC, vol. 11, no. 61,
BSs are different. It is thus desirable to develop Aug. 1993, pp. 885–89.
the optimal handover strategy for the best CSCN [2] J. Hoydis, M. Kobayashi, and M. Debbah, “Green Small-
performance. Cell Networks,” IEEE Vehic. Tech. Mag., vol. 6, no. 1,
Mar. 2011, pp. 37–43.
POWER CONTROL [3] T. S. Rappaport, Wireless Communications: Principles &
Practice, Prentice Hall PTR, 1999.
It is well known that power control is a funda- [4] T.-S. P. Yum and W.-S. Wong, “Hot-Spot Traffic Relief in
mental approach for capacity improvement and Cellular Systems,” IEEE JSAC, vol. 11, no. 6, Aug. 1993,
pp. 934–40.
interference mitigation. This article does not dis- [5] W. Song, W. Zhuang and Y. Cheng, “Load Balancing for
cuss power control problem due to the lack of Cellular/WLAM Integrated Networks,” IEEE Network,
space. In future work, cooperative joint power vol. 21, no. 1, Jan./Feb. 2007, pp. 6–12.
and spectrum allocation algorithms should be [6] V. Chandrasekhar, J. G. Andrews, and A. Gatherer,
“Femtocell Networks: A Survey,” IEEE Commun. Mag.,
developed with easy implementation and low vol. 46, no. 9, Sept. 2008, pp. 59–67.
complexity. [7] A. Saleh, A. Rustako, and R. Roman, “Distributed Antennas
In addition, it would also be interesting to for Indoor Radio Communications,” IEEE Trans. Commun.,
develop a CSCN testbed such that the system vol. 35, no. 12, Dec. 1987, pp. 1245–51.
[8] P. Ralf et al., “Relay-Based Deployment Concepts for
performance can be demonstrated in a realistic Wireless and Mobile Broadband Radio,” IEEE Commun.
wireless environment. Such a CSCN testbed can Mag., vol. 42, no. 9, Sept. 2004, pp. 80–89.
not only validate the theoretical results, but also [9] H.-C. Lu and W. Liao, “Cooperative Strategies in Wire-
reveal new practical constraints that should be less Relay Networks,” IEEE JSAC, vol. 30, no. 2, Feb.
2012, pp. 323–30.
considered in the modeling and analysis, as well [10] R. R.Choudury et al., “On Designing MAC Protocols for
as identify new research problems. Wireless Networks with Directional Antennas,” IEEE Trans.
Mobile Comp., vol. 5, no. 5, pp. 477–91, May 2006.
CONCLUSIONS [11] J. Mitola and G. Q. Maguire, “Cognitive Radio: Making
Software Radios More Personal,” IEEE Pers. Commun.,
vol. 6, no. 4, Aug. 1999, pp. 13–18.
In this article, we present a CSCN architecture [12] V. K. Garg and J. E. Wilkes, Wireless and Personal
to meet the high capacity demand in hotspots. Communications Systems, Prentice Hall PTR, 1996.
The proposed architecture adopts sectorized [13] Y. I. Choni and Ali Hassan, “Optimal Adaptive Antenna
Arrays for Asynchronous Communication Systems,” IEEE
base station deployment, dynamic resource man- Trans. Antennas Propagation, vol. 60, no. 6, June 2012,
agement based on dynamic traffic load, and pp. 3071–76.
interference coordination to achieve enhanced [14] J.-S. Wu, J.-K. Chung, and C.-C. Wen, “Hot-Spot Traffic
capacity and reduced outage probability. The Relief with a Tilted Antenna in CDMA Cellular Net-
works,” IEEE Trans. Vehic. Tech., vol. 47, no. 1, Feb.
key is to leverage the cooperation of neighboring 1998, pp. 1–9.
BSs for better spectrum reuse, interference miti- [15] ITU-R Rec. M.1225, “Guidelines for Evaluation of Radio
gation, and capacity enhancement. The proposed Transmission Technologies for IMT-2000,” 1997.

IEEE Wireless Communications • December 2014 115

JIANG_LAYOUT_Layout 12/18/14 4:15 PM Page 116

BIOGRAPHIES His research interests include cognitive radio networks, fem-

tocell networks, and full-duplex communication. He is a
TAO JIANG [M’06, SM’10] is currently a Distinguished Pro- recipient of a Woltosz Fellowship at Auburn University.
fessor in the Department of Electronics and Information
Engineering, Huazhong University of Science and Technol- DA CHEN received his B.S. from Huazhong University of Sci-
ogy, Wuhan, P. R. China. He received B.S. and M.S. ence and Technology in 2009. He is currently working on
degrees in applied geophysics from China University of his Ph.D. degree at the same university. From September
Geosciences, Wuhan, P. R. China, in 1997 and 2000, 2012 to August 2013, he was a visiting Ph.D. student in
respectively, and his Ph.D. degree in information and com- the Department of Electrical Engineering and Computer
munication engineering from Huazhong University of Sci- Science, Northwestern University. Since September 2013,
ence and Technology in April 2004. From August 2004 to he is a visiting Ph.D. student in the Department of Electri-
December 2007, he worked at some universities, such as cal and Computer Engineering, University of Delaware. His
Brunel University and the University of Michigan-Dearborn. current research interests include the areas of wireless
He has authored or co-authored over 200 technical papers communications, especially for femtocell networks and
in major journals and conferences, and six books/chapters FBMC systems.
in the areas of communications and networks. He served
or is serving as a Technical Program Committee member- SHIWEN MAO [S’99-M’04-SM’09] received his Ph.D. in electri-
ship of major IEEE conferences, including INFOCOM, cal and computer engineering from Polytechnic University,
GLOBECOM, and ICC. He served as TPC Symposium Chair Brooklyn, New York. He is the McWane Associate Professor
for IEEE GLOBECOM 2013 and IEEE WCNC 2013. He has in the Department of Electrical and Computer Engineering,
served or is serving as an Associate Editor of technical Auburn University. His research interests include wireless
journals in communications, including IEEE Communica- networks and multimedia communications, with a current
tions Surveys and Tutorials, IEEE Transactions on Vehicular focus on cognitive radio, small cells, 60 GHz mmWave net-
Technology, and the IEEE Internet of Things Journal. He is works, free space optical networks, and smart grid. He is a
a recipient of the NSFC for Distinguished Young Scholars Distinguished Lecturer of the IEEE Vehicular Technology
Award in P. R. China. Society (VTS) in the Class of 2014. He is on the Editorial
Boards of IEEE Transactions on Wireless Communications,
M INGJIE F ENG received his B.E. and M.E. degrees from IEEE Internet of Things Journal, and IEEE Communications
Huazhong University of Science and Technology in 2010 and Surveys and Tutorials, among others. He received the 2013
2013, respectively, both in electrical engineering. He was a IEEE ComSoc MMTC Outstanding Leadership Award and
visiting student in the Department of Computer Science, the NSF CAREER Award in 2010. He is a co-recipient of The
Hong Kong University of Science and Technology, in 2013. IEEE ICC 2013 Best Paper Award and the 2004 IEEE Com-
He is currently a Ph.D. student in the Department of Electri- munications Society Leonard G. Abraham Prize in the Field
cal and Computer Engineering, Auburn University, Alabama. of Communications Systems.

116 IEEE Wireless Communications • December 2014