7882 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO.
10, OCTOBER 2016
Vehicular Communications for 5G Cooperative
Small-Cell Networks
Xiaohu Ge, Senior Member, IEEE, Hui Cheng, Guoqiang Mao, Senior Member, IEEE,
Yang Yang, Senior Member, IEEE, and Song Tu
Abstract—Cooperative transmission is an effective approach
for vehicular communications to improve wireless transmission
capacity and reliability in fifth-generation (5G) small-cell net-
works. Based on distances between the vehicle and cooperative
small-cell base stations (BSs), the cooperative probability and
the coverage probability have been derived for 5G cooperative
small-cell networks where small-cell BSs follow Poisson point
process distributions. Furthermore, the vehicular handoff rate
and the vehicular overhead ratio have been proposed to evaluate
the vehicular mobility performance in 5G cooperative small-cell
networks. To balance the vehicular communication capacity and
the vehicular handoff ratio, an optimal vehicular overhead ratio
can be achieved by adjusting the cooperative threshold of 5G
cooperative small-cell networks.
Index Terms—Cooperative transmission, mobility performance,
small cell, vehicular communications.
I. I NTRODUCTION
I N future fifth-generation (5G) cellular networks, denser
and smaller cells are expected to provide a high transmis-
sion rate for users [1]–[3]. Different from traditional personal
users, vehicles are sensitive to transmission scenarios in 5G
Manuscript received July 13, 2015; revised December 2, 2015; accepted
February 15, 2016. Date of publication April 1, 2016; date of current version
October 13, 2016. This work was supported in part by the International Science
and Technology Cooperation Program of China under Grant 2015DFG12580
and Grant 2014DFA11640; by the National Natural Science Foundation of
China (NSFC) under Grant 61471180; by the NSFC Major International Joint
Research Project under Grant 61210002; by the Fundamental Research Funds
for the Central Universities under Grant 2015XJGH011; by the EU FP7-
PEOPLE-IRSES, project acronym S2EuNet (Grant 247083), project acronym
WiNDOW (Grant 318992), and project acronym CROWN (Grant 610524); by
the National International Scientific and Technological Cooperation Base of
Green Communications and Networks under Grant 2015B01008; and by the
Hubei International Scientific and Technological Cooperation Base of Green
Broadband Wireless Communications. The work of G. Mao was supported by
the ARC Discovery Project DP120102030 and the NSFC Project 61428102.
The work of Y. Yang was supported in part by the NSFC under Grant 61231009
and Grant 61461136003, by the National Science and Technology Major
Project under Grant 2016ZX03001024, and by the Science and Technology
Commission of Shanghai Municipality under Grant 14ZR1439700. The re-
view of this paper was coordinated by Dr. T. Luan. (Corresponding author:
Yang Yang.)
X. Ge, H. Cheng, and S. Tu are with the School of Electronic Information
and Communications, Huazhong University of Science and Technology, Wuhan
430074, China (e-mail: xhge@hust.edu.cn; hc_cathy@hust.edu.cn; songtu@
hust.edu.cn; songtu@mail.hust.edu.cn).
G. Mao is with the School of Computing and Communications, University of
Technology Sydney and National ICT Australia, Sydney, NSW 1466, Australia
(e-mail: guoqiang.mao@uts.edu.au; g.mao@ieee.org).
Y. Yang is with the Shanghai Institute of Microsystem and Information
Technology (SIMIT), Chinese Academy of Sciences, Shanghai 200050, China
(e-mail: yang.yang@wico.sh).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TVT.2016.2539285
0018-9545 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
cooperative small-cell networks [4], [5]. Moreover, due to the
mobility nature of vehicles and the related high vehicular speed,
the topology of vehicular wireless networks becomes highly
dynamic and prone to a recurrent link intermediate [6]–[8].
In this case, cooperative transmissions are recommended as a
promising solution for vehicles in 5G cooperative small-cell
networks [9]. However, there still exist some problems, such
as the frequent handoff and coverage problems for vehicles
in 5G cooperative small-cell networks [10]. Therefore, it is a
great challenge to investigate vehicular communications for 5G
cooperative small-cell networks.
To meet the communication requirements from vehicles,
some studies have been investigated for vehicular communi-
cations in cellular networks [11]–[15], [39]. Congestion and
awareness control techniques have been investigated for coop-
erative vehicular communications, which is based on wireless
communications between vehicles and with other infrastructure
nodes [11]. To minimize the cost of transmission or, alterna-
tively, transmission time in vehicular heterogeneous networks,
performing vertical handoff is an appreciable choice at lower
speeds, whereas it would be better to avoid vertical handoff and
stay in the cellular network at higher speeds [12]. Based on a
traffic model of two-tier cellular networks composed of macro-
cells and small cells, the impact that the user traffic dynamics,
the mobility of users, and the capacity constraint of the small-
cell backhaul have on the system performance has been eval-
uated in [13]. Accounting for vehicular mobility and network
load in cellular/802.11p heterogeneous networks, an analytical
model was proposed for estimating the average achievable
individual throughput, and an optimal handoff threshold was
derived in [14]. To resolve problems resulting from limited
roadside units and insufficient resources in vehicular ad hoc
networks, the vehicles were configured as special vehicular
small cells that have been further integrated into the layered
heterogeneous networks [15]. Considering that small-cell base
stations (BSs) are deployed at vehicles, a closed-form outage
probability was derived for evaluating the user gain in two-tier
cellular networks [16].
Considering that the coverage of small cells is smaller, the
cooperative communication is widely used for small-cell net-
works [9], [17]–[24]. In [17], multicell multiple-input multiple-
output (MIMO) cooperation concepts were examined from
different perspectives, including an examination of the funda-
mental information-theoretic limits, a review of the coding and
signal processing algorithmic developments, and consideration
of scalability and system-level integration. Based on random
cellular networks, a general methodology was proposed to treat
GE et al.: VEHICULAR COMMUNICATIONS FOR 5G COOPERATIVE SMALL-CELL NETWORKS
problems of cooperation in cellular networks, in the case where
data exchange is allowed only between pairs of nodes [18].
Taking into account the irregular BS deployment typically en-
countered in practice, the signal-to-interference-plus-noise ratio
(SINR) distribution with cooperation was precisely character-
ized in a generality-preserving form, and a tractable model was
further proposed for analyzing noncoherent joint-transmission
BS cooperation [19]. To mitigate the impact of the cross-tier
interference in multitier wireless networks, a scheme was pro-
posed for location-aware cross-tier cooperation between BSs
in different tiers for downlink coordination multipoint (CoMP)
transmission in two-tier cellular networks [20]. Utilizing the
average user throughput under CoMP and non-CoMP transmis-
sions after taking into account the downlink training overhead,
each user was allowed to select a transmission model between
coherent CoMP and non-CoMP to avoid the extra overhead
outweighing the cooperative gain in cellular networks [21]. To
fully exploit the benefits of heterogeneous networks, a radio
resource allocation scheme was proposed for cooperative relays
where the relay nodes with in-band backhaul act as micro BSs
and are able to serve users either independently or cooperatively
with macrocell BSs [22]. By defining the cooperation region as
a function of the user quality of service (QoS) requirements
and network load, a QoS-aware cooperative downlink schedul-
ing approach was proposed for cell-edge and handoff users,
which offers more reliability and higher effective capacity
[23]. Using stochastic-geometry-based heterogeneous cellular
networks, the coverage probability, the average achievable rate,
and energy efficiency were derived for K-tier heterogeneous
wireless networks with different cooperative sleep models for
small cells [24].
However, in all of the aforementioned vehicular communi-
cation studies, only simple scenarios, such as two cooperative
cells with a single antenna or multiple antennas, were con-
sidered, and the underlying vehicular communications were
limited to simple point-to-point wireless communications.
Moreover, the exact coverage probability of cooperative MIMO
small-cell networks with cochannel interference has not been
investigated. Moreover, a detailed investigation of the vehicle
mobility performance used for 5G cooperative MIMO small-
cell networks is surprisingly rare in the open literature. Moti-
vated by these gaps, in this paper, we consider the scenarios
of vehicular communications for vehicle-to-infrastructure, and
for urban roads, we derive the vehicular handoff rate and the
vehicular overhead ratio to evaluate the vehicular mobility
performance in 5G cooperative MIMO small-cell networks
considering cochannel interference. The contributions and nov-
elties of this paper are summarized as follows.
1) Based on distances between the vehicle and cooperative
small cells, the cooperative probability and the coverage
probability of cooperative small-cell networks have been
derived for vehicles equipped with multiple antennas.
2) From the proposed cooperative probability and cover-
age probability, the vehicle handoff rate and the vehicle
overhead ratio are proposed to evaluate the vehicle mo-
bility performance in 5G cooperative MIMO small-cell
networks considering cochannel interference.
7883
Fig. 1. System model.
3) Numerical results imply that there exists a minimum ve-
hicle overhead ratio for 5G cooperative MIMO small-cell
networks considering different cooperative thresholds.
This result can be used for optimizing vehicular commu-
nications in 5G cooperative MIMO small-cell networks.
The rest of this paper is organized as follows. Section II
describes the system model of 5G cooperative MIMO small-cell
networks where small-cell BSs follow Poisson point process
distributions. In Section III, the cooperative probability has
been derived for 5G cooperative MIMO small-cell networks.
Moreover, the coverage probability has been derived for 5G
cooperative MIMO small-cell networks considering cochannel
interference from adjacent small cells in Section IV. Further-
more, in Section V, the vehicular handoff rate and the vehicular
overhead ratio have been proposed to evaluate the vehicular
mobility performance in 5G cooperative MIMO small-cell net-
works. Numerical results indicate that there exists a minimum
vehicular overhead ratio for 5G cooperative MIMO small-cell
networks considering different cooperative thresholds. Finally,
Section VI concludes this paper.
II. S YSTEM M ODEL
Fig. 1 shows the system model of 5G cooperative small-
cell networks, which is a two-tier cellular network including
macrocell BSs and small-cell BSs. Macrocell BSs take charge
of control information for vehicles and small-cell BSs. Small-
cell BSs transmit the desired data to vehicles. In this case,
macrocells form the control plane (C-plane), and small cells
form the user plane (U-plane) in 5G cooperative small-cell net-
works. Without loss of generality, both the control zone signal-
ing and the L1/L2/L3 signaling, which are carried by a physical
downlink shared channel and scheduled by a physical downlink
control channel in the data zone of downlink subframes, are
assumed to belong to the C-plane information. Only the user
traffic data are carried by U-plane subframes in the C/U plane
split architecture [25]. Macrocells with the same radius are
assumed to be regularly deployed in the infinite plane R2 .
Small-cell BSs are assumed to be randomly deployed in the
infinite plane R2 . Moreover, the locations of small-cell BSs
follow independent Poisson point processes Φs with intensity
λs . Every small-cell BS has the same transmission power Ps .
7884 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO. 10, OCTOBER 2016

In this paper, the vehicle is assumed to be associated with of intensity λπ, and the expectation and cumulative distribution
the closest BS, which would suffer the least path loss during function (cdf) of Ri is expressed as
wireless transmissions. Every small cell is assumed to include i
only one BS and a few vehicles. Then, the cell boundary, which E[Ri ] = (3a)
(λπ)
can be obtained through the Delaunay triangulation method Γic (i, λπr2 )
by connecting the perpendicular bisector lines between each FRi (r) = 1 − (3b)
Γ(i)
pair of small-cell BSs [26], splits the plane R2 into irregular
polygons that correspond to different small-cell coverage ar- where Γic (·, ·) is the incomplete Gamma function. When the
eas. Such stochastic and irregular topology forms a so-called differential is operated on (3b), the probability density function
Poisson–Voronoi tessellation (PVT) [27]. An illustration of one (pdf) of Ri is derived as
i
2e−λπr (λπr2 )
2
macrocell scenario is shown in Fig. 1, where each small cell
fRi (r) = . (4)
is denoted as Eq (q = 1, 2, 3, . . .). Despite its complexity, an rΓ(i)
outstanding property of PVT random small-cell networks is that
the geometric characteristics of any small cell Eq coincide with B. Cooperative Probability
that of a typical PVT small cell E1 , according to the Palm theory
[28]. This feature implies that the analytical results for a typical Since the radius of 5G small cells is usually less than 100 m,
PVT small cell E1 can be extended to the whole random small- the vehicle has to frequently execute the handoff operation
cell networks. when the high-speed vehicle is only associated with one small-
Without loss of generality, the initial location of the vehicle cell BS. Even so, it is a great challenge to keep the wireless link
U E 0 located at E1 is assumed as the origin position. The reliability for vehicular communications in 5G small-cell net-
distance between U E0 and the closest small-cell BS BS1 is works. To solve these problems, the cooperative transmission
denoted as R1 . Moreover, the distance between U E0 and the based on adjacent small-cell BSs is a promising candidate. In
ith closest small-cell BS BSi is denoted as Ri (i = 2, 3, 4, . . .). this paper, cooperative small-cell BSs are selected according to
In this paper, adjacent small-cell BSs can cooperatively transmit the following cooperative scheme. Considering that the radius
data to a specified vehicle. Moreover, this paper is focused on of a small cell is much less than the radius of macrocells, the
the downlinks of 5G cooperative small-cell networks. wireless link is assumed be a line-of-sight transmission in this
study. To simplify derivations, the path loss and Rayleigh fading
are considered, but the shadowing effect is ignored in wireless
III. C OOPERATIVE P ROBABILITY channels, as commonly done in the area [30], [31].
IN S MALL -C ELL N ETWORKS Cooperative Scheme: When the ratio of distance Ri to dis-
tance R1 is less than or equal to the given cooperative threshold
In this paper, cooperative small-cell BSs are selected by
ρ, the small-cell BS BSi being Ri apart from vehicle U E0 is se-
distances between vehicle U E0 and adjacent small-cell BSs.
lected for cooperative transmissions, which can be expressed as
How to evaluate the distance distribution of cooperative small-
Ri
cell BSs is the basis for the cooperative transmission of small- ≤ ρ. (5)
cell networks. R1
Therefore, the cooperative probability of BS, i.e., BSi , is
expressed as
A. Distance Distribution of Cooperative Small-Cell BSs   +∞
Ri
In a homogeneous M -dimension Poisson point process with Pr ≤ρ = Pr (Ri ≤ ρy, R1 = y)dy
R1
intensity λ, the probability of finding N nodes in a bounded 0
Borel space A ⊂ RM is expressed as +∞
= Pr (Ri ≤ ρy|R1 = y)Pr (R1 = y)dy
(λA)N
Pr [N nodes in A] = e−λA . (1) 0
N! +∞ 
+∞

−λs D (λs D)k
= e · fR1 (y)dy
For a homogeneous 2-D Poisson point process with intensity λ k!
k=i−1
and A = πr2 , the distance Rn between a point and its nth clos- 0

est point is governed by the generalized Gamma distribution, (6)

i.e., where D = π(ρy) − πy is the area between circles with
2 2

different radii of R1 and ρR1 . Substitute (4) into (6), the

n
2(λπr2 ) cooperative probability of BS BSi is further derived by the
fRn (r) = e−λπr
2
(2)
rΓ(n) following equation:
  ∞ 
Ri
1 − e−λs [π(ρy) −πy ]
2 2
where Γ(·) is the Gamma function. Pr ≤ρ =
Corollary 1 [29]: Let y ∈ R2 and Xi ∈ R2 be the points of R1
a homogeneous point process of intensity λ in the R2 plane
0 
Γ i − 1, λs π(ρy)2 − πy 2
· 2λs πye−λs πy dy. (7)
2
ordered according to their Euclidean distance to y. Then, Ri := ·
Γ(i − 1)
y − Xi 2 has the same distribution as the 1-D Poisson process
GE et al.: VEHICULAR COMMUNICATIONS FOR 5G COOPERATIVE SMALL-CELL NETWORKS 7885

Fig. 2. Cooperative probability with respect to the cooperative threshold Fig. 3. Cooperative probability with respect to the number of cooperative
considering different cooperative small-cell BSs. small-cell BSs and the cooperative threshold ρ.

When k small-cell BSs are close to vehicle U E0 , the coop- Fig. 3 shows the impact of the number of cooperative small-
erative probability of k adjacent small-cell BSs is derived as cell BSs and the cooperative threshold ρ on the cooperative
follows: probability of small-cell BSs. When the number of cooperative
 
Rk Rk+1 small-cell BSs is fixed as 1, i.e., only one small-cell BS is
Pk =Pr ≤ρ∩ >ρ selected for cooperative transmissions, the cooperative proba-
R1 R1
+∞ bility monotonically decreases with the increase in cooperative
= Pr (Rk ≤ ρy, Rk+1 > ρy, R1 = y)dy threshold ρ. When the number of cooperative small-cell BSs
is larger than 1, the cooperative probability first increases with
0
the increase in cooperative threshold ρ. When the cooperative
+∞
probability achieves the maximum, the cooperative probability
= Pr (Rk ≤ ρy, Rk+1 > ρy|R1 = y)Pr (R1 = y)dy
decreases with the increase in cooperative threshold ρ. In the
0 end, the cooperative probability approaches a saturated value
+∞ when threshold ρ is larger than 4. When the threshold is fixed,
(λs D)k−1
= e−λs D · · fR1 (y)dy the cooperative probability decreases with the increase in the
(k − 1)!
0 number of cooperative small-cell BSs.
+∞
e−λs [π(ρy) −πy ]
2 2
=
0

k−1 IV. C OVERAGE P ROBABILITY OF C OOPERATIVE
λs π(ρy)2 − λs πy 2 S MALL -C ELL N ETWORKS
2λs πye−λs πy dy
2
·
(k − 1)!
 k−1 A. Interference Model
+∞ λs π(ρy) 2
− λs πy 2
When the orthogonal frequency-division multiplexing
2λs πye−λs π(ρy) ·
2
= dy.
(k − 1)! scheme is assumed to be adopted by small-cell BSs to support
0 multiuser transmission in a small cell, there is no cochannel
(8)
interference generated from the intracell in this paper. For
vehicle U E0 , no more than one cochannel interfering vehicle
is assumed to exist in each adjacent small cell. Vehicle U E0
C. Performance Analysis of Cooperative Probability is interfered by the downlinks of cochannel vehicles in the
To validate the proposed cooperative probability, some per- adjacent small cells, which is transmitted from their associated
formance analysis is simulated by numerical results in Figs. 2 small-cell BSs. The small-cell BS is equipped with nt anten-
and 3. The intensity of small-cell BSs is configured as λs = nas, and the vehicle is equipped with nr antennas. Hence, in
1/(π × 502 ). Fig. 2 shows the impact of the cooperative thresh- this paper, the vehicular communication is a type of MIMO
old ρ on the cooperative probability of BS BSi . When a communications. Without loss of generality, for vehicle U E0 ,
small-cell BS is selected, the cooperative probability of BS C ⊂ Φs is the cooperative small-cell BS set that can simul-
BSi increases with the increase in cooperative threshold ρ. In taneously transmit data to a given vehicle, and B ⊂ Φs \ C is
this paper, the cooperative small-cell BSs are ordered by the the interfering small-cell BS set. Considering the cooperative
distance between BS BSi and vehicle U E0 . When threshold ρ transmission from adjacent small-cell BSs, the received signal
is fixed, the cooperative probability of BS BSi decreases with at vehicle U E0 is expressed as
the increase in the distance between BS BSi and vehicle U E0 .  √Ps  √Ps
When threshold ρ is larger than 3.5, the cooperative probability y= η Hi0 xi + η Hj0 xj + Z (9)
i∈C Ri
2
j∈B Rj
2
of BS BSi (i = 2, 3, 4) approaches a saturated value.
7886 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO. 10, OCTOBER 2016

where y ∈ Cnr ×1 is the received signal vector at vehicle U E0 , point process with intensity λ(r) = 2πλs r [34]. As a conse-
Ri is the distance between vehicle U E0 and the cooperative quence, the Laplace transform of the aggregate interference at
small-cell BSs, Rj is the distance between vehicle U E0 and the vehicle U E0 is derived as
interfering small-cell BSs, η is the path-loss coefficient, xi ∈
Cnt ×1 is the desired signal vector from the cooperative trans- LIagg (s)
mission small-cell BS BSi , xj ∈ Cnt ×1 is the interfering signal = E(e−sIagg )
⎡ ⎛  n n ⎞⎤
vector from the adjacent interfering small-cell BS BSj , and ∞  r  t
Z ∈ Cnt ×1 is the additive white Gaussian noise with variance Ps −η
= E ⎣exp ⎝−s R |hj,m,n |2 ⎠⎦
σ 2 in wireless channels. Hi0 ∈ Cnr ×nt is the small-scale fad- nt j m=1 n=1
j=k+1
ing channel matrix between vehicle U E0 and the cooperative    
Ps
small-cell BS BSi , hi,m,n (m = 1, 2, . . . , nr ; n = 1, 2, . . . , nt ) = EΦ,gj Π exp −s Rj−η gj
nt
is the element of the channel matrix Hi0 and is governed by a ⎡ j>k ⎤
 
complex Gaussian distribution, i.e., hi,m,n ∼ CN (0, 1), and its −η
= exp⎣−2πλs Egj 1 − e−s nt r gj rdr⎦
(a) Ps

magnitude |hi,m,n | is a Rayleigh-distributed random variable

[20], where hi,m,n is the channel coefficient between the mth ⎡ r>Rk
⎛∞
receiving antenna at vehicle U E0 and the nth transmission   nt nr −1
antenna at the cooperative small-cell BS BSi ; Hj0 ∈ Cnr ×nt is = exp ⎣−2πλs ⎝ g e−g
Γ(nt nr )
the small-scale fading channel matrix between vehicle U E0 and 0
r>Rk
⎞ ⎤
the interfering small-cell BS BSj , hj,m,n (m = 1, 2, . . . , nr ; 
× 1 − e−s nt r−η g dg ⎠ rdr⎦
Ps
n = 1, 2, . . . , nt ) is the element of the channel matrix Hj0 and
is governed by a complex Gaussian distribution, i.e., hj,m,n ∼
CN (0, 1), and its magnitude |hj,m,n | is Rayleigh-distributed ⎡ ⎛ ⎞ ⎤
 ∞
−2πλ Ps −η
random variables, where hj,m,n is the channel coefficient be- = exp⎣ ⎝1− g nt nr −1 e ( nt
− ) dg ⎠rdr⎦
s s r +1 g
tween the mth receiving antenna at vehicle U E0 and the nth Γ(nt nr )
transmission antenna at the interfering small-cell BS BSj . ⎡ r>Rk
⎛ 0 ⎞ ⎤

When k cooperative small-cell BSs are assumed to jointly 1
transmit data to vehicle U E0 and considering maximum ratio = exp⎣−2πλs ⎝1 −  nt nr
⎠ rdr⎦
Ps −η
1 + s nt r
transmission/maximum ratio combining [32], [33], the SINR r>R k

received by vehicle U E0 is derived as (12)

 where E(·) is the expectation operation, and (a) is due to the

Ps Ri−η |Hi,0 |2
i∈C
probability-generating function for a Poisson point process.
SINRc = 
σ2 + Ps Rj−η |Hj,0 |2
j∈B
B. Coverage Probability

k 
Ps
nt Ri−η λmax Hi,0 HHi,0
For cooperative transmissions of small-cell BSs, the cover-
≈ i=1
 n n  age of cooperative small-cell BSs can be extended from every

∞
−η r t coverage of cooperative small-cell BSs. The extended coverage
σ2 + Ps
nt Rj |hj,m,n | 2
j=k+1 m=1 n=1 of cooperative small-cell BSs can provide for a better reliable
  link service for vehicular communications. When the outage

k 
nr 
nt
Ps
nt Ri−η |hi,m,n |2 threshold is configured as ε for vehicle links, the coverage
≈ i=1
 n n
m=1 n=1
 (10) probability of k cooperative small-cell BSs is expressed as

∞
−η r t
σ2 + Ps
nt Rj |hj,m,n | 2
Pkc =Pr (SINRc > ε)
j=k+1 m=1 n=1 ⎛ ⎞
Ps 
k
−η
⎜ nt Ri gi ⎟
where λmax (Hi,0 HH ⎜ i=1 ⎟
i,0 ) is the maximum singular value of the =Pr ⎜ > ε⎟. (13)
H
matrix Hi,0 Hi,0 . Furthermore, the interference aggregated at ⎝ σ 2 + Iagg ⎠
vehicle U E0 can be expressed as

∞
  However, the analytical expression cannot be derived for (13)
Ps  −η 
nr 
nt
Iagg = Rj |hj,m,n | 2
. (11) when distance Ri is a random variable. In most cases, cooper-
nt m=1 n=1
ative transmissions are related with the limited adjacent small-
j=k+1
cell BSs in 5G cooperative small-cell networks. Moreover, the
nr nt number of cooperative small-cell BSs is less than or equal to
To simplify the derivation, let gi = m=1 n=1 |hi,m,n |2 five in realistic scenarios. When vehicle U E0 is assumed to
Δ
and Φ = {Rj |j ∈ B}. From the distribution of hi,m,n , we can be located at the edge of small cells, the distance between
derive that the pdf of gi is fgi (x) = (xnt nr −1 /Γ(nt nr ))e−x . vehicle U E0 and cooperative small-cell BSs is approximated
Based on the mapping theorem, Φ is an inhomogeneous Poisson to be equal. Therefore, the distance between vehicle U E0 and
GE et al.: VEHICULAR COMMUNICATIONS FOR 5G COOPERATIVE SMALL-CELL NETWORKS 7887

cooperative small-cell BSs is configured as D in the following Based on (12), the differentiation of the aggregate interfer-
derivations. To simplify the derivation, the transmission power ence at vehicle U E0 is derived by the following equation:
of a small-cell BS is normalized as 1, and the noise is ignored
(1) d
in wireless channels considering that the noise power is obvi- LIagg (s) = LI (s)
ds agg⎛ ⎞
ously less than the desired signal power and the interference
∞  
power [35]. As a consequence, the coverage probability of d ⎜ 1 ⎟
k cooperative small-cell BSs is further derived as = exp ⎝−πλs 1− −η/2 nt nr du⎠
ds (su + 1)
R2k
  ⎛ ⎞

k ∞  
Pkc = Pr Ri−η gi > εIagg ⎜ 1 ⎟
= exp ⎝−πλs 1− −η/2 nt nr du⎠
i=1 (su + 1)
  R2k
(a) 
k ⎛ ⎞(1)
= Pr D−η gi > εIagg ∞  
⎜ 1 ⎟
i=1 · ⎝−πλs 1− n n du⎠
 kn n −1  (14) (su−η/2 + 1) t r

t r
(εDη )n R2k
(b)
= ED EIagg n
Iagg e−εD
η
Iagg ⎛ s⎞
n! ∞  
n=0 ⎜ 1 ⎟
kn n −1  = exp ⎝−πλs 1− n n du⎠
 (su−η/2 + 1) t r
(c)
t r
(−εDη )n (n)
= ED LIagg (εDη ) ⎛
R2k
⎞
n=0
n! ∞
⎜ nt nr u−η/2 ⎟
· ⎝−πλs n n +1 du⎠
where condition (a) is the assumption that the distance be- su−η/2 + 1 t r
R2k
tween vehicle U E0 and the cooperative small-cell BSs is ⎛ ⎞
∞
equal, condition (b) is based on the cdf of the Gamma dis- ⎜ nt nr u−η/2 ⎟
= ⎝−πλs n n +1 du⎠ × LI (s)
tribution, and condition (c) is the Laplace transform prop- su−η/2 + 1 t r # $% &
erty EIagg [Iagg
n
e−sIagg ] = (−1)n (dn /dsn )LIagg (s). Let xn = R2k v(s)
(n) # $% &
((−εDη )n /n!)LIagg (εDη ), (14) is simply expressed as ω(s)
(16)
kn n −1   

t r t nr −1
kn
Pkc = ED xn = ED x0 + xn . (15) Furthermore, the nth-order derivative of (12) is derived by
n=0 n=1 (17), shown at the bottom of the page.

 (n−1) 
n−1
(n) (1)
LIagg (s) = LI (s) = i
Cn−1 ω (n−1−i) v (i)
i=0


n−1
(i)
= i
Cn−1 ω (n−1−i) × LI (s)
i=0
⎛ ⎞(n−1−i)

n−1 ∞ 
⎜ −nt nr −1 ⎟
u−η/2 (1 + su−η/2 )
(i)
= πλs nt nr i
Cn−1 ⎝− du⎠ × LI (s)
i=0
R2k
⎛

n−1 ∞ 
⎜ (n − i − 1 + nt nr )!
= πλs nt nr i
Cn−1 ⎝− u−η/2 (−1)n−i−1
i=0
(nt nr )!
R2k
⎞

−nt nr −1−n+i+1 n−i−1 ⎟
×(1 + su−η/2 ) (u−η/2 )
(i)
du⎠ × LI (s)

⎛ ⎞

n−1 ∞  −η/2 n−i

(n − i − 1 + nt nr )! ⎜ (u ) ⎟ (i)
= πλs i
Cn−1 (−1)n−i ⎝ du⎠ × LI (s) (17)
i=0
(nt nr − 1)! (1 + su−η/2 )
n−i+nt nr
R2k
7888 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO. 10, OCTOBER 2016

Let v −η/2 = su−η/2 , then du = sη/2 dv. Equation (17) can algebra. Let
be further derived as
xknt nr = [x1 , x2 , . . . , xknt nr ]T (21a)
(n)

n−1
T
yknt nr = [y1 , y2 , . . . , yknt nr ]
LIagg (s) = πλs i
Cn−1 · (−1)n−i 
i=0 nt nr (1 + nt nr )
= nt nr k1 , k2 ,
(n − i − 1 + nt nr )! 2/η−n+i 2
· s T
(nt nr − 1)! nt nr
nt nr Ckn t nr −1+nt nr
..., kknt nr (21b)
∞ n−i knt nr
(v −η/2 ) (i)
× dv × LIagg (s).
(1 + v −η/2 )n−i+nt nr then (20) can be represented in a matrix form as (22a)–(22c),
ε−2/η
shown at the bottom of the next page.
(18)
Since Gknt nr is a strictly lower triangular matrix, we have
Gnknt nr = 0, n ≥ knt nr . According to this property, after iter-
Substituting s = εDη into (12), the following expression is
ating, xknt nr can be rewritten as
derived:
xknt nr = ax0 yknt nr +aGknt nr xknt nr
x0 = LIagg (s) = exp(−πλs k0 D2 ) (19a)
= ax0 yknt nr+aGknt nr (ax0 yknt nr+aGknt nr xknt nr )
= ax0 yknt nr +a2 x0 Gknt nr yknt nr+a2 G2knt nr xknt nr
with
=···
∞   
knt nr
2 1
k0 = ε η 1− dv. (19b) = an x0 Gn−1
knt nr yknt nr . (23)
(1 + v −η/2 )nt nr n=1
ε−2/η

Similarly, denote xknt nr −1 = [x1 , x2 , . . . , xknt nr −1 ]T , then

When n ≥ 1, the following expression is derived by formula
(20a), shown at the bottom of the page, with t nr −1
kn
xknt nr −1 = an x0 Gn−1
knt nr −1 yknt nr −1 . (24)
∞ n=1
2 1
ki = ε η
η i
dv i ≥ 1. (20b)
n n
1 + v 2 (1 + v −η/2 ) t r  t nr −1
ε−2/η Define sum(xknt nr −1 ) = kn n=1 xn , which is the sum of
elements in the vector xknt nr −1 . In the end, the coverage
Based on (20a), a linear recurrence relation of xn is derived probability of k cooperative small-cell BSs can be expressed in
for the explicit expression of the coverage probability via linear an explicit form as (25), shown at the bottom of the next page.

sn (n)
xn = (−1)n LIagg (s)
n!
⎛ ∞ − η n−i ⎞

n−1 
(n − i − 1 + n n )! ν 2
s η +i ⎝ dv ⎠ LIagg (s)
t r 2 (i)
i
= πλs Cn−1 (−1)i 
i=0
n!(n t n r − 1)! 1 + ν − η2 n−i+nt nr
ε−2/η
⎛ ∞ − η n−i ⎞

2 
n−1
i i!(n − i − 1 + n n
t r ⎝)! ν 2 i
⎠ (−1)i s L(i) (s)
= πλs s η Cn−1  dv
i=0
n!(nt nr − 1)! 1 + ν− 2
η n−i+nt nr
i! Iagg
ε−2/η
⎛ ∞ − η n−i ⎞

n−1 
(n − 1)! i!(n − i − 1 + nt nr )! ⎝ ν 2
dv ⎠ xi
2
= πλs s η

i=0
i!(n − i − 1)! n!(n t n r − 1)! 1 + ν − η2 n−i+nt nr
ε−2/η
⎛ ⎞
∞ 
 nt nr Cn−1−i+n
n−1 nt nr
2 ⎜ ν− 2
η n−i
⎟
εη ⎜ ⎟ xi
s=εDη t nr
= πλs D2 ⎝ n−i+n n
dv ⎠
n 1 + ν −2
η t r
i=0 2 −
ε η


n−1 nt nr
nt nr Cn−1−i+n t nr
= πλs D2 kn−i xi (20a)
i=0
n
GE et al.: VEHICULAR COMMUNICATIONS FOR 5G COOPERATIVE SMALL-CELL NETWORKS 7889

Without loss of generality, a scenario of three small-cell

cooperative BSs is considered as follows. In this case, the
distance D between vehicle U E0 and the cooperative small-cell
BSs is given by [35]

fD (D) = 2(λπ)2 D3 e−λπD .

2
(26)

Substitute (26) into (25), the coverage probability of three

cooperative small-cell BSs is simply derived by (27), shown at
the bottom of the page.
C. Performance Analysis of Coverage Probability
Based on the proposed coverage probability of three coop-
erative small-cell BSs in (27), some performance evaluations
are numerically analyzed in Figs. 4–7. In the following cover-
age probability analysis, default parameters are configured as
follows: The antenna number at the small-cell BS is nt = 4,
the antenna number at the vehicle is nr = 2, the small-cell BS
transmission power is normalized as 1, the path-loss coefficient
is η = 4, the intensity of small-cell BSs is λs = 1/(π × 502 ),
the outage threshold of vehicle links is ε = 0 dB [36], and the
radius of a small cell is 50 m.
Fig. 4 shows the impact of the outage threshold and the path-
loss coefficient on the cooperative probability. When the path-
loss coefficient is fixed, the coverage probability decreases with
the increase in the outage threshold. When the outage threshold
Fig. 4. Coverage probability with respect to the outage threshold considering
different path-loss coefficients.
is fixed, the coverage probability increases with the increase
in the path-loss coefficient. This result can be explained as
follows: The signal propagation attenuation is obviously in-
creased with the increase in the propagation range when the
path-loss coefficient is increased. Compared with cooperative
small-cell BSs, the interfering small-cell BSs are far away from
the received vehicle. Therefore, the interference attenuation is
larger than the desired signal attenuation. When the outage
threshold and the transmission power of small-cell BSs are
fixed, the coverage probability of three cooperative small-cell
BSs increases with the increase in the path-loss coefficient.
Fig. 5 analyzes the impact of the number of transmission
antennas at cooperative small-cell BSs on the coverage prob-
ability. When the outage threshold is less than 2.5 dB, the

xknt nr = ax0 yknt nr + aGknt nr xknt nr (22a)

2
a = πλs D (22b)
⎡ ⎤
0
⎢ 1 ⎥
⎢ 2 nt nr k1 0 ⎥
⎢ nt nr (1+nt nr )
k2 nt nr ⎥
Gknt nr =⎢ 3 3 k1 0 ⎥ (22c)
⎢ .. ⎥
⎢ ⎥
⎣ n n
.
n n
0 ⎦
nt nr Cknt t nrr +nt nr −2 nt nr Cknt t nrr +nt nr −3
knt nr kknt nr −1 knt nr kknt nr −2 ··· nt nr
knt nr k1 0

 
t nr −1
kn
Pkc = ED x0 + xn
n=1
∞  
t nr −1
kn
= fD (D) x0 + xn dD
0 n=1
⎛ ⎛ ⎞ ⎞
∞ ∞  
1
fD (D)⎝exp ⎝−πλs D2 ε dv ⎠ + sum (xknt nr −1 )⎠ dD
2
= η 1− nt nr (25)
(1 + v −η/2 )
0 ε−2/η

⎞ ⎞
∞   ∞  
1
2(λπ)2 D3 e−λπD dv ⎠ + sum (xknt nr −1 )⎠ dD
2 2
P3c = exp −πλs D2 ε η × 1− nt nr (27)
(1 + v −η/2 )
0 ε−2/η
7890
Fig. 5. Coverage probability with respect to the number of transmission
antennas at cooperative small-cell BSs considering different outage thresholds.
Fig. 6. Coverage probability with respect to the radius of small cells consider-
ing different transmission antennas at cooperative small-cell BSs.
Fig. 7. Coverage probability with and without cooperative communication
schemes.
coverage probability increases with the increase in the number
of transmission antennas at cooperative small-cell BSs. When
the outage threshold is larger than or equal to 2.5 dB, the
coverage probability decreases with the increase in the number
of transmission antennas at cooperative small-cell BSs.
Fig. 6 shows the impact of the radius of small cells and the
number of transmission antennas at cooperative small-cell BSs
on the coverage probability. When the number of transmission
antennas is fixed, the coverage probability decreases with the
increase in the radius of small cells. When the radius of a small
cell is fixed, the coverage probability increases with the increase
in the number of transmission antennas at cooperative small-
cell BSs.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO. 10, OCTOBER 2016
Fig. 7 shows the coverage probability with and without co-
operative transmission schemes in small-cell networks. When
the cooperative transmission scheme is adopted in small-cell
networks, the number of cooperative small-cell BSs is config-
ured as k = 3. When the cooperative communication scheme is
not adopted in small-cell networks, the number of cooperative
small-cell BSs is configured as k = 1. Compared with two
curves in Fig. 7, the cooperative transmission scheme can
improve the coverage probability in small-cell networks, and
the gain of coverage probability with the cooperative transmis-
sion scheme achieves the maximum when the outage threshold
is −1 dB.
V. V EHICLE M OBILITY A NALYSIS OF C OOPERATIVE
S MALL -C ELL N ETWORKS
A. Vehicle Handoff Rate of Cooperative Small-Cell Networks
When the initial position of the vehicle is assumed as the
origin, the vehicle will arrive at a new position after a minimal
slot τ . Without loss of generality, the Gauss–Markov mobil-
ity model is adopted for the vehicle mobility in this study.
Assuming that the vehicle moves with the velocity of ϕn−1
and the direction of ϑn−1 at the (n − 1)th time instant, then
the velocity and direction of the nth)time instant are calcu-
lated by ϕn = αϕn−1 + (1 − α)ϕ̄ + (1 − α2 )ϕxn−1 , ϑn =
)
αϑn−1 + (1 − α)ϑ̄ + (1 − α2 )ϑxn−1 , respectively, where ϕ̄
and ϑ̄ represent the mean value of velocity and direction as
n → ∞, and ϕxn−1 and ϑxn−1 are random variables governed
by Gaussian distribution. To simplify the derivation, we assume
that α = 1. In this case, the vehicle keeps the velocity ϕ and the
direction ϑ constant. When the distance between vehicle U E0
and the ith closest small-cell BS BSi is assumed as Ri in the
last slot, the new distance between vehicle U E0 and the small-
cell BS BSi in the current slot is expressed as
*
Rinew = Ri2 + (ϕτ )2 + 2Ri ϕτ cos ϑ. (28)
If we let ξ = arg min Rinew , the distance between vehicle U E0
i
and the closest small-cell BS BSξ is denoted as Rξnew in the
current slot. The vehicle passes through different cooperative
small cells when the vehicle moves for a slot τ . The trigger of
vehicle handoff in a slot τ is expressed as
  
Rinew Ri
ΔHτ = 1 ∃i ∈ N , +
>ρ∩ ≤ρ
Rξnew R1
 
Rinew Ri
∪ ≤ρ∩ >ρ (29)
Rξnew R1
where 1[·] is the indicator function, which is equal to 1 when
the condition inside the bracket is satisfied and 0 otherwise.
For a long time, e.g., T = tτ , t  1, and t ∈ N + , the handoff
number of the vehicle ΔHT is the sum of (29) in time T .
Furthermore, the vehicle handoff rate in cooperative small-cell
networks is expressed by
ΔHT
HO = . (30)
T
GE et al.: VEHICULAR COMMUNICATIONS FOR 5G COOPERATIVE SMALL-CELL NETWORKS 7891

Furthermore, the handoff rate of the type-l vehicle traffic is

expressed as
HOl = pA (l) · HO. (33)
Let al be the flow rate of type-l vehicle traffic, and the av-
erage handoff duration is χ, and the overhead of the type-l
vehicle traffic generated by a handoff over the X2-U link is
expressed as
βl = al · χ. (34)
When all handoff requests are assumed to be accepted in
small-cell networks, the overhead traffic of X2-U links is
expressed as
Fig. 8. Vehicular handoff rate with respect to the vehicular velocity and the
cooperative threshold.
L
TX2−U = βl · HOl . (35)
l=1
Fig. 8 analyzes the impact of the vehicular velocity and the
cooperative threshold on the vehicular handoff rate. Without Based on (31) and (35), the total overhead traffic of X2 links
loss of generality, the traditional scenario of fixed two coop- in cooperative small-cell networks can be expressed as
erative small cells is compared with the proposed multicell
TX2 = TX2−C + TX2−U . (36)
cooperative scenarios in Fig. 8. When the cooperative threshold
is fixed, the vehicular handoff rate increases with the increase in Without loss of generality, a change in the cooperative status
the vehicular velocity in both scenarios. When ρ = 1, it means is triggered by a vehicular handoff in cooperative small-cell
only one small cell transmits signals to the vehicle. In this networks. In general, the change in vehicular cooperative status
case, the curve of the multicell cooperative scenario coincides conduces to some overhead traffic in cooperative small-cell
with the curve of the scenario with fixed two cooperative networks. The size of overhead traffic is expressed as
small cells. When ρ = 1, the number of cooperative small cells
C = k · TX2 . (37)
is changed, accounting for the distance between the vehicle
and the small-cell BSs for the proposed multicell cooperative Therefore, the expectation of the size of overhead traffic is
scenarios. Therefore, the vehicular handoff rate of the proposed expressed as

multicell cooperative scenarios is obviously larger than the E[C] = Pk · k · TX2 . (38)
vehicular handoff rate of the traditional scenario with fixed two k
cooperative small cells.
When the SINR of vehicular communication is configured as
the outage threshold ε, the vehicular communication capacity is
B. Vehicle Overhead Ratio of Cooperative
expressed as
Small-Cell Networks ∞ 

When the vehicle passes through different small cells, the co- ∂= Pc · Pk · Bw · log(1 + ε)
k
(39)
operative status has to be changed. Based on small-cell network k=1
architectures [37], the cooperative link of small cells is defined where Bw is the bandwidth for vehicle wireless links.
as the X2 link, which has been used for transmitting handoff To evaluate the cooperative communication overhead for
and cooperative information in cooperative small-cell networks. vehicular communications, the vehicular overhead ratio in
The X2 link of cooperative small-cell networks is composed of cooperative small-cell networks is defined as
two parts, i.e., the X2-U link and the X2-C link. In general,
E[C]
the X2-U link is used for transmitting handoff information, Ω= . (40)
and the X2-C link is used for transmitting cooperative control ∂
information among small cells. Therefore, the overhead traffic Considering the cooperative transmission in small-cell net-
of X2-C links is expressed as works, the gain, i.e., the vehicular communication capacity, and
the cost, i.e., the vehicular overhead ratio, can be evaluated by
TX2−C = δ · HO (31)
(39) and (40).
where δ is the average cooperative control data size per small
cell when a handoff is triggered in small-cell networks [38]. C. Performance Analysis
When the total of L types of traffic is assumed for vehicular
Based on the proposed vehicular communication capacity
communications, ψl and ζl−1 , 1 ≤ l ≤ L are the traffic arrive
and vehicular overhead ratio, some performance evaluations are
rate and the average vehicle session duration of the type-l
numerically analyzed in Figs. 9–11. In the following analysis,
vehicle traffic. Based on the queue theory, the active probability
the default parameters are configured as follows: The outage
of the type-l vehicle traffic is expressed as
threshold is ε = 0 dB, the bandwidth is Bw = 10 Mb/s, the
ψl average cooperative control data size is δ = 480 bits, the path-
pA (l) = . (32)
(ψl + ζl ) loss coefficient is η = 4, the time slot is τ = 15 ms, the radius
7892 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO. 10, OCTOBER 2016
Fig. 9. Vehicular communication capacity with respect to the radius of small
cells considering different cooperative thresholds.
Fig. 10. Vehicular overhead ratio with respect to the vehicular velocity consid-
ering different cooperative thresholds.
Fig. 11. Vehicular overhead ratio with respect to the radius of small cells
considering different cooperative thresholds.
of a small cell is 50 m, the vehicular velocity is ϕ = 10 m/s,
and the handoff duration is χ = 0.05 s. Without loss of gen-
erality, two types of traffic are considered in this paper. The
type-1 traffic has the following configuration parameters: a1 =
12.2 Kb/s, ψ1 = 1.5, and ζ1−1 = 0.03333. The type-2 traffic
has the following configuration parameters: a2 = 353.8 Kb/s,
ψ2 = 0.5, and ζ2−1 = 0.05 [38].
Fig. 9 evaluates the vehicular communication capacity with
respect to the radius of small cells considering different coop-
erative thresholds. When the cooperative threshold is fixed, the
vehicular communication capacity decreases with the increase
in the radius of small cells. When the radius of small cells
is fixed, the vehicular communication capacity increases with
the increase in the cooperative threshold. Fig. 10 shows the
vehicular overhead ratio with respect to the vehicular velocity
considering different cooperative thresholds. When the cooper-
ative threshold is fixed, the vehicular overhead ratio increases
with the increase in the vehicular velocity. When the vehicular
velocity is fixed, the vehicular overhead ratio increases with
the increase in the cooperative threshold. The capacity gain
is increased when the cooperative threshold is less than 1.5.
When the cooperative threshold is larger than or equal to 1.5,
the cooperation probability of k small cells decreases with the
increase in the cooperative threshold. This result is validated
in Fig. 3. It implies that the number of cooperative small cells
will be limited with the increase in the cooperative threshold
in practical applications. Therefore, the capacity gain is limited
with the increase in the cooperative threshold. Fig. 11 analyzes
the vehicular overhead ratio with respect to the radius of small
cells considering different cooperative thresholds. When the
radius of small cells is fixed, the vehicular overhead ratio
increases with the increase in the cooperative threshold. When
the cooperative threshold is fixed, the vehicular overhead ratio
first decreases with the increase in the radius of small cells.
Numerical results show that there exist turning points, i.e., the
radius of small cells is 75, 80, and 85 m, corresponding to
the cooperative thresholds 1.5, 1.2, and 1. When the radius
of small cells is larger than the turning points, the vehicular
overhead ratio increases with the increase in the radius of small
cells. Therefore, there exists a minimal value for the vehicu-
lar overhead ratio under different cooperative thresholds. The
minimal vehicle overhead ratio is 4.5 × 10−4 , 3.4 × 10−4 , and
1.3 × 10−4 , corresponding to the cooperative thresholds of 1,
1.2, and 1.5. This result provides a guideline for optimizing the
vehicular overhead ratio in 5G cooperative small-cell networks
with different radii of small cells.
VI. C ONCLUSION
In this paper, the vehicular mobility performance has been
analyzed for 5G cooperative MIMO small-cell networks con-
sidering cochannel interference. Based on distances between
the vehicle and cooperative small-cell BSs, the cooperative
probability and the coverage probability have been derived
for 5G cooperative small-cell networks where small-cell BSs
follow Poisson point process distributions. Furthermore, the
vehicular handoff rate and the vehicular overhead ratio have
been proposed to evaluate the vehicular mobility performance
in 5G cooperative MIMO small-cell networks. Numerical re-
sults indicate that the cooperative transmission scheme in-
creases the vehicular communication capacity and the vehicular
handoff rate in 5G cooperative MIMO small-cell networks.
Therefore, there exists a tradeoff between vehicular commu-
nication capacity and vehicular handoff ratio. By evaluating
the vehicular overhead ratio, numerical results show that there
exists a minimum vehicular overhead ratio for 5G cooperative
MIMO small-cell networks considering different cooperative
thresholds. These results provide a guideline for optimizing
vehicular communications in 5G cooperative MIMO small-cell
networks.
GE et al.: VEHICULAR COMMUNICATIONS FOR 5G COOPERATIVE SMALL-CELL NETWORKS
R EFERENCES
[1] S. Chen and J. Zhao, “The requirements, challenges and technologies
for 5G of terrestrial mobile telecommunication,” IEEE Commun. Mag.,
vol. 52, no. 5, pp. 36–43, May 2014.
[2] M. Chen, Y. Zhang, Y. Li, S. Mao, and V. Leung, “EMC: Emotion-aware
mobile cloud computing in 5G,” IEEE Netw., vol. 29, no. 2, pp. 32–38,
Mar. 2015.
[3] X. Ge, S. Tu, G. Mao, C.-X. Wang, and T. Han, “5G ultra-dense cel-
lular networks,” IEEE Wireless Commun., vol. 23, no. 1, pp. 72–79,
Feb. 2016.
[4] L. Maviel, Y. Lostanlen, and J. Gorce, “A hybrid propagation model for
large-scale variations caused by vehicular traffic in small cells,” in Proc.
IEEE GLOBECOM, Dec. 2012, pp. 5021–5026.
[5] X. Ge, H. Cheng, M. Guizani, and T. Han, “5G wireless backhaul net-
works: Challenges and research advances,” IEEE Netw., vol. 28, no. 6,
pp. 6–11, Nov. 2014.
[6] Y. Zhu, Y. Bao, and B. Li, “On maximizing delay-constrained coverage of
urban vehicular networks,” IEEE J. Sel. Areas Commun., vol. 30, no. 4,
pp. 804–817, May 2012.
[7] M. J. Khabbaz, W. F. Fawaz, and C. M. Assi, “A probabilistic and
traffic-aware bundle release scheme for vehicular intermittently con-
nected networks,” IEEE Trans. Commun., vol. 60, no. 11, pp. 3396–3406,
Aug. 2012.
[8] M. Chen, Y. Hao, Y. Li, C. Lai, and D. Wu, “On the computation offload-
ing at ad hoc cloudlet: Architecture and service models,” IEEE Commun.,
vol. 53, no. 6, pp. 18–24, Jun. 2015.
[9] C. Li, J. Zhang, M. Haenggi, and K. B. Letaief, “User-centric inter-
cell interference nulling for downlink small cell networks,” IEEE Trans.
Commun., vol. 63, no. 4, pp. 1419–1431, Feb. 2015.
[10] J. Yu, Y. Wang, X. Lin, and Q. Zhang, “Statistical analysis of capacity
in joint processing coordinated multi-point systems,” in Proc. IEEE/CIC
ICCC, Oct. 2014, pp. 496–501.
[11] M. Sepulcre, J. Mittag, P. Santi, and H. Hartenstein, “Congestion and
awareness control in cooperative vehicular systems,” Proc. IEEE, vol. 99,
no. 7, pp. 1260–1279, Jun. 2011.
[12] K. Shafiee, A. Attar, and V. C. M. Leung, “Optimal distributed vertical
handoff strategies in vehicular heterogeneous networks,” IEEE J. Sel.
Areas Commun., vol. 29, no. 3, pp. 534–544, Mar. 2011.
[13] E. Bernal-Mor, V. Pla, J. Martinez-Bauset, and L. Guijarro, “Performance
analysis of two-tier wireless networks with dynamic traffic, backhaul
constraints and terminal mobility,” IEEE Trans. Veh. Technol., vol. 65,
no. 1, pp. 241–250, Jan. 2016, doi: 10.1109/TVT.2015.2397317.
[14] L. Li, Y. Xu, and L. Ma, “Vertical handoff strategy on achieving through-
put in vehicular heterogeneous network,” in Proc. IEEE VTC—Spring,
May 2014, pp. 1–5.
[15] X. Li, H. Zhang, and Q. Xu, “Optimal access scheme for mobile vehicular
small cells in layered heterogenous networks,” in Proc. IEEE IC-NIDC,
Sep. 2014, pp. 46–50.
[16] M. F. Feteiha, M. H. Qutqut, and H. S. Hassanein, “Outage probability
analysis of mobile small cells over LTE-A networks,” in Proc. IEEE
IWCMC, Aug. 2014, pp. 1045–1050.
[17] D. Gesbert, S. Hanly, H. Huang, and S. Shamai Shitz, “Multi-cell MIMO
cooperative networks: A new look at interference,” IEEE J. Sel. Areas
Commun., vol. 28, no. 9, pp. 1380–1408, Oct. 2010.
[18] F. Baccelli and A. Giovanidis, “A stochastic geometry framework for
analyzing pairwise-cooperative cellular networks,” IEEE Trans. Wireless
Commun., vol. 14, no. 2, pp. 794–808, Sep. 2014.
[19] R. Tanbourgi, S. Singh, J. G. Andrews, and F. K. Jondral, “A
tractable model for noncoherent joint-transmission base station cooper-
ation,” IEEE Trans. Wireless Commun., vol. 13, no. 9, pp. 4959–4973,
Jul. 2014.
[20] A. H. Sakr and E. Hossain, “Location-aware cross-tier coordinated mul-
tipoint transmission in two-tier cellular networks,” IEEE Trans. Wireless
Commun., vol. 13, no. 11, pp. 6311–6325, Aug. 2014.
[21] Q. Zhang and C. Yang, “Transmission mode selection for downlink co-
ordinated multipoint systems,” IEEE Trans. Veh. Technol., vol. 62, no. 1,
pp. 465–471, Sep. 2012.
[22] Q. Li, R. Q. Hu, Y. Qian, and G. Wu, “Intracell cooperation and resource
allocation in a heterogeneous network with relays,” IEEE Trans. Veh.
Technol., vol. 62, no. 4, pp. 1770–1784, May 2013.
[23] S. Agarwal, S. De, S. Kumar, and H. M. Gupta, “QoS-aware downlink
cooperation for cell-edge and handoff users,” IEEE Trans. Veh. Technol.,
vol. 64, no. 6, pp. 2512–2527, Aug. 2014.
[24] C. Liu, B. Natarajan, and H. Xia, “Small cell base station sleep strate-
gies for energy efficiency,” IEEE Trans. Veh. Technol., vol. 65, no. 3,
pp. 1652–1661, Mar. 2016.
7893
[25] H. Song, X. M. Fang, and L. Yan, “Handover scheme for 5G CU plane
split heterogeneous network in high-speed railway,” IEEE Trans. Veh.
Technol., vol. 63, no. 9, pp. 4633–4646, Apr. 2014.
[26] J. S. Ferenc and Z. Neda, “On the size distribution of Poisson Voronoi
cells,” Phys. A, Statist. Mech. Appl., vol. 385, no. 2, pp. 518–526,
Nov. 2007.
[27] D. Stoyan, W. S. Kendall, and J. Mecke, Stochastic Geometry and Its
Applications, 2nd ed. New York, NY, USA: Wiley, 1996.
[28] S. G. Foss and S. A. Zuyev, “On a Voronoi aggregative process re-
lated to a bivariate Poisson process,” Adv. Appl. Probab., vol. 28, no. 4,
pp. 965–981, Dec. 1996.
[29] M. Haenggi, “On distances in uniformly random networks,” IEEE Trans.
Inf. Theory, vol. 51, no. 10, pp. 3584–3586, Oct. 2005.
[30] C. Wang, S. J. Tang, X. Y. Li, and C. J. Jiang, “Multicast capacity
of multihop cognitive networks,” in Proc. IEEE MOBHOC, Oct. 2009,
pp. 274–283.
[31] C. Fan, Y. J. Zhang, and X. Yuan, “Dynamic nested clustering for parallel
PHY-layer processing in Cloud-RANs,” IEEE Trans. Wireless Commun.,
vol. 15, no. 3, pp. 1881–1894, Mar. 2016.
[32] P. A. Dighe, R. K. Mallik, and S. S. Jamuar, “Analysis of transmit–receive
diversity in Rayleigh fading,” IEEE Trans. Commun., vol. 51, no. 4,
pp. 694–703, Apr. 2003.
[33] X. Ge, K. Huang, C. Wang, X. Hong, and X. Yang, “Capacity analy-
sis of a multi-cell multi-antenna cooperative cellular network with co-
channel interference,” IEEE Trans. Wireless Commun., vol. 10, no. 10,
pp. 3298–3309, Oct. 2011.
[34] M. Haenggi, G. J. Andrews, F. Baccelli, O. Dousse, and M. Franceschetti,
“Stochastic geometry and random graphs for the analysis and design
of wireless networks,” IEEE J. Sel. Areas Commun., vol. 27, no. 7,
pp. 1029–1046, Sep. 2009.
[35] G. Nigam, P. Minero, and M. Haenggi, “Coordinated multipoint joint
transmission in heterogeneous networks,” IEEE Trans. Commun., vol. 62,
no. 11, pp. 4134–4146, Nov. 2014.
[36] C. Li, J. Zhang, and K. B. Letaief, “Throughput and energy efficiency
analysis of small cell networks with multi-antenna base stations,” IEEE
Trans. Wireless Commun., vol. 13, no. 5, pp. 2505–2517, Mar. 2014.
[37] J. Robson, “Small cell backhaul requirements,” Next Gener. Mobile Netw.
(NGMN) White Paper, Frankfurt, Germany, Jun. 2012, pp. 1–40.
[38] I. Widjaja and H. La Roche, “Sizing X2 bandwidth for inter-connected
eNBs,” in Proc. IEEE VTC—Fall, Sep. 2009, pp. 1–5.
[39] W. Zhang et al., “Multi-hop connectivity probability in infrastructure-
based vehicular networks,” IEEE J. Sel. Areas Commun., vol. 30, no. 4,
pp. 740–747, Apr. 2012.
Xiaohu Ge (M’09–SM’11) received the Ph.D. de-
gree in communication and information engineering
from Huazhong University of Science and Technol-
ogy (HUST), Wuhan, China, in 2003.
He is currently a Full Professor with the School of
Electronic Information and Communications, HUST.
He is an Adjunct Professor with the Faculty of En-
gineering and Information Technology, The Univer-
sity of Technology Sydney, Ultimo, NSW, Australia.
Since November 2005, he has been with HUST. Prior
to that, he was a Researcher with Ajou University,
Suwon, South Korea, and Politecnico di Torino, Turin, Italy, from January 2004
to October 2005. He was a Visiting Researcher with Heriot-Watt University,
Edinburgh, U.K., from June to August 2010. He has published more than
100 papers in refereed journals and conference proceedings and has been
granted about 15 patents in China. His research interests include mobile
communications, traffic modeling in wireless networks, green communications,
and interference modeling in wireless communications.
Dr. Ge received the Best Paper Awards from the 2010 IEEE Global Telecom-
munications Conference. He is leading several projects funded by the National
Natural Science Foundation of China (NSFC), the Chinese Ministry of Science
and Technology (MOST), and industries. He is taking part in several interna-
tional joint projects, such as the EU FP7-PEOPLE-IRSES: project acronym
WiNDOW (Grant 318992) and project acronym CROWN (Grant 610524). He
is a Senior Member of the China Institute of Communications and a member of
the NSFC and MOST Peer Review College. Since 2005, he has been actively
involved in organizing more than ten international conferences. He served
as the General Chair for the 2015 IEEE International Conference on Green
Computing and Communications. He serves as an Associate Editor for the
IEEE A CCESS , the Wireless Communications and Mobile Computing Journal
(Wiley), and the International Journal of Communication Systems (Wiley). He
also served as the Guest Editor for the IEEE C OMMUNICATIONS M AGAZINE
Special Issue on 5G Wireless Communication Systems.
7894 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 65, NO. 10, OCTOBER 2016
Hui Cheng received the Bachelor’s degree in in-
formation engineering from Wuhan University of
Technology, Wuhan, China, in 2013. She is currently
working toward the Master’s degree in communica-
tion and information systems with Huazhong Univer-
sity of Science and Technology, Wuhan.
From June to September 2015, she was an ex-
change student with the University of Linkoping,
Linkoping, Sweden. Her research interests include
mobile backhaul traffic and user mobility models for
small-cell networks.
Guoqiang Mao (S’98–M’02–SM’08) received the
Ph.D. degree in telecommunications engineering
from Edith Cowan University, Perth, WA, Australia,
in 2002.
He is with the School of Computing and Com-
munications, The University of Technology Sydney,
Ultimo, Australia. He is also with Beijing University
of Posts and Telecommunications, Beijing, China,
and Huazhong University of Science and Technol-
ogy, Wuhan, China. Between 2002 and 2014, he
was with the School of Electrical and Information
Engineering, The University of Sydney, Sydney, NSW, Australia. In February
2014, he joined The University of Technology Sydney as a Professor of wireless
networking and the Director of the Center for Real-time Information Networks.
The Center is among the largest university research centers in Australia in the
field of wireless communications and networking. He has published more than
150 papers in international conferences and journals, which have been cited
more than 3500 times. His research interests include intelligent transport sys-
tems, applied graph theory and its applications in telecommunications, wireless
sensor networks, wireless localization techniques, and network performance
analysis.
Dr. Mao has been an Editor of the IEEE T RANSACTIONS ON W IRELESS
C OMMUNICATIONS (since 2014) and the IEEE T RANSACTIONS ON V EHIC -
ULAR T ECHNOLOGY (since 2010). He received the Top Editor Award for
his outstanding contributions to the IEEE T RANSACTIONS ON V EHICULAR
T ECHNOLOGY in 2011 and 2014. He is a Cochair of the IEEE Intelligent
Transport Systems Society Technical Committee on Communication Networks.
He has served as a Chair, Cochair, and Technical Program Committee Member
for a large number of international conferences.
Yang Yang (S’99–M’02–SM’10) received the
B.Eng. and M.Eng. degrees in radio engineering
from Southeast University, Nanjing, China, in 1996
and 1999, respectively, and the Ph.D. degree in infor-
mation engineering from The Chinese University of
Hong Kong, Hong Kong in 2002.
He is currently a Professor with Shanghai Insti-
tute of Microsystem and Information Technology,
Chinese Academy of Sciences (CAS), Shanghai,
China, serving as the Director of the CAS Key Labo-
ratory of Wireless Sensor Network and Communica-
tion and the Director of the Shanghai Research Center for Wireless Communi-
cations (WiCO). He is also an Adjunct Professor with the School of Information
Science and Technology, ShanghaiTech University, Shanghai, China. Prior to
that, he was a Senior Lecturer with the Department of Electronic and Electri-
cal Engineering, University College London, London, U.K.; a Lecturer with
the Department of Electronic and Computer Engineering, Brunel University,
London, U.K.; and an Assistant Professor with the Department of Information
Engineering, The Chinese University of Hong Kong. He has coedited a book
on heterogeneous cellular networks (2013, Cambridge University Press) and
coauthored more than 100 technical papers. His research interests include
wireless ad hoc and sensor networks, software-defined wireless networks,
fifth-generation mobile systems, intelligent transport systems, wireless testbed
development, and practical experiments.
Dr. Yang has been serving in the organization teams of about 50 international
conferences, e.g., as a Cochair of the Ad Hoc and Sensor Networking Sympo-
sium at the 2015 IEEE International Conference on Communications and of
the Communication and Information System Security Symposium at the 2015
IEEE Global Telecommunications Conference.
Song Tu received the B.E. degree from Huazhong
University of Science and Technology, Wuhan,
China, in 2014, where he is currently working toward
the Master’s degree with the School of Electronic
Information and Communications.
His research interests include green communica-
tions and distributed wireless networks.