OUTAGE PERFORMANCE OF COMP NOMA NETWORKS WITH SELECTIVE CELL AND TRANSMIT DIVERSITY

Received: April 1, 2020. Revised: April 27, 2020. 196

 Outage Performance of CoMP NOMA Networks with Selective Cell and
 Transmit Diversity

 Nam-Soo Kim1*

 1
 Deptartment of Electronic Engineering, Cheongju University, Republic of Korea
 * Corresponding author’s Email: nskim@cju.ac.kr

 Abstract: Most of the coordinated multipoint (CoMP) transmission studies have been focused on the sum capacity.
 However, the increase of the sum capacity does not guarantee the individual user’s required outage performance. In
 this paper, we consider the outage performance of non-orthogonal multiple access (NOMA) networks with CoMP
 transmission under multi-cell environments. We adapt the Alamouti code for CoMP transmission to improve the
 outage performance of a cell-edge user. Also to provide the spatial diversity gain, cell selection with maximum channel
 gain is adapted. Outage probabilities of users are derived in closed-form and verified through Monte Carlo simulation.
 It is noticed that the outage performance improves as a near user approaches a base station, however, the error floor
 which does not improve the performance as the transmit power increases is observed. Since the inter cell interference
 (ICI) from the selected cell with NOMA causes the error floor, the selected cell with orthogonal multiple access (OMA)
 shows better outage performance for the near user than the cell with NOMA. Furthermore, the performance is more
 susceptible to the distance between the base station and the near user rather than power allocation to the user. Also,
 the performance of the cell-edge user shows significant improvement with the combination of the Alamouti code and
 the cell selection compared to the cell selection only.
 Keywords: CoMP system, NOMA system, Outage performance, Selective diversity, Transmit diversity.

 systems concentrate on the downlink analysis, uplink
1. Introduction analysis has recently been commenced [10].
 In [5], Choi et al. introduced the Alamouti code
 Recently, coordinated multipoint (CoMP)
 [11] into a CoMP NOMA system for improvement on
transmission which sends messages from more than
 sum rate with two base stations. And an opportunistic
two base stations has been focused to improve the
 CoMP which selects the best base station among
performance of the cellular system [1-4]. The cell-
 multi-cells has been proposed to NOMA systems [6],
edge user generally receives poor signal strength
 [7, 8]. The recent study expands the CoMP
caused by the severe fading and path loss attenuation.
 transmissions to all users not only to a cell-edge user,
This phenomenon can be improved by transmitting
 and controls the power allocation to each user to
the same messages from different base stations.
 increase the sum capacity [12]. In [9], an adaptive
However, as the number of cell-edge users increases
 relay-aided multiple access (RAMA) has been
the more communication resources such as frequency,
 proposed to increase the sum capacity as well as to
time slot, codes are required for the CoMP
 lessen the transmit power of the cell-edge user for
transmissions. Consequently, inefficient spectral
 uplink communication.
usages of the conventional orthogonal multiple
 Most of the studies, including upper mentioned
access (OMA) systems are expected. To overcome
 researches for CoMP transmission have been focused
the inefficient spectral usages without degradation of
 on the rise of the sum capacity [5, 6, 9, 12]. On the
system performance, non-orthogonal multiple access
 network side the sum capacity increase is important,
(NOMA) has been introduced to CoMP systems [5-
 however, on the user side the outage performance can
9]. While most of the studies of CoMP NOMA
International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 197

be a matter of primary concern. For example, a users are given, and the analytical results are verified
certain network structure can increase sum capacity, with the simulation in section 4. Finally, this paper
but can be offered the poor performance for a concludes with the future research direction in
designated user. In a NOMA system, the inter user section 5.
interference can be eliminated using successive Notation: ℎ denotes the channel coefficient
interference cancellation (SIC), however, the inter between BS ( ∈ {0, 1}) and User j ( j  1, 2,3 ),
cell interference (ICI) can’t be removed. Though the
 which has complex Gaussian distribution with zero
outage performance of a cell-edge user can be
 mean and variance , hij ~ CN (0,  ij ) . The variance
improved by CoMP transmission, the ICI from the
 − 
base stations which are participated in CoMP is given by = , where is the distance
transmission can cause severe degradations to a near between BS and User j , and is the propagation
user. In some cases, even if the transmit power of the loss coefficient which has typically between 3 to 6 in
base stations increases, the near user can’t be reached urban area [13]. E[·] and max{∙} denote the statistical
to the required outage performance due to the ICI. For expectation and the maximum value, respectively . 
this reason, it is important to analyse the outage is Gaussian noise, ~ (0, 0 ), ∈ {1, 2, 3}.
performance of individual users. For convenience, we present the list of symbols
 Motivated by the observation of the above used in this paper in Table 1.
mentioned reasons, we modified Choi’s system
model which is applied to improve the sum capacity Table 1. Table of symbols
[5]. Our proposed system model includes a multi-cell the number of base stations
environment and adapts a selected cell transmission
for CoMP. We derive the outage probabilities of the ( ) the messages to user ( ∈ {1,2,3}) from BS 
near and cell-edge users analytically. The main ( ∈ {0, 1}) at timeslot t (t∈ {1,2})
contributions can be summarized as follows: ∗ ( ) the complex conjugate of the message ( )
- Assumes a multi-cell environment with NOMA
 protocol, and proposes a selected CoMP NOMA average power of ( ) ( ∈ {0, 1},
 ∈ {1,2,3})
 system model. A selected base station which has
 maximum channel gain among the neighbour received SINR of a at b
 base stations transmits the Alamouti code for _ outage probability of a
 CoMP.
- The outage probabilities of the near and cell-edge 2. System model
 users are derived in closed-form, and verified In this study, we consider downlink CoMP
 through Monte Carlo simulation. networks with a home base station BS0 and
- The effect of the interference from the selected neighbour base stations (BS1, … , BS ) in Fig.1.
 cell is analysed, and finds there exists an error We assume the NOMA transmit protocol is applied
 floor which does not decrease the outage to every BSs. Generally, the signal strength for a cell-
 edge user is poor due to the fading and path
 probability even if the transmit power increases.
 attenuation. Moreover, the interferences from
- Analyse the space diversity gain caused by the
 cell selection, and performance enhancement
 from the Alamouti code.
- Performance comparisons are made to the
 NOMA and OMA cell.
- The effect of the power allocation among users in
 the NOMA cell is analysed.

 The rest of the paper is organized as follows.
Section 2 presents the proposed selective CoMP
NOMA system model and analyses the signal-to-
interference plus noise ratio (SINR) of users. The
outage probabilities of users have derived
analytically in section 3. The numerical examples of Figure. 1 Multi-cell system model
the outage probabilities of the near and cell-edge
International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 198

 0 (1) = 01 (1) + 03 (1) 1 (1) = 12 (1) + 13 (1) Similarly, the received signal of U2 is given by
 ∗
 ∗
 0 (2) = 01 (2) − 13 (1) 1 (2) = 12 (2) + 03 (1)
 2 (1) = ℎ02 0 (1) + ℎ12 1 (1) + 2
 ℎ02 ℎ11 = ℎ02 { 01 (1) + 03 (1)} + ℎ12 { 12 (1) + 13 (1)} + 2
 ℎ03
 1 (2) = ℎ02 0 (2) + ℎ12 1 (2) + 2
 ℎ13 ∗ ∗
 ℎ01 ℎ12 = ℎ02 { 01 (2) − 13 (1)} + ℎ12 { 12 (2) + 03 (1)} + 2
 (2)
 BS0 U1 U3 U2 BS1
 The received signal of U3 is
 Figure. 2 CoMP system model
 3 (1) = ℎ03 0 (1) + ℎ13 1 (1) + 3
adjacent cells degrade the performance of the cell- = ℎ03 { 01 (1) + 03 (1)} + ℎ13 { 12 (1) + 13 (1)} + 3
edge user and causes poor spectral efficiency. To 3 (2) = ℎ03 0 (2) + ℎ13 1 (2) + 3
alleviate these degradations, we assume CoMP ∗
 = ℎ03 { 01 (2) − 13 ∗
 (1)} + ℎ13 { 12 (2) + 03 (1)} + 3 .
transmission for a cell-edge user. (3)
 The more base stations participated in CoMP
transmissions, the better performances are expected. Also, the received signal of U1 after SIC can be given
However, the more communication resources such as as
frequency, time, and codes are required.
Consequently, it leads to the inefficient spectral 1 (1) = ℎ01 01 (1) + ℎ11 12 (1) + 1
efficiency. To enhance the spectral efficiency and the 1 (2) = ℎ01 01 (2) + ℎ11 12 (2) + 1 (4)
user performance, we adapt the selection of the strong
BS among neighbour cells and introduce the transmit Similarly, the received signal of U2 after SIC can be
diversity of the Alamouti code. written by
 Fig. 2 shows the simplified model of Fig.1 for
easier analysis. We assume two BSs, Bs0 and BS1, 2 (1) = ℎ02 01 (1) + ℎ12 12 (1) + 2
where BS1 denotes the selected base station among 2 (2) = ℎ02 01 (2) + ℎ12 12 (2) + 2 . (5)
 BSs without loss of generality. Each BS has two
Users, U1 and U3 for BS0 and U2 and U3 for BS1. Since the Alamouti coded signals are transmitted
U1 and U2 represent near users of BS0 and BS1, from BS0 and BS1, the received signals are combined.
respectively. And U3 is the common cell-edge user The combined signals can be given as [11]
of BS0 and BS1. Notice the channel coefficient
 (1) ℎ∗ ℎ1 (1)
between BS1 and U3 is denoted by ℎ13 , which is the [ ] = [ 0 ][ ] (6)
 (2) ∗
 ℎ1 −ℎ0 ∗ (2)
strongest channel from the selected BS, for easy
distinction between channel coefficients. We assume
 where ∈ {1, 2, 3} . While the case of = 3 , the
all channels are independent Rayleigh flat fading
 channel coefficients between the selected BS1 and
channels. As mentioned in the introduction section, ∗
the Alamouti code is applied to increase the SINR of U3 will be denotes ℎ13 and ℎ13 instead of ℎ13 and
 ∗
the cell-edge user U3 [11]. ℎ13 , respectively, in distinction to the other channel
 In timeslot 1, BS0 and BS1 transmit 0 (1) = coefficients.
 01 (1) + 03 (1) and 1 (1) = 12 (1) + 13 (1) , The received SINR of 03 (1) and 13 (1) at U1
respectively. In timeslot 2, BS0 and BS1 transmit can be obtained from Eq. (6) by replacing = 1,
 ∗
 0 (2) = 01 (2) − 13 (1) and 1 (2) = 12 (2) + (1) (1)
 ∗ 103 = 113 =
 03 (1), respectively. The average power of messages 2
 (|ℎ01 |2 +|ℎ11 |2 ) 3
is E[| 01 (1)|2 ] = E[| 01 (2)|2 ] = 1 , ∗ 2 ∗ 2 .
 2] 2] |ℎ01 | 1 +|ℎ01 ℎ11 | 1 +|ℎ01 ℎ11 | 2 +|ℎ11 |4 2 +(|ℎ01 |2 +|ℎ11 |2 ) 0
 4
E[| 12 (1)| = E[| 12 (2)| = 2 , and
 (7)
E[| 03 (1)|2 ] = E[| 13 (1)|2 ] = 3 . Since more
power is allocated to the cell-edge user in the NOMA
 Assume 1 = 2 = , then Eq. (7) becomes
system, 3 is bigger than 1 or 2 .
The received signal of U1 in Fig 2 can be written by (1) (1) (|ℎ01 |2 +|ℎ11 |2 ) 3
 103 = 113 = (|ℎ 2 2 . (8)
 01 | +|ℎ11 | ) + 0
 1 (1) = ℎ01 0 (1) + ℎ11 1 (1) + 1
 = ℎ01 { 01 (1) + 03 (1)} + ℎ11 { 12 (1) + 13 (1)} + 1 Similarly, SINR of 03 (1) and 13 (1) at U2 can be
 1 (2) = ℎ01 0 (2) + ℎ11 1 (2) + 1
 ∗
 = ℎ01 { 01 (2) − 13 ∗
 (1)} + ℎ11 { 12 (2) + 03 (1)} + 1 .
 obtained from Eq. (6) by replacing = 2,
 (1)

International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 199

 (1) (1) (|ℎ02 |2 +|ℎ12 |2 ) 3
 203 = 213 = (|ℎ 2 2 . (9)
 02 | +|ℎ12 | ) + 0 (1)
 ( 101 < Γ1 ) = (|ℎ01 |2 < ) (16)
After SIC, the SINR of 01 (1) and 01 (2) at U1 can
be obtained from Eq. (4), where = Γ1 1 / 1 and 1 = [ 11 ] 2 + 0 . For
 the notational simplicity, we denote |ℎ11 |2 = 11 .
 (1) |ℎ01 |2 1 (2) The outage probability of Eq. (14) can be rearranged
 101 =
 [|ℎ11 |2 ] 2 + 0
 = 101 (10)
 using Eqs. (15) and (16), and can be written by
From Eq. (5), the SINR of 12 (1) and 12 (2) at U2 _ 1 = (|ℎ01 |2 < , |ℎ11 |2 < )
after SIC is given by + {(|ℎ01 |2 ≥ , |ℎ11 |2 < ), |ℎ01 |2 < } . (17)
 (1) |ℎ12 |2 2 (2)
 212 =
 [|ℎ02 |2 ] 1 + 0
 = 212 . (11) Denote |ℎ01 |2 = 01 . Under the Rayleigh fading
 environment, Eq. (17) can be expressed as [14], [16]
Similarly, SINR of 03 (1) and 13 (1) at U3 can be
 
obtained from Eq. (6), − (
 [ 01 ]
 ,
 [ 01 ]
 )
 0_ 1 = 1 − . (18)
 (1) (1)
 303 = 313 =
 2 2
 Alternatively, Eq. (18) can be expressed as
 (|ℎ03 |2 +|ℎ13 | ) 3
 2 4
 |ℎ03 |4 1 +|ℎ03 |2 |ℎ13 | ( 1 + 2 )+|ℎ13 | 2 +(|ℎ03 |2 +|ℎ13 | ) 0
 2 . (12) 
 − (1)
 1− [ 01] = ( 103 < Γ3 ) , η ≥ ξ
 0_ 1 = { . (19)
Assume 1 = 2 = , SINR of 03 (1) and 13 (1), 1− 
 −
 [ 01] = ( 101
 (1)
 < Γ1 ) , η < ξ
 3 is given by
 2 From Eq. (18) and Eq. (19), we can conclude that the
 (|ℎ03 |2 +|ℎ13 | ) 3
 3 = 2 . (13) outage probability of U1 becomes
 (|ℎ03 |2 +|ℎ13 | ) + 0

 (1) (1)
 0_ 1 = { ( 101 < Γ1 ) , ( 103 < Γ3 ) }
3. Outage analysis
 (20)
 In this section, we consider the outage probability
of the proposed system in Fig.2. When the received (1)
 where ( 101 < Γ1 ) equals 1 − − / [ 01 ] from
SINR bellows the predetermined threshold, the
 (1)
outage event occurs. The outage event of U1 happens Eq. (9). And ( 103 < Γ3 ) has two random
two cases; firstly, the SINR of 03 (1) at U1 is less variables as shown in Eq. (15), it can be written by
than the threshold, and secondly, though the SINR of
 03 (1) at U1 is greater than the threshold, the SINR (1)
 ( 103 < Γ3 ) =
 1 
 ∫ {− −( − )/ [ 01] } − / [ 11] ,
 [ 11 ] 0
of 01 (1) is less than the threshold. Hence, the 3
outage probability of U1 can be written by [14-15] Γ3 <
 
 (21)

 0_ 1 = ( 103
 (1)
 < Γ3 ) + ( 103
 (1) (1)
 ≥ Γ3 , 101 < Γ1 ) After rearrangement, we have Eq. (22) at the middle
 (14) of the next page.
 Similar to the outage probability of U1, the
 outage probability of U2 can be written by
where Γ1 and Γ3 are the thresholds of U1 and U3,
respectively, Γ1 = 2 1 − 1 and Γ3 = 2 3 − 1 . 1 (1) (1) (1)
 0_ 2 = ( 213 < Γ3 ) + ( 213 ≥ Γ3 , 212 < Γ2 )
and 3 are spectral efficiency [bps/Hz] of U1 and U3,
respectively. By replacing Eq. (8) into Eq. (14), the (23)
first probability of Eq. (14) can be rearranged by
 Also Eq. (23) can be arranged similar to Eq. (20), we
 (1)
 ( 103 < Γ3 ) = (|ℎ01 |2 < − |ℎ11 |2 , |ℎ11 |2 < ) have
 3
 = (|ℎ01 |2 < , |ℎ11 |2 < ) , Γ3 < (15) (1) (1)
 0_ 2 = { ( 212 < Γ2 ) , ( 213 < Γ3 ) }
 (24)
 where = Γ3 0 /( 3 − Γ3 ) and η = − |ℎ11 |2 .
 (1)
The outage probability of 101 in the second (1)
 where ( 212 < Γ2 ) can be obtained from Eq.
probability of Eq. (14) can be obtained from Eq. (10),
 (11),
International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 200

 (1)
 ( 212 < Γ2 ) = 1 − 
 −
 [ 12] (25) In the case of 1 = 2 = , the outage probability
 can be rearranged from Eq. (13)
where ε = Γ2 2 / 2 and 2 = [ 01 ] 1 + 0 . For 2 2
 ( 3 < Γ3 ) = {|ℎ03 |2 < − |ℎ13 | , |ℎ13 | < },
simplicity, define |ℎ02 |2 = 02 , |ℎ12 |2 = 12 . By 3
replacing E[ 01 ] and E[ 11 ] with E[ 02 ] and E[ 12 ], Γ3 <
 
 . (28)
 (1)
respectively, the probability of ( 213 < Γ3 ) in
 2
Eq. (24) is given by Eq. (26). Define |ℎ03 |2 = 03 and |ℎ13 | = 13, then Eq. (28)
 The outage event of U3, which receives both becomes
signal form BS0 and BS1, happens when the received
 − 
SINR is less than the threshold. Hence, the outage ( 3 < Γ3 ) = ∫0 [1 − {− ( )}] 13 ( ) ,
 [ 03 ]
probability can be defined by Γ3 <
 3
 (29)
 
 (1)
 0_ 3 = ( 303 < Γ3 ) . (27) where 13 ( ) is the probability density function

 1 1
 (1) − [ 01 ] − −( − ) 3
 ( 103 < Γ3 ) = 1 − [ 11] − [ 01] {1 − [ 11 ] [ 01] } , Γ3 < . (22)
 [ 01 ]− [ 11 ] 

________________________________________________________________________________________

 1 1
 (1) − [ 02 ] − −( − ) 3
 ( 213 < Γ3 ) = 1 − [ 12] − [ 02] {1 − [ 12] [ 02 ] } , Γ3 < . (26)
 [ 02 ]− [ 12 ] 

________________________________________________________________________________________

 1
 −1 (1 − − [ [ 03 ] − −( ̂ − ) 
 0_ 3 = ∑ 
 =1 ( ) (−1)
 ̂ 13] − [ 03 ] {1 − [ 13 ] [ 03 ]
 }), Γ3 < 3 . (31)
 [ 03 ]− [ 13 ] 

________________________________________________________________________________________

 (pdf) of 13 . Since the selected BS1 has the
maximum channel gain between BS j and U3 ( = 4. Numerical examples
1,2, … , ), the pdf in Rayleigh fading channel is This section considers the numerical examples of
given by the outage performance of U1 and U3. We assumed
 −1 
 the distances between BS0 and U1 and between BS1
 − −
 13 ( ) =
 [ 13 ]
 [1 − ̂
 [ 13] ] ̂
 [ 13 ] and U3 are assumed identical (i.e., 01 = 12). The
 distances are normalized to the distance between BS0
 = ∑ 
 =1 ( ) (−1)
 −1
 − / [ 13] , Γ3 < 3
 [ 13 ] and U3. Also, we assumed the transmit powers to U3
 (30) from BS0 and BS1 are identical. From this
 assumption, the outage probabilities of U1 and U2 are
where the second equality holds from the binomial the same. Hence, we focus on the outage performance
expansion [17]. Replacing Eq. (30) into Eq. (29), the of the U1 and U3.
outage probability of U3 can be obtained in Eq. (31). Fig. 3 shows the outage probability of U1 ( _ 1 ),
 As a special case, if the number of neighbour BSs “*” denotes the results of Monte Carlo simulation,
is single (i.e., = 1 ) and [ 03 ] = [ 13 ] , the which is obtained from 1 × 108 iterations. It shows
outage probability of U3 can be obtained from Eqs. that the results from the analysis show a perfect
(29) and (30), match for the simulation. As we expected, the outage
 
 probability degrades as the distance between BS0 and
 −1 (1 − − [ −
 0_ 3 = ∑ 
 =1 ( ) (−1)
 ̂ 13 ] − [ 03 ] ) , U1 increases. Also, it shows the error floor - the
 [ 13 ]
 3 outage does not decrease according to the increase of
 Γ3 < , = 1, [ 03 ] = [ 13 ] . (32)
 the transmit SNR. When 01 = 0.2 and 01 = 0.3,
 the outage probability of 1 × 10−3 can’t be reached.

International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 201

 0 0
 10 10
 01 = 0.3
 0_ 3_ _ 
 01 = 0.2
 -1 -1 =1
Outage prob. of U1, 0_ 1

 Outage prob. of U3, 0_ 3
 10 01 = 0.1 10
 =2
 _ 1 =3
 -2 -2
 10 10

 -3 -3
 10 10 0_ 3
 =1

 -4 -4
 10 10

 _ 1_ _ 12 =2
 -5 -5
 10 10
 =3

 -6 -6
 10 10
 0 5 10 15 20 25 30 0 2 4 6 8 10 12 14 16 18 20

 Tx SNR, 0 ( ) Tx SNR, 0 ( )

 Figure. 3 Outage probability of U1 ( 1 = 3 = 1, Figure. 4 Outage probability of U3 for different 
 = 3, 1 = 2 = , 3 = 4 ) ( 3 = 1, = 3, 01 = 0.1, 1 = 2 = , 3 = 4 )

 0
 10
This is because of the interference, the message of U2,
from BS1. Though the inter user interference from
 -1
BS0 can be cancelled by SIC, the interference from
 Outage prob. of Users

 10

BS1 which is the inter cell interference (ICI) can’t be _ 1 01 = 0.3
 -2
cancelled. 10
 01 = 0.2
 _ 1_ _ 12 in Fig. 3 means the outage
 -3
probability of U1 when BS1 is OMA cell. In that case, 10
 01 = 0.1
BS1 does not transmit the messages for U2; BS1
 ∗
transmits 1 (1) = 13 (1), 1 (2) = 03 (1). There is 10
 -4
 =1
no interference from BS1, hence the error floor does
 _ 3
not exist; the outage probability decreases as the 10
 -5

 =2
transmit SNR increases. From this observation, we =3
can conclude that the OMA cell which transmits a 10
 -6
 0.4 0.5 0.6 0.7 0.8 0.9 1
message for U3 only is preferred for the performance
 Power allocation coefficient, α
improvement of U1.
 The outage probability of U3 for different Figure. 5 Outage probability of U1 and U3 with limited
numbers of neighbour cells is shown in Fig. 4, the transmit power ( 1 = 3 = 1, = 3,
results from the analysis and simulations are well = (1 − ) 1 + 3 , 1 = 2 , / 0 = 20 )
matched. If the outage probability of 1 × 10−3 is
required, the transmit SNR is 8.72 dB, 3.43 dB, and U1. The minimum outage performance is noticed at
0.78 dB is necessary with = 1 , 2, and 3, 0.51 of power allocation coefficient α. While above
respectively. When = 1 which is the system model 0.9 of α, the power allocation to U1 decreases and the
of Choi [5], it shows the worst performance. As the interference from BS1 consists. Hence SINR of U1
number of increases, the required transmit SNR to decreases, consequently, increase the outage
maintain the required outage probability decreases probability.
due to the space diversity gain. Therefore the cell Notice that the outage probability of the U3 is not
selection in this paper is effective to reduce the a function of distance as shown in Eq. (31) and Eq.
outage probability. As a reference, the outage (32), it is why the distance is normalized to itself. We
probability without the Alamouti code ( _ 3_ _ ) observed that the outage probability decreases as the
is shown also in this figure, it revels higher outage number of the neighbour cells and power allocation
performance than with the code. increases.
 The outage probabilities of U1 and U3 with
limited transmit power of a base station is shown in 4. Conclusions
Fig. 5. The outage probability of U1 ( _ 1 ) improves In this paper, we consider selective CoMP
with the decrease of the distance between BS0 and NOMA systems in multi-cell environment. In the
 proposed system, we adapt the Alamouti code for
International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 202

cell-edge user and cell selection with maximum Scenarios an Operational Challenges”, IEEE
channel gain. The outage probability is derived in Communications Magazine, Vol. 50, No. 2,
closed-form and the results are verified with Monte pp.148-155, 2012.
Carlo simulation. For the simulation, 1 × 108 [5] J. Choi, “Non-orthogonal Multiple Access in
iterations are performed.
 From the analysis, we noticed that the error floor Downlink Coordinated Two-point Systems”,
of the near user due to ICI from the selected cell IEEE Communications Letters, Vol. 18, No. 2, pp.
happens. This phenomenon can be removed from the 313-316, 2014.
selected cell with OMA which has no ICI to near user. [6] Y. Tian, S. Lu, A. Nix, and M. Beach, “A Novel
On the while, the outage performance of the cell-edge Opportunistic NOMA in Downlink Coordinated
user significantly improves with the combination of Multi-point Networks”, In: Proc. of Vehicular
the Alamouti code and the selected cell which
 Technology Conference (VTC2015-Fall), Boston,
provides the spatial diversity gain. However, only the
selected cell without the Alamouti code, it is noticed USA, 2015. doi: 10.1109/VTCFall.2015.
that the error floor still exists and the performance 7390804
improvement is not significant. [7] Y. Tian, A. R. Nix, and M. Beach, “On the
 The outage probability of the near user is a Performance of Opportunistic NOMA in
function of the distance from the home base station Downlink CoMP Networks”. IEEE
and power allocation, it is more susceptible to the Communications Letters, Vol. 20, No. 5, pp. 998-
distance than the power allocation coefficient
 1001, 2016.
between 0.51 and 0.9. However, the performance of
the cell-edge user improves with the power allocation. [8] Y. Tian, A. Nix, and M. Beach, “On the
Further research will be focused on the improved Performance of Multi-tier NOMA Strategy in
network structure which enhances both the sum rate Coordinated Multi-point Networks”, IEEE
and outage performance of users in CoMP NOMA Communications Letters, Vol. 21, No. 11, pp.
system. 2448-2451, 2017.
 [9] Q. Zhou, Y. Ma, L. Bai, J. Choi, and Y.-C. Liang,
Conflicts of Interest
 “Relay-aided Multiple Access Scheme in Two-
 The authors declare no conflict of interest. Point Joint Transmission”, IEEE Transactions on
 Vehicular Technology, Vol. 68, No. 6, pp. 5629-
References
 5641, 2019.
[1] M. K. Karakayali, G. J. Foschini, and R. A. [10] Y. Sun, Z. Ding, X. Dai, and O. A. Dobre, “On
 Valenzuela, “Network Coordination for the Performance of Network NOMA in Uplink
 Spectrally Efficient Communications in Cellular CoMP System: a Stochastic Geometry
 Systems”, IEEE Wireless Communications, Vol. Approach”, IEEE Transactions on
 13, No. 4, pp. 56-61, 2006. Communications, Vol. 867, No. 7, pp. 5084-
[2] M. Sawahashi, Y. Kishiyama, A. Morimoto, D. 5098, 2019.
 Nishikawa, and M. Tanno, “Coordinated [11] S. M. Alamouti, “A Simple Transmit Diversity
 Multipoint Transmission/reception Techniques Technique for Wireless Communications”,
 for LTE-advanced,” IEEE Wireless IEEE Journal on Selected Areas in
 Communications, Vol. 17, No. 3, pp.26-34, 2010. Communications, Vol. 16, No. 8, pp. 1451-1458,
[3] S. Brueck, L. Zhao, J. Giese, and M. A. Amin, 1998.
 “Centralized Scheduling for Joint Transmission [12] Z. Liu, G. Kang, L. Lei, N. Zhang, and S. Zhang,
 Coordinated Multi-point in LTE-advanced”, In: “Power Allocation for Energy Efficiency
 Proc. of International ITG Workshop on Smart Maximization in Downlink CoMP Systems with
 Antennas (WSA), Bremen, Germany, pp. 177-184, NOMA”, In: Proc. of IEEE Wireless
 2010. Communications and Networking Conference
[4] D. Lee, H. Seo, B. Clerckx, E. Hardouin, D. (WCNC), San Francisco, USA, pp.1-6, 2017.
 Mazzarese, S. Nagata, and K. Sayana, [13] A. Goldsmith, Wireless Communications,
 “Coordinated Multipoint Transmission and Cambridge University Press, New York, 2005.
 Reception in LTE-advanced: Deployment
International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17

Received: April 1, 2020. Revised: April 27, 2020. 203

[14] Y. Liu, Z. Ding, M. Elkashlan, and H. V. Poor,
 “Cooperative Non-orthogonal Multiple Access
 in 5G Systems with SWIPT”, In: Proc. of the
 23rd European Signal Processing Conf.
 (EUSIPCO), Nice, France, pp. 1999-2003, 2015.
[15] N.-S. Kim, “Overlay Cognitive Radio NOMA
 Networks with Selected Relay and Direct link”,
 International Journal of Intelligent Engineering
 and Systems, Vol. 13, No. 1, pp. 181-190, 2020.
[16] N.-S. Kim, “Cooperative Overlay Cognitive
 Radio NOMA Network with Channel Errors and
 Imperfect SIC”, International Journal of
 Intelligent Engineering and Systems, Vol. 12,
 No. 5, pp. 224-231, 2019.
[17] K. Tourki, H.-C. Yang, and M. –S. Alouini,
 “Accurate Outage Analysis of Incremental
 Decode-and-forward Opportunistic Relaying”,
 IEEE Transactions on Wireless
 Communications, Vol. 10, No. 4, pp. 1021-1025,
 2011.

International Journal of Intelligent Engineering and Systems, Vol.13, No.4, 2020 DOI: 10.22266/ijies2020.0831.17