Non-Orthogonal Multiple Access

(NOMA)
Orthogonal Multiple Access (OMA)
• The concept of orthogonal multi-access (MA) was proposed for better
utilization of the resources (i.e., bandwidth and time) in wireless
communication to increase the spectral efficiency
• When the same time and same frequency band are shared by the
more than one user orthogonally, is called OMA technology, e.g.,
frequency division multiple access (FDMA) and time division multiple
access (TDMA), etc.
• After decades of the evolution in OMA, orthogonal FDMA (OFDMA)
became the most prominent multiplexing technique in fourth
generation (4G) of wireless communication.
From OMA to NOMA
• Dilemma to realize a better trade-off between system throughput and user
fairness
• The limitation of the number of orthogonal time, frequency, and code
resources
• The deleterious effects of the wireless channel may erode the
orthogonality. For example, the channel-induced dispersion may ’smear’
the originally orthogonal time-slots of a TDMA system into each other,
because the transmitted signal is convolved with the dispersive channel’s
impulse response (CIR)
• Low bandwidth efficiency
A promising solution is to break the orthogonality
• NOMA: power domain (PD)-NOMA, code domain (CD)-NOMA
Multiple Access Technologies

IDMA: Interleave-division multiple access

LDS-CDMA: Low-density spreading-CDMA
BDM: Bit division multiplexing
MUSA: Multi-user shared access
SCMA: Sparse code multiple access
PDMA: Pattern division multiple access

Fig. 1 The milestone of multiple access technology [1]

[1] L. Dai at el., A survey of non-orthogonal multiple access for 5G IEEE Commun. Surv. & Tut.
 PD-NOMA: Basics
• Support multiple users within a given resource block
(time/frequency/space/code), using different power levels for
distinguishing/separating them.
• Users with better channel conditions get less power
• Applying successive interference cancellation (SIC) at the receiver
for separating the users
OMA Versus NOMA

Fig. 2: A pictorial comparison of OMA and NOMA

Why NOMA for 5 G?
• If a user has poor channel condition
a) The bandwidth allocated to this user via OMA cannot be used at a
high rate
b) NOMA - improves the bandwidth-efficiency.
• If a user only needs a low data rate, e.g. IoT networks
a) The use of OMA gives the IoT node more capacity than it needs
b) NOMA - heterogeneous QoS and massive connectivity
• The use of NOMA can accommodate the 5G requirements
a) High system throughput
b) Low latency
c) Massive connectivity
Basic Transmitter and Receiver Model For NOMA

x
Superposition Coding (SC)
• It allows the transmitter to transmit multiple users’ information at
the same time.

x2

x1
x2 x1
 Successive Interference Cancellation (SIC)
• The basic idea of SIC is that user signals are successively decoded.
After one user’s signal is decoded, it is subtracted from the combined
signal before the next user’s signal is decoded [1].

[1] S. M. R. Islam et al., Power-Domain Non-Orthogonal Multiple Access (NOMA) in 5G Systems: Potentials and Challenges
Downlink PD-NOMA

|h1|2>|h2|2

Fig. 5. Two user downlink PD-NOMA [2]

[2] S. M. R. Ismal et al., Non-Orthogonal Multiple Access (NOMA): How It Meets 5G and Beyond
Cont…
• Suppose that x1 and x2 are the signals to be transmitted from the
base station (BS) to users 1 and 2, respectively. The BS transmits the
superposition coded signal as
x= 𝑃1 x1+ 𝑃2 x2 (1)
where Pi , i =1,2, is the transmit power for user i and message signal xi,
i =1,2, is of unit power.
• The total transmit power of users 1 and 2 can then be written as 𝑃 =
𝑃1 + 𝑃2
• The received signal at the user i can be expressed as
yi = h i x i + n i (2)
where hi is a channel gain coefficient and ni is a additive noise. For
multi-cell scenario, the inter cell interference is also induced in ni
Cont…
• To separate users’ signals, SIC is used at the receivers. The optimal
decoding order of SIC is the decreasing order of the strength of the
users’ channels, determined by |hi|2
• User 1 (with the maximum channel strength |h1|2), alternatively
called the strong user, can cancel the interference from user 2 (with
the least channel strength |h2|2), referred to as the weak user.
• BS periodically performs the SIC ordering based on the channel
state information (CSI) feedback received from users
• A user with a weaker channel strength (i.e., weak user) is allocated
higher power compared to a user with a stronger channel strength
(strong user) to increase its signal-to-interference-plus-noise ratio
(SINR)
SINR Analysis
• For two user NOMA system shown in Fig. (5), where user-1 is a near
user with |h1|2 > |h2|2 (and hence, P1<P2), only user-1 perform the SIC,
• User 1 first decode the user-2 signal x2, and subtracts it from the
received signal (2), after which it decodes its own signal
• In the case of perfect SIC, an instantaneous SINR for user-1 is given by
SINR1= P1|h1|2/Ω0, (3)
where Ω0 is a variance of noise n1
• User-2 treats user-1 signal as a noise and thus it directly decode its
own signal from (2), the instantaneous SINR for user-2 is given by
SINR2= P2|h2|2/(P1|h1|2 + Ω0), (4)
Achievable Data Rate Analysis
• The achievable channel capacity for Gaussian channel is upper
bounded by
R = log2(1+SINR) bps/Hz (5)
• Form (3) and (5), the data rate of the user-1 can be expressed as
R1 = log2(1+SINR1) bps/Hz (6)
• From (4) and (5), the data rate of the user-1 can be expressed as
R2 = log2(1+SINR2) bps/Hz (7)
• It is observed from (6) and (7) that the BS can optimize the users
data rate by adjusting the power allocation factor β with P1=
β*P and P2= (1-β)*P
 Downlink OMA
• To get a comparative understanding of the data rate performances of
NOMA and OMA, we consider the 2-user FDMA scheme shown in Fig. 6
• Total bandwidth (BW) 1 Hz is divided for the two users –– user 1 uses w
Hz while user 2 uses the remaining 1 − w Hz of the BW and the power
ratio β:(1- β)=P1:P2 remains the same as for the NOMA scheme

Fig. 6. Two-user downlink FDMA

OMA Users Data Rate Analysis
• The achievable data rate for user-1 can be written as
R1OMA = w log2 (1 + P1|h1|2/w Ω0) (8)

• The Rate of user-2 is given by

R2OMA = (1-w) log2 (1 + P2|h2|2/(1-w) Ω0) (9)

• The achievable sum capacity is ROMA = R1OMA + R2OMA . Eqs. (8) and
(9) suggest that no OMA user suffers from the interference from
the signal of the other user, unlike NOMA
Data Rate Comparison
Optimal Power Allocation
• A widely used fairness metric is the maximin fairness (MMF), which is
achieved by maximizing the worst (i.e.,minimum) user rate
• According to (6) and (7), the power allocation to achieve the MMF is
given by the solution to the following optimization problem:
βopt : max
P P
min{R1(P1, P2), R2(P1,P2)} (10)
1, 2

s.t., 0 ≤ P1 ≤ P2, P1 + P2 ≤ P, β= P1/ P

After solving the Eq. (10) according to [2], we have
βopt = (-(ϒ1+ϒ2)+ (ϒ1+ϒ2)2 + 4 ϒ1ϒ22𝑃)/(2 Pϒ1ϒ2),
where ϒ1= |h1|2/Ω0, ϒ1= |h2|2/Ω0

[2] J. Zhu, et al, On optimal power allocation for downlink non-orthogonal multiple access
systems. IEEE J. Sel. Areas Commun. (2017)
Uplink PD-NOMA
• Uplink NOMA system model is shown in Fig. 7, where user 1 and
2 simultaneously transmit their signals x1 and x2 to the BS

Fig. 7. Two user power domain uplink NOMA

Cont..
• The received signal at the BS can be written as
y = 𝑃1 x1h1+ 𝑃2 x2h2+n,
where n is a additive Gaussian noise
• In general, the BS broadcasts a downlink reference signal based on which
each user performs channel estimation. Thus, they can regulate their
transmit power to either 𝑃1 or 𝑃2, depending upon their channel gains.
• User 1 is again marked as the strong user experiencing a higher channel
gain compared to user 2, the weak user
• The BS first decodes the signal of user 1 by treating the signal of user 2 as
noise, then subtracts the decoded signal of user 1 (𝑥̂1) from the received
signal (𝑦). From the remaining, the signal of user 2 (𝑥̂2) is then decoded
Cont….
• The achievable data rate of the user 2, 𝑅2 for a transmission BW of
1 Hz
R1 = log2(1+ P1|h1|2/(P2|h2|2 + Ω0)) bps/Hz

• If SIC is perfect, the achievable data rate of the user 2, 𝑅2 for a

transmission BW of 1 Hz is expressed as
R1 = log2(1+ P2|h2|2/Ω0) bps/Hz

• The achievable sum capacity is 𝑅= 𝑅1+ 𝑅2

Sum Rate Comparison

Fig. 8. Sum capacity in 2-user downlink NOMA and OMA

Interplay Between NOMA & Cognitive Radio
• NOMA in cognitive radio network [3].

[3] Y. Liu et al., “Non-orthogonal Multiple Access in Large-Scale Underlay Cognitive Radio Networks”
Cont….
• NOMA can be viewed as a special case of cognitive radio
- The user with poorer channel condition can be viewed as a primary user
- With orthogonal MA, bandwidth allocated to this user cannot be reused, which
. leads to poor spectral efficiency
- By using NOMA, a user with better CSI is admitted
- Slight loss at the primary user, but significant improvement on system throughput
- More secondary users can be admitted by using NOMA
Underlay Cognitive-NOMA

Fig. 9. Underlay cognitive-NOMA architectures [4]

[4] L. Lv et. al., “ Cognitive Non-Orthogonal Multiple Access with Cooperative Relaying: A New Wireless
Frontier for 5G Spectrum Sharing” IEEE Communication Magazine
Outage Analysis
• Secondary transmitter (ST) communicates with all SUs by applying the
NOMA transmission protocol
• The power allocated to the ST is constrained to limit the interference
at the PR’s
• According to underlay CR, the transmit power Pt at the ST is
constrained as follows:

(11)

where Ip is the maximum permissible interference power an PRs and Ps

is the maximum transmission power at the ST, gl is the channel
coefficient between ST to lth PR.
Cont…
• According to NOMA, the ST sends a combination of messages to all NOMA users,
and the observation at the mth SU is given by
(12)

where nm is the additive white Gaussian noise with variance σ2 , an is the power
allocation coefficient for the nth SU with σ𝑀
𝑛=1 𝑎𝑛 = 1, xn is the information for the
nth user, and hm is the channel coefficient between the mth user and the ST
• Without loss of generality, all the channels of SUs are assumed to follow the
order as |h1|2 ≤ |h2|2 ≤···≤|hM|2. The power allocation coefficients are assumed
to follow the order as a1 ≥ a2 ≥ ···≥ aM.
• According to the NOMA principle, SIC is carried out at the receivers and assumed
that 1 ≤ j ≤ m <i. In this case, the mth user can decode the message of the jth user
and treats the message for the ith user as interference
Cont…
• mth user first decodes the messages of all the (m − 1) users and then successively
subtracts these messages to obtain its own information.
• The received signal to-interference-plus-noise ratio (SINR) for the mth user to
decode the information of the jth user is given by
(13)

where IB denotes the interference between PT and SU

Cont…
• In NOMA, an outage occurs if the mth user cannot detect any of the jth
user’s messages, where j ≤ m due to SIC.
• Denote Xm = |hm|2 γt/(ρbIB + 1), based on (13), the CDF of Xm is given
by

• We denote εj = τj/(aj − τj σ𝑀 𝑎
𝑖=𝑗+1 𝑖 ) for j<M, τj = 2Rj − 1, R is the target
j
data rate for the jth user, εM = τM/aM, and = max{ε1, ε2,...,εm}.
Cont…
• The outage probability at the mth user can be expressed as follows:

(14)

• The average of (14) is derived in [5]

[5] Y. Liu et al., Non-orthogonal Multiple Access in Large-Scale Underlay Cognitive Radio Networks .
, IEEE Trans. on Veh. Tech.
Result
a1=0.5
a2=0.4
a3=0.1

RD=stands for the radius of disc on SUs exist

Target rate for each user 0.1 bps