52 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.16, NO.

2 August 2018

On the Performance of the

Layered-Division-Multiplexing using
Maximal-Ratio Combining

Tanapong Khumyat , Non-member

ABSTRACT tems [1], including mobile HDTV and Ultra HDTV

This article proposes the diversity gain enhance- (UHDTV). The key requirement of this standard is

ment in layered division multiplexing (LDM) systems to service the high performance transmission inclu-

by applying maximal-ratio combining (MRC) techni- ding high data rate and large area coverage in low

que. LDM systems is adopted as a baseline techno- Signal to Noise Ratio (SNR). To enhance this stan-

logy of the ATSC 3.0 systems which the two transmis- dard, layered division multiplexing (LDM), the Non-

sion layers are simultaneously transmitted over two Orthogonal Multiplexing (NOM) technique [2], which

dierent types of fading channels for xed and mobile has been originally proposed as Cloud transmission

services. Hence, the performance analysis of each la- [3] that support multiple multimedia services in di-

yer need to be evaluated on dierent type of fading erent scenarios including mobile, indoor and statio-

channel. Starting with the moment generating func- nary channels. Because of advantage and a simple

tion (MGF) of the MRC output signal-tonoise ratio implementation of LDM [4]-[7], a two-layer LDM has

(SNR), we propose the new method to derive closed- been promoted as an optional feature in the ATSC

form expressions for average symbol-error rate (SER) 3.0 physical layer. LDM technique is dierent from

of the proposed systems over Rayleigh and Nakagami- the frequency division multiplexing (FDM) and time

m fading channels in the presence of error propaga- division multiplexing (TDM), the multi-layered sig-

tion (EP) that is generated from the rst layer detec- nals of LDM can be simultaneously transmitted to

tion. Analytical results show that the specic value the receiver by using a single carrier. The two-layer

of injection level at the transmitter need to be pre- LDM consist of the core layer (CL) stream and the

cisely dened to obtain the best mutual benet for enhanced layer (EL) stream, the higher power CL

both LDM layers, and the proposed technique has signal can be directly detected at receiver while the

signicant performance advantage over conventional lower power EL signal is acted as additional noise,

LDM systems. Simulation results demonstrate the after that the signal interference cancellation (SIC)

tightness of the author's analysis receiver is applied to cancel CL stream from the re-
ceived signal, the EL stream is then detected from

Keywords: Layered division multiplexing, Maximal- the regenerated signal.

ratio combining. The diversity gain and channel capacity of the

ATSC 3.0 system can be increased by Multiple-Input
1. INTRODUCTION Multiple-Output (MIMO) technique [8], [9]. Howe-

In the next generation digital broadcasting systems ver, due to the complexity of MIMO system in prac-

and wireless broadband communication systems, the tical detection, the simple technique called Maximal-

main challenge is to deliver the high data rate mul- Ratio Combining (MRC) can be adopted to improve

timedia services on the limited frequency resources the diversity gain in LDM system [10], [11] instead

due to the rapidly increasing demand of the com- of MIMO technique. In fact, CL stream is norma-

mercial broadband wireless services. The adopted lly encoded by Low-density parity-check (LDPC) co-

technology should be supported for many required des that oer error free in low SNR region. The EL

functions to achieve more ecient spectrum utiliza- stream can be then detected without performance de-

tion and deliver high data rate multimedia services to gradation by error propagation (EP) in SIC process

customers. To facilitate this challenge, the Advance [6]. However in very low SNR scenario, EP in the EL

Television Systems Committee (ATSC) call for pro- stream can be occurred at the receiver and the system

posal for the a new international broadcasting stan- must take into account the eect of EP in error rate

dard, called ATSC 3.0, to support high data rate mul- analysis. SER of M-PSK and M-QAM signals with

timedia services in the next generation wireless sys- diversity technique over generalized fading channels
has been derived by starting with evaluation of the

Manuscript received on August 20, 2018 ; revised on August moment generating function (MGF) [13], [14]. To the
30, 2018. best of our knowledge, there are no works in litera-
The author is with Rajamangala University of Tech-
ture review that demonstrate the analysis method to
nology Lanna Tak, 41 Phaholyothin Rd., Tak, Email:
k_tnp@hotmail.com. acquire a closed-form symbol error rate (SER) ex-
On the Performance of the Layered-Division-Multiplexing using Maximal-Ratio Combining 53

pression for the conventional LDM and MRC-LDM The LDM signal in (2) is a combined signal inclu-
system in the presence of EP [12]. (d) (d)
ding CL stream (S1 ) and EL stream (S2 ) that are
In this paper, the MRC technique is adopted to simultaneously transmitted in dierent layers have di-
increase diversity gain of LDM systems. The advan- erent specications for the delivery of dierent ser-
tage of the proposed scheme can be demonstrated by vices as shown in Fig. 2. The injection level, w de-
analyzing the closed-form error rate performance ver- termines the power allocation among the two layers.
sus the general LDM systems. The MGF of the MRC Meanwhile the higher power of CL stream can be al-
output SNR is rst evaluated and SER can be then located by adjusting the suitable real value w in [0,
derived for CL stream. Due to the eect of EP, the 1), where w=0 results in a single-layer system. The
total probability theorem [9], [15] need to be adopted symbol energy per transmit antenna is assumed equal
to calculate the average SER for EL stream. In order to Es , which the energy Es is divided into two por-
to derive SER performance and demonstrate the ob- tions for CL and EL symbol while the energy ratio
vious advantage of MRC-LDM systems, we consider for both layers can be adjusted by w . The energy Ep
the uncoded MRC-LDM systems without the bene- and w2 Ep are allocated for CL symbol and EL sym-
t of LDPC code. The error rate of BPSK signals Es
bol, respectively, where Ep =
(1+w2 ) . From Fig.1,
over Rayleigh fading channel and QPSK signals over the channel between the transmit antenna and the re-
Nakagami-m fading channel are analyzed for CL and −jϕi
ceive antenna i is denoted by hi where hi = αi e .
EL streams, respectively. Numerical results demon- The receive signal after passing through the fading
strate the tightness of the author's analysis. channel is disturbed by complex AWGN ni with a
The remainder of this paper is organized as follows. one-sided power spectral density, which is represen-
The system and channel models are introduced in the ted by 2N0 (W/Hz). The received baseband signal
next section while in Section 3, we derive the ave- on each receive antenna branch is represented by ri
rage bit-error rate (BER) of binary signals over Ra- and can be expressed as
yleigh and Nakagami-m fading channels by applying
the MGF-based approach. The average SER of M- ri = hi S (d) + ni
PSK signals over Nakagami-m fading channels is des- ( ) (3)
(d) (d)
cribed in Section 4. In Section 5, the average BER of
= hi S1 + wS2 + ni
MRC-LDM system is derived, and numerical exam-
ples is also presented. A conclusion of results are In our analysis, the signal combination at MRC

given in Section 6. receiver will be considered on the MGF of the MRC

2. SYSTEM AND CHANNEL MODELS
This section presents LDM signal at transmitter
and receiver and show the method to derive the SNR
output SNR in the next part. The rst layer CL
(d) (d)
(S1 ) can be detected by considering EL signal (S2 )
as additional noise, thus the detected symbol of CL
stream can be acquired from

= h−1
per symbol of each receive antenna. Finally, the chan- (d)
S̃1 i ri
nel models for the two layers of LDM system are de- (4)
(d)
monstrated. = S1 + ñi

ñi = wS2 +h−1

(d)
2. 1 LDM Signals and Channel Models where i ni . In Appendix 1 we show
that the interference-plus-noise energy of ñi can be
At LDM transmitter, the two layers including CL expressed as
and EL stream are continuously broadcasted over L
independent slowly-varying at fading channels and E[ñi ñ∗i ] = w2 Ep + N0
transmitted to the MRC-LDM receiver as shown in Es (5)
Fig.1. The LDM complex signal is transmitted over = w2 + N0
(1 + w2 )
the channel and can be described as

∞
∑ where E[·] is the expectation operator. Based on
(d) −j2πfc t
s (t) = S e P (t − dTs ) (1)
(4) and (5), the post-processing SNR per symbol of
th
d=−∞ the i receive antenna branch for CL stream can be
expressed as
where fc is the carrier frequency, the function P (·)
Es
denotes a pulse shaping waveform of duration Ts se- µN
{ (d) }∞ γ̄i,CL = 0
(6)
conds, and S
d=−∞
denotes the LDM sequence (1 − µ) N
Es
0
+1
symbols that results from mapping continuous b-bit
b
block into one of M = 2 possible waveforms, and where µ = 1/(1 + w2 ). By considering the process
can be written as after detecting the CL stream ,the EL stream can be
then detected by using successive interference cance-
(d) (d)
S (d) = S1 + wS2 (2) llation (SIC) to cancel CL signal from the existing
54 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.16, NO.2 August 2018

Fig.1: Multilink channel model of MRC-LDM systems.

receive signal ri , which can be explained by

h−1
where ń = w ni .
i
Based on (9) and the
interference-plusnoise energy of ń in Appendix 2, the
(d)
ri,EL = ri − hi Ŝ1 SNR per symbol of each receive antenna branch wi-
( ) (7) thout EP for EL stream can be expressed as
(d) (d) (d)
= hi S1 − Ŝ1 + hi wS2 + ni
w2 Ep Es
where ri,EL denotes the EL received signal and γ̄ELt = = (1 − µ) (10)
N0 N0
(d)
Ŝ1 represents the CL symbol that is chosen from
constellation by the maximum likelihood detector
(d) (d)
(MLD) at the receiver. The results of hi (S1 − Ŝ1 ) (d) (d)
can be divided into two manners, and hence the pos-
2) T he EL received signal with EP (S1 ̸= Ŝ1 ) :
This event can be presented when the MLD chooses
sible results of the EL received signal is also occurred
an incorrect symbol from constellation, hence there
in two cases as follows
(d) (d) is an EP from the SIC process, then the EL received
1) T he EL received signal without EP (S1 = Ŝ1 ) :
signal or the regeneration signal can be written by
This event can be presented when the MLD chooses
a correct symbol from constellation, hence there is no
(d)
EP from the SIC process, then the EL received signal ri,EL = hi wS2 + ∆s hi + ni (11)

or the regeneration signal can be written as

where ∆s represents the mean distance of CL sym-
(d) (d) (d)
ri,EL = hi wS2 + ni (8) bols pair (S1 −Ŝ1 ), which the mean distance can be
obtained by averaging the distance of all symbol pairs
(d) in constellation. For generalized modulation scheme
The EL signal S2 can be detected by multiplying
−1 2
in LDM system, ∆s = ρEp , where the coecient ρ is
(8) with hi /w as follows
depending on the modulation type (ρ = 4 for BPSK
(d)
(d) h−1
i (d) (9)
signal). The EL signal S2 can be detected by mul-
S̃2 = ri,EL = S2 + ń −1
tiplying (11) with hi /w as follows
w
On the Performance of the Layered-Division-Multiplexing using Maximal-Ratio Combining 55

Fig.2: Channels of CL and EL layer of LDM systems.

EL signal is usually designed to deliver the higher qu-

ality service, which usually require higher SNRs for
h−1
ri,EL = S2 + n′′
(d) i (d) (12)
S̃2 = successful detection. The large directional antenna of
w xed receiver should be placed at high location that
′′ is capable of providing higher SNRs as compared to
where n = ∆ws + ń Based on (12) and the
′′ mobile services [16]. At the xed receiver, the un-
interference-plus-noise energy of n in Appendix 3,
the SNR per symbol of each receive antenna branch desired CL signal is rst cancelled and then the EL

with EP for EL stream can be expressed as signal can be detected from the regenerated receive
signal. In the transmission link of EL layer, there are
(1 − µ) N
Es
0
two characters of transmission path including direct
γ̄ELf = Es
(13)
LOS path and scattered path. Hence the Rician fa-
ρµ N0
+1
ding channel is an appropriate model to derive the
performance of EL signal transmission, while the ra-
2. 2 Channel Models
tio of signal power in direct LOS path to signal power
In the MRC-LDM transmission process, the trans- in scattered path can be assigned by the Ricain K-
mitted signal are passed over independent L chan- factor. However the MGF of the received SNR over
nels where l denotes the channel index, and the ran- the Rician fading channel cannot be easily calculated
dom channel amplitude, phase, and delays are repre- th
because of integral of the zero -order modied Bessel
L
sented by {αl }l=1 , {ϕl }l=1 and {τl }l=1 , respectively.
L L
function. Thus, in order to obtain the MGF by un-
Due to delay of multilink channel, we assume that complicated way, this work derives the performance
τ1 < τ2 < τ3 < τL while τ1 is zero for the rst chan- of EL signal transmission over the Nakagami-m fa-
nel that is assigned to the reference channel. We de- ding channel that can be estimated to as the Rician
L
ne {αl }l=1 , {ϕl }l=1 and {τl }l=1 are all constant over
L L
distribution [17], [18], while the relation between the
a symbol duration for slow-fading channel in analy- parameter K and m of the two channels can be pre-
sis. In general, CL signal is designed to deliver robust K 2 +2K+1
sented as m =
2K+1 . The instantaneous SNR
mobile broadcasting service to mobile user and indoor th
per symbol of the l channel γl for Rayleigh and
receiver. Thus there is only non-line-of-sight (NLOS)
Nakagami-m distributions can be referred to as (14)
path for transmitting CL signal without direct line-
and (15), respectively.
of-sight (LOS) path. Therefore the Rayleigh fading
( )
channel is assigned to analyze the performance of CL 1 γ
− γ̄l
signal transmission as shown in Fig. 2. Meanwhile
pγl (γl ; γ̄l , r) = e l , γl ≥ 0 (14)
γ̄l
56 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.16, NO.2 August 2018

l ml −1
( )
mm l γl
γ
− γ̄l
pγl (γl ; γ̄l , ml ) = ml e l , γl ≥ 0 (15) from (19) is giving by
γ̄l Γ(ml )
( )
π
∫2
At the output of MRC combiner, the total condi- 1 g
tional SNR per symbol γt from L branches is given
Pb (E) = Mm − , γ̄ dϕ
∑L π sin2 ϕ
by γt = l=1 γl [13]. 0
[ ( √ )]m
gγ̄
1
3. AVERAGE BER OF BINARY SIGNALS = 1− m
gγ̄ ·
2 1+ m
BER of BPSK signal over fading channel for SISO
)[ ( √ )]k
and MRC receiver can be derived by using MGF of ∑(
m−1
m−1+k 1
gγ̄

Rayleigh and Nakagami-m distribution as follows.

1− 1− m

k=0
k 2 1 + gγ̄
m
(20)
3. 1 Average BER for Single-channel Recep-
tion (L=1) The average BER over Nakagami-m fading chan-
nel in (20) can be derived by substituting MGF of
The average BER can be calculated from the con-
Nakagami-m distribution in function JN (c) and de-
ditional BER and envelope probability density func-
tion (pdf ) as
ne N = m.

∫∞ 3. 2 Average BER for Multichannel Reception

Pb (E) = Pb (E|γ)pγ (γ; γ̄, z)dγ (L > 1)
0 When the MRC receiver is used, the multichannel
(16)
∫
π
2 ∫∞ ( gγ ) conditional BER Pb (E|{γl }L
l=1 ) need to be averaged
1 − sin2 ϕ over the joint pdf of the instantaneous SNR sequence
= e pγ (γ; γ̄, z)dγdϕ ∏L
π pγ1,γ2,...,γL(γ1,γ2,...,γL) = l=1 pγl(γl;γ̄l,il) , hence the
0 0
unconditional BER Pb (E) can be written in general

where z can be substituted by r and m for the Ra- form and MGF form in (21) and (22), respectively

yleigh and Nakagami-m pdf 's, respectively. [13].

1) Rayleigh f ading channel : From (14) and (16), the ∫ ∫ ∫ ∫ ( )

average BER over Rayleigh fading channel can be
∞ ∞ ∞
1
π
2 ∏
L gγ
− sin2lϕ
P b(E) = ... e ·
written as 0 0 0 π 0 (21)
l=1
π
∫ 2 ∫∞ ( ) pγl (γl ; γ̄l , il )dϕdγ1 dγ2 ...dγL
1 1 − singγ γ
Pb (E) = e 2ϕ
· e(− γ̄ ) dγdϕ (17)
∫2 ( ( ))L
π
π γ̄l
0 0 1 g
P b(E) = Mi − , γ̄ dϕ (22)
π sin2 ϕ
By deriving the MGF for Rayleigh distribution in 0
Appendix 4, the average BER from (17) is giving by
where all fading paths in (21) are uniformly distri-
buted with the same fading parameter i and the same
) (
π
∫ 2
1 g average SNR per bit γ̄ for all L channels. Hence, the
Pb (E) = Mr − 2 , γ̄ dϕ simple form of average BER can be demonstrated as
π sin ϕ
0 (18) in (22).
( √ )
1 gγ̄ 1) Rayleigh f ading channel : By considering the
= 1− MGF for Rayleigh distribution [13] and the function
2 1 + gγ̄
JN (c), the average BER of MRC receiver from (22)
where (18) can be derived by JN (c) = can be expressed as
∫ π
( )N
1 sin2 ϕ
π 0
2
sin2 ϕ+c
dϕ [13], [20] for N = 1 and c = gγ̄ ∫2 ( (
π
))L
2) N akagami − m f ading channel : 1 g
From (15) and P b(E) = Mi − 2 , γ̄ dϕ
(16), the average BER over Nakagami-m fading chan- π sin ϕ
0
nel can be written as [ ( √ )]
1 gγ̄
π = 1− ·
∫2 ∫∞ ( ) 2 1 + gγ̄
1 mm γ m−1 (− mγ √
∑ ( L − 1 + k )[ ( )]k
− singγ
Pb (E) = e 2ϕ
· e γ̄ ) dγdϕ L−1
1 gγ̄
π γ −m Γ(m) 1− 1−
0 0 k 2 1 + gγ̄
(19) k=0
(23)

By deriving the MGF of Nakagami-m distribution

in Appendix 5 and function Jn (c), the average BER 2) N akagami − m f ading channel : By considering
On the Performance of the Layered-Division-Multiplexing using Maximal-Ratio Combining 57

( M∫M−1 )π L ( )
1 ∏ gP SK
Ps (E) = Mm − , γ̄ dϕ
π sin2 ϕ
l=1
0
( ∫ )π(
M −1
M )mL
1 sin2 ϕ
= dϕ
π sin2 ϕ + gP m
SK γ̄
0
( M∫M−1 )π( )mL−1 ( )
1 sin2 ϕ sin2 ϕ
Fig.3: BER performance comparison between BPSK = dϕ
SISO receiver, and BPSK MRC receive (L = 2) with π sin2 ϕ + áγ̄ sin2 ϕ + áγ̄
0
m = 1, m = 2 and m = 4. (25)

the MGF for Rayleigh distribution [13] and the func-

where gP SK = sin2 (ϕ/M ), M = 4 for QPSK signal
gP SK
and á = Then (25) can be transformed to the
tion JN (c), the average BER of MRC receiver from m
integral of a trigonometric function In+1 (θ; c) [14] as
(22) can be expressed as
(( ) )
∫2 (
π
( ))L M −1
1 g Ps (E) = ImL π; áγ̄ (26)
P b(E) = Mm − , γ̄ dϕ M
π sin2 ϕ
0
[ ( √ )]mL 5. AVERAGE BER OF MRC-LDM SYS-
gγ̄
1 TEMS
= 1− m
gγ̄ ·
2 1+ m Generally, the LDM detection process can be de-
[ ( √ )]
∑ ( mL − 1 + k )
kmonstrated in Fig.2, the receiver of mobile user detect
mL−1 gγ̄
1 CL stream by considering the low power EL stream
1− 1− m

k=0
k 2 1 + gγ̄
m
as an additional noise [7]. Meanwhile the receiver of
xed user detect the undesired CL stream rst by
(24)
considering the low power EL stream as an additio-
nal noise. Then the undesired CL stream is cancel-
3. 3 Numerical Example
led from received signal of the xed user and the EL
Fig. 3 shows the eect of L and m on the average
stream can be subsequently detected in the presence
error rate of BPSK. The BER performance of SISO
of EP. Thus, to analyze the average BER of the CL
and SIMOMRC over Rayleigh fading and Nakagami-
and EL signals at MRC receiver, we should consider
m fading can be derived by applying (18), (20), (23),
the level of EL signal that is controlled by injection
and (24). These curves conrm the benet of MRC
level w, and assign the number of receive branches L
receiver, a diversity gain and BER performance of
appropriately.
wireless link can be signicantly improved by increa-
sing the number of L at MRC receiver of the LDM
5. 1 Average BER of CL Stream
systems. Moreover, the BER performances are va-
ried according to the characteristic of channels when To derive BER performance of BPSK signal over

varying m parameter. Rayleigh fading channel at the mobile receiver, the

postprocessing SNR per symbol of the CL stream in

4. AVERAGE SER OF M -PSK SIGNALS (6) can be substituted into (23) to derive the average
BER of the CL signal as (see (27))
By considering EL signal transmission in this work,
For case of undesired CL stream at a xed re-
the transmitted QPSK signal is broadcasted over
ceiver, the BERCL′ of BPSK and MPSK signal
Nakagami-m fading channel and accepted by the -
over Nakagami-m fading channel can be derived by
xed MRC receiver. Thus the average SER for this
substituting (6) into (24) and (26), respectively.
scenario can be expressed as [13]

  v    v k
u u
u µNEs
( ) u Es
µN
1  u g (1−µ) Es +1  L−1  u (1−µ) Es +1 
1
0 0

  u  ∑ L − 1 + k   u 
=  1 − u  1 − 1 − u 
N0 N0
BERCL (27)
2  t µ Es
 k  2 t µ Es

1 + g (1−µ)NE0s +1 k=0 1 + g (1−µ)NE0s +1
N0 N0
58 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.16, NO.2 August 2018

Fig.4: BER advantage of MRC-LDM (L = 4) over SISO-LDM (L=1).

sired CL signal that is evaluated over the Nakagami-

m fading channel at the xed receiver while the post-
processing SNR per symbol of the CL stream in (6)
5. 2 Average BER of EL Stream
can be substituted into (24), Ps (E, γ̄ELf ) denotes the
In this work, EL and undesired CL streams are mu- average SER in (26) that substituting by γELf from
tually transmitted over the Nakagami-m fading chan- (13), Ps (E, γ̄ELt ) denotes the average SER in (26)

nel and accepted by the xed receiver. The xed re- that substituting by γELt from (10) while the mo-

ceiver detect the undesired CL stream rst and then dulation method can be chosen by varying
( √ ) M, and
g γ̄i,CL
detect the EL stream after cancelling the undesired
δ= 1
2 1− m
g γ̄
CL stream from the existing receive signal which the 1+ i,CL m

possible output of cancellation can be divided into

two manners. For the rst event, the undesired CL
signal is correctly detected and the postprocessing
SNR per symbol of the EL stream can be derived
as in (10). And the second, the undesired CL signal
is incorrectly detected and the post-processing SNR
per symbol of the EL stream is shown in (13). The
two SNRs per symbol of dierent events are directly
aected to the average BER of EL stream. Therefore
the total probability theorem can be applied [9], [15]
to derive the average BER of EL stream by substitu-
ting (10) and (13) into (26) as follows. (see (28))
where BERCL′ denotes the average BER of unde-
5. 3 Numerical Results
Fig. 4 shows the BER performance compari-
son between SISO-LDM, and MRC-LDM system
with L = 4. The BER performances are deri-
ved by applying (26), (27), and (28) with M =
2 while both SISO-LDM and MRC-LDM transmit-
ter send two layers of BPSK signal simultaneously
with w = 0.3 over the single channel, when the CL
stream of mobile receiver and the EL stream of -
xed receiver are transmitted over the Rayleigh fa-
ding and Nakagami-m fading channel, respectively.

Ps (E, γ̄ELf ) Ps (E, γ̄ELt )

BEREL = BERCL′ + (1 − BERCL′ )
log2 M log2 M
 (1−µ)
Es 
N0

( ) gP SK E
ρµ s +1 
ImL  MM−1 π; 
N0
m
∑ (
mL−1
mL − 1 + k
)
k (28)
= δ mL [1 − δ] ·
k log2 M
k=0
( )
{ } I ( M −1 ) Es
gP SK(1−µ) N
∑ ( ) π; 0
mL−1 mL
mL − 1 + k k
M m
+ 1 − δ mL [1 − δ] ·
k log2 M
k=0
On the Performance of the Layered-Division-Multiplexing using Maximal-Ratio Combining 59

0
10
EL Layer
Average Bit Error Rate

CL Layer
10-5
BPSK L=3 m=1 system I
QPSK L=3 m=5 system I
BPSK L=4 m=1 system II
QPSK L=4 m=5 system II
10-10 QPSK L=3 m=5 system III
QPSK L=3 m=5 system III
QPSK L=4 m=5 system IV
QPSK L=4 m=5 system IV
Marker represents all simulation results
10-15
0 5 10 15 20 25 30
Eb/No [dB]

Fig.5: BER comparison of MRC-LDM receiver in system with L = 3 versus L = 4 over dierent fading
channels.

Table 1: Optimum injection level in various trans- receiver.

mission modes Fig. 5 shows BER performance comparison of

Optimum MRC-LDM receiver in system with L = 3 versus L =
System Layer m, L Signal
point 4 by setting w = 0.24. The error rate performances
CL 1,3 BPSK w = 0.2 at of mobile-CL receiver (m = 1) and xed-CL receiver
A −5
EL 5,3 BPSK BER 2×10 (m = 5) are also demonstrated. These curves indicate

CL 1,3 BPSK w= 0.26 at advantage of MRC technique when increase the nu-
B −5
EL 5,3 QPSK BER 6×10 mber of L. The stable BER performances due to the

CL 5,3 BPSK w= 0.39 at imperfect CL signal cancellation are initially occurred

C −7
EL 5,3 QPSK BER 5×10 at 25 dB of Eb/No.

CL 1,4 QPSK w = 0.23 at Again these gures demonstrate the tightness of

D −5
EL 5,4 QPSK BER 3×10 the author's analysis, i.e. (27) and (28). By con-
sidering BERs of among CL layers, QPSK signals of
system III and IV provide a higher performances than
BPSK signals of system I and II because there are di-
rect part and scattering parts at the xed receiver of

From the gure we can see that the analysis, i.e. system III and IV (m = 4 and 5) while the trans-

(27) and (28), show agreement with the simulation mitted signal of system I and II are transmitted over

results. And the results show that the MRC-LDM mobile channel without a direct path (m = 1), the-

system oers higher BER performance than SISO- refore system III and IV are slightly aected from

LDM system clearly. Moreover, it is observed that fading channels. On the other hand, by considering

the CL layer of SISO-LDM without channel coding BERs of among EL layers, system I and II provide

(FEC) cannot apply in practice because the signal qu- higher performances than system III and IV because

ality cannot support any wireless services, meanwhile the mean distance (∆s ) of symbol pair of BPSK sig-

the MRC-LDM without channel coding can increase nal is less than ∆s of symbol pair of QPSK signal.

diversity gain and can be applied in practical use. However, when focus on the mutual benet between

The stable BER performances of CL and EL receiver CL and EL layer, the results show that the perfor-
−5 −12 mance of CL and EL layer of system IV provide the
are initially occurred at 10 and 10 BER, res-
−9 −10
pectively although the system increase transmitted best mutual performances of 10 and 10 BER

power continuously. These results thus indicate that at 26 dB of Eb/No, respectively because both CL

the LDM and MRC-LDM system without channel co- and EL receiver are xed station and they are equi-

ding oer the constrained BER performance due to pped with the maximum number of receive branches

the presence of EP. These results agree with the au- L, and these agree with the author's simulation re-

thor's simulations shown in Fig. 4. Fortunately, by sults in Fig. 5.

applying channel coding in LDM system, the EP from Fig. 6 (a) shows the eect of injection level on the
CL layer can be perfectly cancelled [3], [6] at LDM MRC-LDM system with various transmission modes
60 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.16, NO.2 August 2018
Fig.6: Search the optimal value of injection level.
by varying injection level and set a xed Eb/No = 20
dB. There are four transmission modes in BER ana-
lysis while the transmission details and the optimum
injection level of various systems can be described in
Table 1. The optimum injection level in Table 1 can
be searched by varying the injection level between 0
to 1 versus BER performance, and then the optimum
point of injection level for CL and EL users can be
chosen at the intersection point between pairs of BER
curves. The results show that the optimum injection
levels for system A, B, C and D are occurred at 0.2,
0.26, 0.39 and 0.23, respectively. At the optimum
point of injection level, the highest mutual benet
between CL and EL layer can be acquired when the
power allocation in LDM transmitter can be properly
operated. For example, in system C, the optimum
−7
point of injection level is 0.39 at 5×10 BER while
other points of injection level oer the lower mutual
benet between CL and EL layer.
Fig. 6 (b) shows the magnitude of BER distance
between CL and EL layer while the BER distance
can be calculated from 10log10 |P bCL − P bEL |, where
P bCL and P bEL denote the BER of CL and EL layer,
respectively. The maximum amplitude of all curves
indicate the position of the optimum injection level
that oer the best mutual benet between CL and
EL layer.
6. CONCLUSION
A performance analysis of the MRC-LDM system
over Rayleigh fading channel and Nakagami-m fading
channel are derived and compared with that of con-
ventional LDM. The error rate analysis can be ea-
analysis results demonstrate that the eect of fading
channel on BER performance in MRC-LDM can be
mitigated by increasing the number of receive bran-
ches L. Hence the proposed system can improve the
BER performance and oer the higher performance
over conventional LDM. Moreover, at the optimum
point of injection level, the highest mutual benet
between CL and EL layer can be acquired while the
injection level controller is a necessary unit in the
MRC-LDM transmitter, and the injection level sho-
uld be automatically adjusted when the transmission
mode and scenario are changed. In practical LDM
systems [6], [7], the Ultra-High-Denition (UHD) vi-
deo transmission can be operated by using high order
QAM modulation with LDPC code. Thus the pro-
posed BER analysis method need to be considered
the coding gain of LDPC code and the distance of
QAM symbol for computing error propagation. Fur-
thermore, the proposed technique can be properly
applied in a limited radio spectrum networks, such
as cognitive radio networks or next generation wire-
less networks.
7. APPENDIX
7. 1 Interference-plus-noise Energy of LDM
System
The proposed proofs in this appendix are derived
to present the interference-plus-noise energy of CL
and EL layers in LDM system.
1) Appendix [ 1 : P roof of (5) : ]
E[ñi ñ∗i ] = E (wS2 + h−1 −1 ∗
(d) (d)
i ni ) × (wS2 + hi ni )
sily derived by calculating the MGF of the MRC- [  ] [ ] [ ]
 (d) 2 (d) n∗ (d)∗
LDM SNR output which the Laplace transform can = w2 E S2  + wE S2 h∗i + wE S2 nhii +
i
be applied to solve the complex integration. These
On the Performance of the Layered-Division-Multiplexing using Maximal-Ratio Combining 61

[ ] ( )−m ( )m
N0 E 1 mm m g sin2 ϕ
|hi |2 = γ̄ m γ̄ + sin2 ϕ
= sin2 ϕ+ gmγ̄

Es
= w2 Ep + N0 = w2 (1+w 2 ) + N0
References
2) Appendix 2 : P roof of interf erence − plus − [1] N. Loghin et al., ATSC 3.0 physical layer speci
noise energy f or (10) : cation: An overview, IEEE Trans. Broadcast,
[( ) ( −1 )∗ ] Vol. 62, No. 1, pp.159-171, 2016.
h−1
E [ńń∗ ] = E
hi
i ×
n [2] E. Garro et al., Layered Division Multiplexing
w ni
i
w
With Multi-Radio-Frequency Channel Technolo-

= N0
= N0 gies, IEEE Trans. Broadcast, Vol. 62, No. 2,
w2 |hi |2 w2
pp.365-374, 2016.

3) Appendix 3 : P roof of interf erence − plus − [3] Y. Wu, B. Rong, K. Salehian and G. Gag-

noise energy f or (13) : non, Cloud transmission: A new spectrum-reuse

friendly digital terrestrial broadcasting transmis-
[( ) ( )∗ ]
E [n′′ n′′∗ ] = E ∆ws + n × ∆ws + n sion system, IEEE Trans. Broadcast, Vol. 58,
No. 3, pp.329-337, 2012.
[( ) ( )∗ ]
h−1 h−1 [4] S. Cho et al., Low-Complexity Decoding Algori-
= E ∆ws + w i
ni × ∆ws + w i
ni thms for the LDM Core Layer at Fixed Receivers
[ ] [ ] [ ] in ATSC 3.0, IEEE Trans. Broadcast, Vol. 63,
2 ∆ n∗
= w12 E |∆s | + w12 E hs∗ i + w12 E ∆hsini + No. 1, pp. 293-303, 2017.
[ ] i
N0 1 [5] W. Li et al., Using LDM to Achieve Seamless
w2 E |hi |2
Local Service Insertion and Local Program Cove-
[ ] [ ]
= wρ2 E |S| + N
2 0
E 1 ρE
= w2p + N 0 rage in SFN Environment, IEEE Trans. Broad-
|hi |2
(
w2
)
w2
cast, Vol. 63, No. 1, pp. 250-258, 2017.
ρEs [6] S. -I. Park et al., Low Complexity Layered Di-
= w12 (1+w 2 ) + N0
vision Multiplexing for ATSC 3.0, IEEE Trans.
Broadcast, 2016, 62, (1), pp. 233-243.
7. 2 MGF of Rayleigh and Nakagami-m Distri-
[7] L. Zhang et al., Layered-Division-Multiplexing:
bution
Theory and Practice, IEEE Trans. Broadcast,
This appendix derives MGF of Rayleigh and Vol. 62, No. 1, pp. 216-230, 2016.
Nakagami-m distribution as follows. [8] L. Zhang et al., Capacity Analysis of LDM-
Based DTV System with Flexible MIMO Con-
1) Appendix 4 : P roof of M GF in (18) and (23) : guration,Proc. IEEE Int. Conf. Broadband
( ) ∫∞ Multimedia Syst. Broadcasting, pp. 1-7, 2016.
γ − gγ
Mr − sing2 ϕ , γ̄ = γ̄1 e− γ̄ · e sin2 ϕ dγ [9] X. Zhu, Z. Wang and J. Cao, NOMA-Based
0
( )
Spatial Modulation, IEEE Access, Vol. 5, pp.
∫∞ 1 −γ gγ
γ̄ − sin2 ϕ 1
∫∞ −γ 1
+ sing2 ϕ 3790-3800, 2017.
= γ̄ e dγ = e dγ γ̄
γ̄
[10] L. Zhang et al., Improving LTE eMBMS with
0 0
∫ ∞By−stapplying Laplace transform, we know that Extended OFDM Parameters and, IEEE Trans.
0
e dt = 1
s . Thus the MGF of Rayleigh distri- Broadcast, Vol. 63, No. 1, pp. 33-47, 2017.
bution can be written as [11] R. S. Rabaca et al., Implementation of
LDM/ISDB-T broadcast system using diversity
( ) ( )−1
Mr − sing2 ϕ , γ̄ = 1 1
+ g
= sin2 ϕ Proc. IEEE Int. Conf. Broadband
at reception,
γ̄ γ̄ sin2 ϕ sin2 ϕ+gγ̄
Multimedia Syst. Broadcasting , pp. 1-4, 2017.
[12] J. Montalban et al., Error Propagation in the
2) Appendix 5 : P roof of M GF in (20) and (24) :
Cancellation Stage for a Multi-Layer Signal Re-
( ) ∫∞
( )
ception,Proc. IEEE Int. Conf. Broadband Mul-
− singγ mγ
e(− γ̄ ) dγ
m m−1
Mm − sing2 ϕ , γ̄ = e 2ϕ
· mγ̄ m Γ(m)
γ
timedia Syst. Broadcasting, pp. 1-5, 2014.
0
( )
mm
∫∞ −γ m g
γ̄ + sin2 ϕ
[13] M. –S. Alouini and A. J. Goldsmith, A Uni-
= γ̄ m Γ(m) γ m−1
·e dγ ed Approach for Calculating Error Rates of Li-
0
nearly Modulated Signals over Generalized Fa-

∫ ∞ByP applying Laplace transform, we know that ding Channels, IEEE Trans. Commun., Vol. 47,
−st Γ(P +1)
t e dt = , where s > 0, P > −1. Thus No. 9, pp. 1324-1334, 1999.
0 S P +1
the MGF of Nakagami-m distribution can be written [14] M. S. Alouini and M. K. Simon, An MGF-
as Based Performance Analysis of Generalized Se-
( ) lection Combining over Rayleigh Fading Chan-
Mm − sing2 ϕ , γ̄ = mm
γ̄ m Γ(m) · (
m
Γ(m−1+1)
g
)m−1+1 nels, IEEE Trans. Commun., Vol. 48, No. 3,
+ sin2 ϕ
γ̄ pp. 401-415, 1999.
62 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.16, NO.2 August 2018

[15] S. Mumtaz, J. Rodriguez and L. Dai, mmWave

Massive MIMO A Paradigm for 5G, Academic
Press, 2016.
[16] S. I. Park et al., Field Test Results of Layered
Division Multiplexing for the Next Generation
DTV System, IEEE Trans. Broadcast, Vol. 63,
No. 1, pp. 260-266, 2017.
[17] A. Goldsmith, Wireless Communications, Cam
bridge University Press, 2005.
[18] G. L. Stuber, Principles of Mobile Communica-
tion, Kluwer Academic Publishers, 2002.
[19] J. W. Craig, A new, simple, and exact result
for calculating the probability of error for two-
dimensional signal constellations, Proc. IEEE
Int. Conf. Military Commun., pp. 571-575, 1991.
[20] J. G. Proakis, Digital Communications, McGraw
Hill, 1989.

Tanapong Khumyat was born in Tak,

Thailand, in 1978. He received Ph.D.
degree in March 2014 from Suranaree
University of Technology. His Ph.D. re-
search was concerned with MIMO co-
mmunication systems. He is currently
lecturer in Department of Electrical En-
gineering at the Rajamangala Univer-
sity of Technology Lanna Tak. His re-
search area concerns wireless communi-
cation systems.