Copyright (c) 2011 IEEE. Personal use is permitted.

For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
1
Spatial Pulse Position Modulation for Optical
Communications
Wasiu O. Popoola, Enrique Poves, and Harald Haas
AbstractThis paper proposes a low complexity spatial modu-
lation (SM) scheme that combines spatial shift keying (SSK) with
pulse position modulation (PPM) for optical wireless communi-
cation systems. SM is a multi-transmitter technique for achieving
increased data rate over the traditional on-off keying (OOK) and
PPM signalling methods. Analysis of the error performance of
the system in the presence of noise is presented and validated via
simulations. There is a perfect agreement between the simulation
and the theoretical analysis for the case of M = 2 bits/symbol
and other values of M at symbol error rate (SER) of less
than 10
2
. At higher SER values the analytical prediction is
about 1.2 dB more than that of the simulation. We also show
the energy efciency/bandwidth requirement trade-off involved
when determining the system parameters such as the number
of transmitters and the number of bits per symbol M. Using
fewer transmitters improves the energy efciency but requires
more bandwidth. Moreover, the error performance of SPPM is
dictated by both the individual channel gains of the multiple
transmitters and the difference between these channel gains or
path losses. Hence, distinct channel gains are a prerequisite in
spatial modulation. An experimental set up to measure and show
the dependence of the channel gains on the relative position of
the transmitter to the receiver is also presented. These measured
channel parameters are then used to evaluate the system error
performance. The performance of the SPPM is also compared,
in terms of energy and spectral efciencies, with the classical
SSK and repetition coded (RC) schemes in which the multiple
transmitters are used to transmit the same data simultaneously.
The results show the SPPM as a multi-transmitter signalling
scheme that combines the energy efciency of the PPM with the
high spectral efciency of the SSK.
Index TermsOptical communications, visible light communi-
cations, spatial modulation, PPM, optical wireless communica-
tions, channel gain measurement, optical MIMO.
I. INTRODUCTION
The spatial modulation (SM) is a modulation technique in
which the transmitters spatial position is also used to convey
some information [1][3]. Within a symbol duration only one
transmitter is active while all the rest are idle. Since different
transmitter positions have different channel gain/pathloss, the
signals at the receiver are thus different for each transmitter
position. This therefore makes it possible for the bits impressed
on each unique transmitter position to be retrievable at the
receiver. When the active transmitter is only sending a constant
power with no additional pulse pattern or signal constellation
on top, then we have space shift keying (SSK). The SM
technique is currently attracting a considerable attention in the
W.O. Popoola, E. Poves and H. Haas are with the Institute for Digital
Communications School of Engineering, University of Edinburgh, UK, EH9
3JL. Email: {w.popoola, e.poves, h.haas}@ed.ac.uk
Copyright (c) 2011 IEEE. Personal use of this material is permitted
design of multiple-input/multiple-output (MIMO) systems [3].
Compared to schemes like the optical OFDM [4][6], the SM
is of much lower complexity [7], [8] and is not as sensitive to
device non-linearity [9][12].
However in optical communications much of the work
has been on the optical SSK [13][15] - which is a special
case of SM. The SM with its multiple optical transmitters
could be used as a form of low complexity MIMO technique
when combined with multiple photodetectors as is the case in
[13]. In [14] a MIMO optical wireless system with SSK is
reported. The authors used an 1/2 rate convolutional forward
error correction (FEC) to improve the SSK performance,
resulting a gain of up to 7 dB over the uncoded SSK. In
the optical SM system reported in [16], the transmitters do
not send a contant power output but different power levels.
These varying transmit power levels are then used to carry
additional bits in a style similar to pulse amplitude modulation
(PAM). This approach is however susceptible to increased
errors as the effect of channel gains makes the distinction
between the active transmitter and the received power level
more challenging.
This paper presents a scheme that combines the optical
SSK with pulse position modulation (PPM) to form spatial
PPM (SPPM). Here, the active transmitter is used to transmit
a PPM signal pattern rather than a constant optical power
or PAM. This scheme benets from both the simplicity of
SSK and the energy efciency of PPM. When compared with
the conventional on-off keying technique that transmits only
one bit/symbol, the scheme is capable of increasing the data
rate by many folds. The achievable increase in data rate is
proportional to the number of transmitters and the size of the
signal constellation used. For an SPPM system with N
t
optical
transmitters and an L-slot PPM, the number of bits transmitted
during each symbol duration is log
2
(N
t
L).
The SPPM is particularly an interesting technique for visible
light communications (VLC). When using visible LEDs for
illumination, multiple LEDs are often used within a lamp
due to the limited luminous ux of the individual LEDs. The
presence of these multiple visible LEDs can thus be leveraged
on by SM to provide a low complexity and high data rate VLC
systems . To this end, VLC application of SPPM is considered
in this work and the channel measurements carried out are
based on white LEDs as the optical transmitters.
The contributions of this work are as follows: 1) pre-
sentation of SPPM, a full edged optical SM system that
combines the simplicity of multiple transmitter SSK with
the energy efciency of PPM, 2) theoretical analysis of the
error performance is presented and validated via simulation,
Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
2
3) experimental measurements of the channel gain/path loss
using white LEDs as transmitters and a PIN photodetector
receiver are equally presented. These measurements are then
used to evaluate the error performance of the system, 4) the
spectral and energy efciencies of SPPM are also compared
with that of the classical techniques such as RC-OOK, RC-
PPM and SSK.
The rest of the paper is arranged as follows: the SPPM
is fully described in section II and the error performance is
presented in section III. The experimental set-up that uses
white LEDs to measure the channel gains are presented
in section IV. Section V contains discussion of the results
including the comparison of SPPM with SSK, RC-OOK and
RC-PPM while the concluding remarks are given in section VI.
II. SPPM DESCRIPTION
The SPPM combines the classical space shift keying (SSK)
that is extensively discussed in [15], [17], [18] with the
pulse position modulation to achieve a full edged optical
spatial modulation. In this scheme, the spatial location of each
transmitter (white LED in this case) corresponds to log
2
(N
t
)
bits.
A. Implementation of SPPM
As in the classical SSK, only one source is transmitting at
any given symbol duration while the rest are off or idle. During
a symbol duration, the active source transmits an L-PPM
pulse pattern. As a result, the total number of bits transmitted
during each symbol duration is given by M = log
2
(LN
t
) as
against log
2
(N
t
) and log
2
(L) in the case of SSK and PPM
respectively. The SPPM is further described in Fig. 1 for the
case M = 4 bits/symbol, N
t
= 4 and 4-PPM. For example,
symbol 5 with binary representation 0101 is transmitted
as follows: the rst log
2
(N
t
) most signicant bits (2 in this
case) determine the transmitter number while the remaining
bits determine the pulse position. Therefore, symbol 5 is
transmitted by LED 2 with a pulse in the second slot of the
4-PPM pulse pattern. P
tj
is the peak transmit power of the
jth LED.
With respect to the regular PPM, the advantages of SPPM
can be summarised as:
1) For the same bandwidth requirement (that is slot/chip
duration) as the regular L-PPM, the SPPM provides
an additional log
2
(N
t
) bits/symbol that translates into
higher data rate.
2) For the same number of bits/symbol M as the regular
L-PPM, the SPPM technique will result in a bandwidth
reduction that is proportional to N
t
. The bandwidth
reduction factor B
rf
is obtained as follows. The chip
duration T
ppm
c
and T
sppm
c
for a regular PPM and SPPM
are given as:
T
ppm
c
=
TM
2
M
(1)
T
sppm
c
=
TMN
t
2
M
(2)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
1101
1110
1111
L
E
D

2
L
E
D

1
L
E
D

4
L
E
D

3
(0)
(5)
(10)
(15)
Sample pulse pattern
Symbol
number
Binary
value
Fig. 1. SPPM description
where T is the bit duration. The SPPM bandwidth
reduction factor B
rf
is then the ratio of (1) to (2) as
given by (3).
B
rf
=
1
N
t
(3)
B. SPPM Detection
Considering the jth LED transmitting a single pulse in slot
m of the total L slots in a line-of-sight link conguration, the
transmitted signal is degraded by the additive white Gaussian
noise n(t) and path loss h
j
. For a receiver using a PIN
photodiode with responsivity R, the received electrical signal
is thus given by the following:
r(t) = s
j
m
(t) +n(t), 0 t T
s
, (4)
where the symbol duration T
s
= LT
sppm
c
, and s
j
m
(t) is as
dened by (5) for a constant peak power P
tj
= P
t
.
s
j
m
(t) =
_
h
j
RP
t
for (m1)T
sppm
c
t mT
sppm
c
0 elsewhere
(5)
To determine what symbol was transmitted, the received
signal is passed through a parallel bank of L crosscorrelators
that computes the projection of r(t) onto the L basis functions
{f
k
(t)} given by (6). The correlator output, r
k
= s
j
mk
+ n
k
for k = 1, 2, . . . L is then used to make a decision based on
the maximum likelihood criterion. That is, the detector nds
the signal s
j
m
that is closest to the received signal r(t). The L
noise components {n
k
} are zero-mean uncorrelated Gaussian
random variables with a common variance
2
n
=
1
2
N
0
; where
N
0
represents the noise power spectral density. The receiver
structure is depicted by Fig. 2.
f
k
(t) =
_
1

T
sppm
c
for (k 1)T
sppm
c
t kT
sppm
c
0 elsewhere
(6)
Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
3
D
e
t
e
c
t
o
r
Sample at
.
.
.
Fig. 2. SPPM receiver structure
The distance metric D
_
r, s
j
m
_
, m = 1, 2, . . . , L; j =
1, 2, . . . , N
t
is computed as follows:
D
_
r, s
j
m
_
=
L

k=1
_
r
k
s
j
mk
_
2
(7)
where s
j
mk
=

_
0
s
j
m
(t)f
k
(t) dt, for k = 1, 2, . . . L and the
joint probability density function of r conditioned s
j
m
being
transmitted is given by (8).
p
_
r|s
j
m
_
=
1
(N
0
)
L/2
exp
_

_
L

k=1
_
r
k
s
j
mk
_
2
N
0
_

_ (8)
where m = 1, 2, . . . , L and j = 1, 2, . . . , N
t
. The minimum
of the LN
t
distance metric D(r, s
j
m
) gives the estimated pulse
position and transmitting LED, m and respectively. That is,
[ m, ] = arg max
m,j
p
_
r|s
j
m
_
= arg min
m,j
_
D(r, s
j
m
)

(9)
This process is equivalent to obtaining m from max(r) and
then nding the transmitter estimate that gives the signal
closest to m. The detection of the SPPM symbol, just as in
all other SM systems, is aided by the fact that the path loss
{h
j
}, j = 1, 2, . . . , N
t
is different for each LED.
III. ERROR PERFORMANCE ANALYSIS
Following the detection process of section II-B, the error
performance of SPPM will now be formulated in this section.
A. Symbol Error Rate Evaluation
The error performance analysis will involve evaluating the
probability of a correct symbol detection, P
sppm
c,sym
, from which
the symbol error probability can then be deduced as P
sppm
e,sym
=
1 P
sppm
c,sym
. For a correctly decoded pulse postion, (i.e. m =
m), the probability of correctly decoding the transmitting LED
is:
P
tx
c
= p ( = j| m = m) (10)
Thus,
P
sppm
c
= P
tx
c
p ( m = m)
= p ( = j| m = m) p ( m = m) (11)
B. Pairwise Error Probabilities
In order to evaluate P
tx
c
and p ( m = m), we shall rst
nd their pairwise error probabilities (PEP). For an LED j
transmitting a pulse in slot m in an N
t
= 2 system, the
PEP
pj
m
of deciding in favour of LED p instead of j is given
as:
PEP
pj
m
= p
_
D
_
r, s
j
m
_
> D(r, s
p
m
)
_
(12)
where
D
_
r, s
j
m
_
=
L

i=1
i=m
(n
i
)
2
+
_
r
m
h
j
_
E
s
_
2
(13)
D(r, s
p
m
) =
L

i=1
i=m
(n
i
)
2
+
_
r
m
h
p
_
E
s
_
2
(14)
and the energy per symbol E
s
= (RP
t
)
2
T
sppm
c
. The PEP
is derived from the foregoing as:
PEP
pj
m
=

_
|hphj|

Es/2
1
_
2
2
n
exp
_

x
2
2
2
n
_
dx
= Q
_
|h
p
h
j
|
_

s
2
_
(15)
where
s
=
_
Es
N0
is the symbol SNR. Now considering N
t
equiprobable transmitters and using the union bound technique
[19], the average transmitter error, conditioned on the pulse
occupying slot m is:
P
tx
e

2
N
t
Nt1

p=1
Nt

j=i+1
PEP
pj
m
(16)
From this the average probability of correctly decoding the
transmitter becomes
P
tx
c
= 1
2
N
t
Nt1

p=1
Nt

j=i+1
Q
_
|h
p
h
j
|
_

s
2
_
(17)
In nding p ( m = m) we consider the jth transmitting LED
with a pulse in slot m in an L = 2 system, similarly the
PEP
j
qm
of deciding in favour of slot q instead of m is given
as:
PEP
j
qm
= p
_
D
_
r, s
j
m
_
> D
_
r, s
j
q
__
=

_
hj

Es
1
_
2
2
qm
exp
_

x
2
2
2
qm
_
dx (18)
Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
4
where
2
qm
= N
0
and
D
_
r, s
j
m
_
=
_
r
m
h
j
_
E
s
_
2
+
_
r
q
_
2
(19)
D
_
r, s
j
q
_
=
_
r
m
_
2
+
_
r
q
h
j
_
E
s
_
2
(20)
Since all the slots are equally likely to contain the pulse,
the PEP between any two slots is therefore given as:
PEP
j
= Q(h
j

s
) (21)
Now for N
t
equally likely transmitters and using the union
bound technique, the p ( m = m) becomes:
p ( m = m) =
L 1
N
t
Nt

j=1
_
1
L 1
Q(h
j

s
)
_
(22)
From (11), (17) and (22), the average probability of symbol
error in SPPM given by (23) is thus obtained. This expression
shows that for a given SNR, the error performance strongly
depends on the individual path loss of the transmitters as well
as the difference in their path losses.
IV. CHANNEL GAIN MEASUREMENT
In order to evaluate the system performance, the values
of the channel pathloss {h
j
}, j = 1, 2, . . . , N
t
are measured
using the experimental set-up of Fig. 3. A continuous stream
of pulses is sent from the transmitter positioned at (0,0) to the
receiver at coordinate (x,y) as shown in the set-up. The pulse
duration and its repetition rate are 400 ns and 625000 pulses/s
respectively with an average LED drive current of 30 mA. At
the receiver, the amplitude of the received undistorted pulses
is taken and averaged over several measurements to reduce the
effect of noise. This measured mean amplitude is proportional
to the received optical power. The path loss is thus obtained
from the ratio of the received signal amplitude at point (x,y)
to that of the chosen reference point (30,0). The experiment
covers a range of 100 cm horizontally and 25 cm trans-
verse transmitter-receiver separation. The range limitations are
imposed by the small optical power and radiation angle of
the OVL 5521 white LED used in the experiment. The full
parameters of the components used in the experiment are given
in Table I.
TABLE I
COMPONENTS CHARACTERISTICS
Transmitter
White LED OVL5521
Angle of half intensity,
1/2
7
1
2

Luminous intensity @ 20 mA 18000 mcd

CIE chromaticity X-axis: 0.31
Y-axis: 0.30
Receiver
PIN Photodiode (Si) S6967
Angle of half sensitivity,
1/2
67

Responsivity at = 460 nm 0.25 A/W

Photodiode active area 26.4 mm
2
Blue lter pass band (460 30) nm
The common white LED is produced by mixing the
light from a blue LED chip with the yellow light emitted
Data acquisition
and logging
LPF
TIA
Amp
PIN
diode
Blue filter
x (cm)
separation
y


(
c
m
)
s
e
p
a
r
a
t
i
o
n
LED driver
Signal
gen.
LED
(x,y)
(0,0)
Fig. 3. Pathloss measurement set-up
by a phosphor downconversion material. These phosphor-
downconverted white LEDs have very low modulation band-
width due to the slow response of the phosphor. And one way
around this bandwidth limitation is to allow only the blue
component of the white light to go through the receiver by
inserting a blue lter in the receiver front-end [20].To ensure
that this limited bandwith does not cause distortions in the
received pulses and thereby impair the measured path loss;
the bandwidth of the transmitted pulse (2.5 MHz in this case)
must be less than the 9 MHz measured bandwidth of the
white LED. The measured path loss values are presented in
Fig. 4. These are the values that will hence be used to evaluate
the system error performance in the following section.
25 20 15 10 5 0 5 10 15 20 25
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
00
Transverse separation (cm)
N
o
r
m
a
l
i
s
e
d

C
h
a
n
n
e
l

G
a
i
n

(
d
B
)


30 cm
50 cm
80 cm
100 cm
Horizontal range
Fig. 4. The pathloss values for different horizontal and vertical separations
V. RESULTS AND DISCUSSIONS
Using the error analysis of section III and the channel
gains of section IV, the error performance for different link
congurations are plotted against the SNR per bit
b
=
s
/M
in Fig. 5. The error performance is shown for different values
of bits per symbol M and N
t
= [2, 4]. Locations A(30,0) and
B(30,-5) are used for the case of 2 white LED transmitters
while for 4 transmitters, the white LEDs are positioned at
locations [A(30,0); B(30,-5); C(80,-10); D(100,20)] relative to
the receiver. From the measurement results plotted in Fig. 4,
Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
5
P
sppm
e,sym
1
L 1
N
t
_
_
1
2
N
t
Nt1

p=1
Nt

j=p+1
Q
_
|h
p
h
j
|
_

s
2
_
_
_
Nt

l=1
_
1
L 1
Q(h
l

s
)
_
(23)
the normalised pathloss values for the positions [A, B, C, D]
are given as [1; 0.31; 0.1; 0.04] respectively. To validate the
analysis, simulation results are also included. The error perfor-
mance plots show a perfect agreement between the simulation
and the theoretical analysis results for M = 2 and other
scenarios when SER < 10
2
. However at SER = 10
1
and
M = [3, 5], the theory requires about 1.2 dB in
b
more than
the simulation predicted; this is not unexpected with the union
bound technique used in the analysis.
5 10 15 20 25 30 35
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
SNR,
b
(dB)
S
E
R


Simulation N
t
=4, M=3
Simulation N
t
=4, M=5
Theory N
t
=4, M=3
Theory N
t
=4, M=5
Theory N
t
=2, M=2
Theory N
t
=2, M=4
Simulation N
t
=2, M=2
Simulation N
t
=2, M=4
N
t
= 2
N
t
= 4
Fig. 5. Error performance against
b
for M = [2, 3, 4, 5] at transmitter
locations [A(30,0), B(30,-5)] and [A(30,0), B(30,-5), C(80,-10), D(100,20)]
for Nt = 2 and 4 respectively
For the case of N
t
= [2, 4] under consideration, the
b
required to attain an SER of 10
6
is plotted against M in
Fig. 6. The more LEDs are used in spatial modulation, the
higher the attainable data rate but this requires distinct channel
gains for all the LEDs and higher SNR as shown, (see Fig. 6).
This is the reason why using 2 transmitters requires less
b
(up to about 18 dB less) than using 4 transmitters as Fig. 6
shows. The observed superior
b
for N
t
= 2 is however at
the cost of higher bandwidth requirement that increases with
M as implied by (2). This hence demonstrates the trade-off
involved between bandwidth and energy efciencies.
For a xed number of LEDs (N
t
= 2 in this case), the
impact of different transmitter positions on the system is
illustrated in Fig. 7 for ve different locations. It is observed
that as the difference between the two distinct channel gains
narrows, the SNR increases for a given SER. In addition, the
SNR also depends on the value of each channel gain not
just their difference. For instance, at an SER < 10
6
, the
channel gain sets [0.1, 0.04] and [0.04, 1] produce the same
performance and both require about 33 dB in
b
at SER =
10
6
. It can thus be inferred from Fig. 7 that at high SNR,
the performance is dictated by the value of the smaller channel
2 3 4 5 6 7 8 9 10
15
17
19
21
S
N
R
,

b



(
d
B
)

@

S
E
R

=

1
0

6

,


N
t

=

2













No of bits/symbol M
2 3 4 5 6 7 8 9 10
30
35
S
N
R
,

b


(
d
B
)

@

S
E
R

=

1
0

6
,

N
t

=

4






















N
t
= 2
N
t
= 4
Fig. 6. SNR at SER = 10
6
against M, at transmitter locations [A(30,0),
B(30,-5)] and [A(30,0), B(30,-5), C(80,-10), D(100,20)] for Nt = 2 and 4
respectively
gain and at lower SNR values by the difference in the channel
gains.
10 15 20 25 30 35 40
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
SNR
b
(dB)
S
E
R


[A,B][1,0.31]
[E,B][0.1,0.31]
[C,F][0.1,0.04]
[G,D][0.03,0.04]
[H,A][0.04,1]
[H(30,15),A(30,0)]
[E(50,10),B(30,5)]
[A(30,0),B(30,5)]
[G(100,25),D(100,20)]
[C(80,10),F(80,15)]
Channel gains
Fig. 7. Error performance for Nt = 2, M = 2 and different transmitter
locations
In Table II, we present the expressions that show how
the SPPM compares with other multiple transmitter optical
communications signalling schemes in terms of SER, average
transmitted optical power P
ave
, spectral efciency
spec
and
the number of bits per symbol M. In this table, P
modtype
t
represents the transmitted peak optical power for the stated
modulation type modtype and is the return-to-zero (RZ)
OOK duty cycle. The spectra efciency here is dened as
Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
6
TABLE II
COMPARISON OF DIFFERENT MODULATION SCHEMES
Mod. Type Error Probability
spec
Pave M
RC-OOK(RZ) P
ook
e,sym
= Q
_
1
Nt
Nt

i=1
h
i
_
(RP
ook
t
)
2
Ts
2N
0
_
; 0 < 1

2
P
ook
t
1
RC-L-PPM P
ppm
e,sym
(L 1) Q
_
1
Nt
Nt

i=1
h
i
_
(RP
ppm
t
)
2
T
ppm
c
N
0
_
1
L
log
2
L
P
ppm
t
L
log
2
L
SSK P
ssk
e,sym

2
Nt
Nt1

j=1
Nt

i=j+1
Q
_
|h
j
h
i
|
_
(RP
ssk
t
)
2
Ts
2N
0
_
log
2
Nt P
ssk
t
log
2
Nt
SPPM P
sppm
e,sym
expression is as given by (23)
1
L
log
2
(NtL)
Pt
L
log
2
(NtL)
the ratio of the bit rate to the bandwidth requirement of the
system. Where the bandwidth requirement itself is taken as
the reciprocal of the pulse duration in each of the signalling
technique.
5 10 15 20 25 30
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Number of slots, L
S
p
e
c
t
r
a
l

e
f
f
i
c
i
e
n
c
y

s
p
e
c



(
b
i
t
s
/
s
/
H
z
)


SPPM
RCPPM
SSK
OOK, =1
(a)
5 10 15 20 25 30
15
20
25
30
35
40
45
Number of slots, L
S
N
R

b

(
d
B
)

a
t

S
E
R

=

1
0

6


SPPM
RCPPM
SSK
OOK, =1
(b)
Fig. 8. Performance comparison of SPPM with RC-PPM, RC-OOK and
SSK (a) Spectral Efciency and (b) SNR to attain SER of 10
6
for Nt = 2
positioned at A(30,0) and B(30,-5)] and for different values of L
To graphically illustrate how the SPPM compares with other
modulation techniques, we consider the case of N
t
= 2 and
using the expressions in Table II, we show in Fig. 8 the spectral
efciency
spec
as a function of L and the SNR required to
attain an SER of 10
6
for the different modulation techniques.
For a fair comparison, all the modulation techniques are as-
sumed to have the same P
ave
as the SPPM. For the conditions
shown, the maximum
spec
for PPM is 0.5 while the SPPM
could attain twice this value. It should be noted that SSK also
has a maximum
spec
of 1 but requires about 6 dB more SNR
at the stated SER. Moreover, as L increases the energy saving,
in terms of SNR, by the SPPM over OOK and SSK increases.
For instance at L = 16 SPPM requires 20 dB and 26 dB less
SNR than OOK and SSK respectively. Furthermore, at lower
values of L, the SPPM has higher
spec
than PPM (double at
L = 2) and as L increases its
spec
approaches that of the PPM
as shown in Fig. 8a. These results therefore show very clearly
how the SPPM combines the energy efciency of the PPM
with the high spectral efciency of the OOK and/or SSK.
VI. CONCLUSION
An optical SM scheme that combines optical SSK with PPM
has been introduced and analysed. An analytical expression for
the error performance has also been presented and validated
via simulation. There is a perfect agreement between the
simulation and the presented theoretical analysis results for the
case of 2 bits/symbol and other scenarios when SER < 10
2
.
We have also shown the trade-off that exists between energy
efciency and bandwidth requirement when choosing system
parameters such as the number of transmitter and M. Using
fewer transmitters results in improved energy efciency but
requires more bandwidth. Moreover, the performance of SPPM
is dictated by both the individual channel gain values and
the difference between them. Hence, distinct channels are a
prerequisite in spatial modulation. And an experimental set up
to measure and show the dependence of these channel gains on
the relative position of the transmitter to the receiver has also
been presented. Error performance have also been presented
using the measured channel gains. The performance of the
SPPM has also been compared with that of OOK, PPM and
SSK in terms of energy and spectral efciencies. The results
show clearly the SPPM as a multi-transmitter scheme that
Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
7
combines the simplicity and spectral efciency of the SSK
with the energy efciency of the PPM modulation technique.
REFERENCES
[1] R. Mesleh, Spatial Modulation: A Spatial Multiplexing Technique
for Efcient Wireless Data Transmission, Ph.D. dissertation, Jacobs
University, Bremen, Germany, Jun. 2007.
[2] R. Mesleh, H. Haas, S. Sinanovi c, C. W. Ahn, and S. Yun, Spatial
Modulation, IEEE Trans. Veh. Technol., vol. 57, no. 4, pp. 2228
2241, July 2008.
[3] M. Di Renzo, H. Haas, and P. M. Grant, Spatial Modulation for
Multiple-Antenna Wireless Systems: A Survey, IEEE Commun. Mag.,
vol. 49, no. 11, pp. 182191, Nov. 2011.
[4] J. Armstrong, OFDM for Optical Communications, Journal of Light-
wave Technology, vol. 27, no. 3, pp. 189204, Feb. 2009.
[5] O. Gonzalez, R. Perez-Jimenez, S. Rodriguez, J. Rabadan, and A. Ayala,
OFDM Over Indoor Wireless Optical Channel, Optoelectronics, IEE
Proceedings, vol. 152, no. 4, pp. 199204, Aug. 2005.
[6] J. Armstrong and A. Lowery, Power Efcient Optical OFDM, Elec-
tronics Letters, vol. 42, no. 6, pp. 370372, Mar. 16, 2006.
[7] R. Mesleh, H. Haas, C. W. Ahn, and S. Yun, Spatial Modulation
A New Low Complexity Spectral Efciency Enhancing Technique, in
IEEE International Conference on Communication and Networking in
China (CHINACOM), Beijing, China, Oct. 2527, 2006, pp. 15.
[8] J. Jeganathan, A. Ghrayeb, and L. Szczecinski, Spatial Modulation:
Optimal Detection and Performance Analysis, IEEE Commun. Lett.,
vol. 12, no. 8, pp. 545547, 2008.
[9] D. Dardari, V. Tralli, and A. Vaccari, A Theoretical Characterization
of Nonlinear Distortion Effects in OFDM Systems, IEEE Transactions
on Communications, vol. 48, no. 10, pp. 17551764, Oct. 2000.
[10] H. Elgala, R. Mesleh, and H. Haas, Impact of LED nonlinearities
on optical wireless OFDM systems, in 2010 IEEE 21st International
Symposium on Personal Indoor and Mobile Radio Communications
(PIMRC), sept 2010, pp. 634 638.
[11] , Non-linearity Effects and Predistortion in Optical OFDM Wire-
less Transmission Using LEDs, Inderscience International Journal of
Ultra Wideband Communications and Systems (IJUWBCS), vol. 1, no. 2,
pp. 143150, 2009.
[12] , A Study of LED Nonlinearity Effects on Optical Wireless Trans-
mission using OFDM, in Proceedings of the 6
th
IEEE International
Conference on wireless and Optical communications Networks (WOCN),
Cairo, Egypt, Apr. 2830, 2009.
[13] R. Mesleh, R. Mehmood, H. Elgala, and H. Haas, Indoor MIMO
Optical Wireless Communication Using Spatial Modulation, in Proc. of
the IEEE International Conference on Communications (ICC10), Cape
Town, South Africa, May 2227 2010, pp. 15.
[14] R. Mesleh, H. Elgala, and H. Haas, Optical Spatial Modulation,
IEEE/OSA Journal of Optical Communications and Networking, vol. 3,
no. 3, pp. 234244, Mar. 2011, ISSN: 1943-0620.
[15] T. Fath, M. D. Renzo, and H. Haas, On the performance of space
shift keying for optical wireless communications, in IEEE Global
Communications Conference (Globecom 2010) - Workshop on Optical
Wireless Communications, Miami, FL, USA, December 2010, pp. 990
994.
[16] T. Fath, H. Haas, Marco Di Renzo, and R. Mesleh, Spatial Mod-
ulation Applied to Optical Wireless Communications in Indoor LOS
Environments, in Proc. of the Global Communications Conference
(GLOBECOM), Houston, Texas, USA, 59 Dec. 2011, pp. 15, 5 pages.
[17] M. Di Renzo and H. Haas, Improving the performance of space
shift keying (SSK) modulation via opportunistic power allocation,
Communications Letters, IEEE, vol. 14, no. 6, pp. 500 502, 2010.
[18] J. Jeganathan, A. Ghrayeb, L. Szczecinski, and A. Ceron, Space Shift
Keying Modulation for MIMO Channels, IEEE Transaction on Wireless
Communications, vol. 8, no. 7, pp. 36923703, Jul. 2009.
[19] J. G. Proakis, Digital Communications, 4th ed., ser. McGraw-Hill Series
in Electrical and Computer Engineering, S. W. Director, Ed. McGraw-
Hill Higher Education, December 2000.
[20] J. Grubor, S. Randel, K. Langer, and J. Walewski, Bandwidth Efcient
Indoor Optical Wireless Communications with White Light Emitting
Diodes, in In the Proceeding of the 6
th
International Symposium
on Communication Systems, Networks and Digital Signal Processing,
vol. 1, Graz, Austria, Jun. 2325, 2008, pp. 165169.
Wasiu O. Popoola holds a National Diploma in electrical Engineering from
The Federal Polytechnic, Ilaro, Nigeria, a First Class (Hons.) in Electronic
and electrical engineeringfrom Obafemi Awolowo University, Nigeria in 2002
and an MSc with Distinction in optoelectronic and communication systems
from Northumbria University at Newcastle upon Tyne, UK in 2006. He was
later awarded his PhD degree in 2009 at the same Northumbria University for
his research work in free-space optical communications. During his PhD, he
was awarded the Xcel Best Engineering and Technology Student of the year
2009.
He worked briey,during his national service year, as a Teaching Assistant
at Nnamdi Azikiwe University, Awka, Nigeria between 2003 and 2004 and as
a Research/Teaching Assistant in the Northumbria Communication Research
Laboratory of Northumbria University between 2006 and 2009. He is currently
a researcher with the Institute for Digital Communications, University of
Edinburgh, UK working on visible light communications. He has published
over 30 peer reviewed articles in the area of optical wireless communications
and co-authored 2 invited book chapters. One of his book chapters with the
title Terrestrial Free-Space Optical Communications pulished with InTech
has been downloaded over 5000 times as of June 2012. Another of his papers
was ranked 2nd most downloaded paper in terms of the number of full
text downloads within IEEE Xplore database in 2008, from the hundreds
of papers published by IET Optoelectronics since 1980. He is also a co-
author of the book titled: Optical Wireless Communications: System and
Channel Modelling with MATLAB published by CRC Press in August 2012.
His research interests include Optical (wireless and bre) Communications,
Digital Communications and Signal Processing. Dr. Popoola is an associate
member of the Institute of Physics (IoP).
Enrique Poves (S06-M10) was born in Madrid, Spain, in 1978. He
received the M.Sc and Ph.D degrees in telecommunication engineering from
Universidad Politecnica de Madrid, Spain, in 2004 and 2011, respectively.
Until 2011 he continued his research on optical wireless systems and
applications for indoor and aerospace environments. In 2011 he joined
the Institute for Digital Communications of the University of Edinburgh
as a Research Fellow. His main research interests are in optical wireless
communication systems, with particular focus on MIMO techniques and
visible light communication applications, and circuit design.
Harald Haas holds the Chair of Mobile Communications at the Institute
for Digital Communications (IDCOM) at the University of Edinburgh and he
currently is the CTO of a university spin-out company, PureVLC Ltd. His main
research interests are interference coordination in wireless networks, spatial
modulation and optical wireless communications. Prof. Haas holds 23 patents.
He has published more than 50 journal papers including a Science Article and
more than 150 peer-reviewed conference papers. Nine of his papers are invited
papers. Prof. Haas has co-authored a book entitled Next Generation Mobile
Access Technologies: Implementing TDD with Cambridge University Press.
Since 2007 Prof. Haas has been a Regular High Level Visiting Scientist
supported by the Chinese 111 program at Beijing University of Posts and
Telecommunications (BUPT). He was an invited speaker at the TED Global
conference 2011, and his work on optical wireless communications was listed
among the 50 best inventions in 2011 in Time Magazine.