Experimental comparison of terahertz and infrared
data signal attenuation in dust clouds
Ke Su,
1,
* Lothar Moeller,
2
Robert B. Barat,
3
and John F. Federici
1
1
Department of Physics, New Jersey Institute of Technology, 322 King Blvd., Newark, New Jersey 07102, USA
2
Bell Laboratories/Alcatel-Lucent, 791 Holmdel-Keyport Road, Holmdel, New Jersey 07733, USA
3
Otto York Department of Chemical Engineering, New Jersey Institute of Technology, 322 King Blvd.,
Newark, New Jersey 07102, USA
*Corresponding author: ks265@njit.edu
Received July 6, 2012; revised September 26, 2012; accepted September 27, 2012;
posted September 27, 2012 (Doc. ID 172090); published October 18, 2012
In order to study and compare propagation features of terahertz (THz) links with infrared (IR) links under different
weather conditions such as turbulence, fog, and dust particles, THz and IR free space communication links
at 625 GHz carrier frequency and 1.5 μm wavelength, respectively, with a maximum data rate of 2.5 Gb∕s have
been developed. After propagating through the same channel perturbation caused by dust, attenuation of the
carrier frequencies by dust as well as scintillation effects on both channels are analyzed by measuring the power
attenuation and bit error rates. Attenuation by the presence of dust degrades the IR channel but exhibits almost no
measurable impact on the THz signal. Numerical simulations of THz attenuation with different dust concentra-
tions are presented and agree with the measured results. © 2012 Optical Society of America
OCIS codes: 060.2605, 040.2235, 140.3070, 290.5930, 010.1300.
1. INTRODUCTION
Recently, terahertz (THz) communication has attracted more
attention because of its potential to support higher data rates
[
1,2]. According to Edholm’s law of bandwidth [3], data rates
of 5–10 Gb∕s will be needed 10 years from now. The increas-
ing demand of higher data rates for wireless communication
systems will likely lead soon to operational systems in the
lower THz frequency range.
Both indoor and outdoor scenarios of THz communications
have been considered. Research on indoor THz links has fo-
cused on various important channel features, and several
experiments have been conducted [
4–7]. Koch and his co-
workers mainly focus on the ray-tracing simulations of THz
indoor communication [
8]. However, our focus is on perfor-
mance degradation of outdoor THz communication under cer-
tain weather conditions such as humidity, rain, fog, dust, and
scintillation effects. Towards higher carrier frequencies, THz
can possibly enable reliable signaling at certain atmospheric
conditions where free space infrared (IR) communication
would fail. Some experiments have been done to investigate
the impact of weather conditions on the THz transmission link.
For example, Yang et al. [
9] recently characterize attenuation
of the water vapor form 0.2 to 2 THz using THz time-domain
spectroscopy and show higher attenuation than previously
measured or predicted in these THz transmission windows.
At a frequency region of above 100 GHz, the effect of rain has
been characterized at 103 GHz [
10], 120 GHz [11], and
355.2 GHz [
12]. Yamaguchi et al. [13] show the measurement of
scintillations due to wind effects on a propagating 125 GHz data
signal. Recently, we performed a direct experimental perfor-
mance comparison between THz and IR communication links
through fog, meaning that both signals experience the same fog
conditions [
14]. Mann [15] predicts that smoke has little or no
effect up to 1 THz due to the relatively small size of atmospheric
particulates compared to THz wavelengths. Federici and
Moeller [
1] theoretically estimate attenuation due to smoke
particles for both THz and IR range and conclude that IR
wavelengths are strongly attenuated while THz radiation can
propagate well through small airborne particulates such as
dust. However, to the best of our knowledge, no experimental
study of the attenuation from THz data signal in dust has been
published. We focus here on channel impairments caused by
dust, which are mainly due to particle scattering and refractive
index fluctuation (scintillation effects).
As discussed in [
1], local temperature, pressure, or humid-
ity gradients, which are generated by thermals and turbu-
lences near ground level, cause small real refractive index
fluctuations across the wave front of the beam. The fluctua-
tions of the real refractive index through the path of a beam
can cause random deflection and interference between differ-
ent portions of the wavefront and destroy the flat phase front
of an IR light beam; therefore, the beam cross section on the
receiver side appears as a speckle pattern (Fig.
1) with huge
local and temporal intensity variations. Scintillation effects
are a major limitation to the maximum transmission distance
limitation of IR communication links. According to the analy-
sis in [
1], since optical path length variations in the THz and IR
ranges are comparable, the relative magnitudes of the phase
variations are predominately determined by the electromag-
netic wavelength. The wavelength of THz at ∼625 GHz is ap-
proximately 320 times longer than the wavelength of 1.5 μm
light. Since scintillation effects are driven by phase variation,
consequently, scintillation and speckle effects in the THz
beam are expected to be significantly smaller than in the IR.
A THz and IR communications lab setup with a maximum
data rate of 2.5 Gb∕s at 625 GHz carrier frequency and
1.5 μm wavelength has been developed [
14]. The performance
2360 J. Opt. Soc. Am. A / Vol. 29, No. 11 / November 2012 Su et al.
1084-7529/12/112360-07$15.00/0 © 2012 Optical Society of America
degradation in both channels due to dust attenuation can be
simultaneously recorded and analyzed. We investigate attenua-
tion and scintillation effects caused by dust through bit error
rate (BER) performance and received power levels. Our inves-
tigations show for the first time how differently THz and IR
communication signals are degraded when passing through
the same dust conditions. Simulations of THz attenuation with
different dust concentrations are presented and show good
agreement with the experimental measurements.
2. EXPERIMENTAL SETUP
Figure 2(a) shows a block diagram and Fig. 2(b) shows a
photo of our experimental setup. Detailed features of the
THz link [
7,14] and IR link [14] were presented earlier. In the
present work, we focus on the details of the links which are
important for the communication channels through dust.
As shown in Fig.
2(a), the output of the horn antenna from
the THz transmitter is collimated by a THz lens with short
focal length (∼32 mm) to a beam with ∼20 mm diameter,
transmitted over a few meters distance and finally coupled
into a THz receiver horn similar to the transmitter antenna.
A 2.5 Gb∕s non-return-to-zero (NRZ) format signal is gener-
ated by a pulse pattern generator. A duobinary modulation
technique [
16,17] is utilized in the system for driving the THz
source, which enables signaling at high data rate, with rela-
tively compact spectrum and higher output power. An iris
with 8.5 mm aperture, inserted concentrically into the beam,
limits its total power to an amount that results in a BER of
about 10
−6
for an unloaded dust chamber. We chose to adjust
our setup at a BER of 10
−6
because statistical fluctuations
of the BER at levels of 10
−6
and higher do not significantly
impact conclusions we draw from our measurement results
[
14]. Also, BER of 10
−6
is a typical threshold for modern for-
ward error correction technology used in lightwave commu-
nication systems [
18].
The IR transmitter is driven by the same 2.5 Gb∕s NRZ data
pattern as our THz source. A 90∶10 single mode fiber coupler
launches a small fraction of the signal into photo detector
PD_1, which serves as monitor for the power entering the dust
Source
Receiver
IR Beam
Dust Clouds
Phase Front Distortions
Speckle
Fig. 1. (Color online) Air turbulence causes refractive index fluctua-
tions resulting in speckle (intensity variations at receiver).
Fig. 2. (Color online) (a) Schematic diagram of the THz and IR wireless communication link through a dust chamber, (b) photo of setup, and
(c) schematic diagram of the dust chamber design.
Su et al. Vol. 29, No. 11 / November 2012 / J. Opt. Soc. Am. A 2361
chamber. We expand the IR beam to about 20 mm diameter
by a fiber collimator, which is comparable to the THz beam
size before the iris. The collimated IR beam is superimposed
with the THz beam using a thin (2 μm) nitrocellulose mem-
brane beam splitter (Thorlabs) with 55% reflection ratio at
45° incident angle, and transmitted through the dust chamber.
After emerging from the chamber, the beam is deflected with a
similar beam splitter to spatially separate THz and IR signals.
A second beam splitter taps off a fraction of the IR power
leaving the chamber and launches it towards a large area
photo diode (effective area 19.6 mm
2
) of detector PD_2
which, in combination with PD_1, is used to determine the
power loss caused by the chamber load. The remaining beam
power enters via a fiber collimator a 1 × 2 multimode (MM)
fiber coupler with 50∶50 splitting ratio. Its output power is
launched to a low bandwidth photo detector (PD_3) with a
1 m long standard single mode fiber (SSMF) and to a dc-
coupled IR lightwave converter (Agilent 81495A) with 9 GHz
bandwidth. The lightwave converter which is accessible via
general purpose interface bus (GPIB) is used both for data
detection and measuring the optical power of the incoming
signal.
The output of PD_1, PD_2, and PD_3 are recorded via a
DAQ board with 16 bit resolution and maximum 10 kHz sam-
pling rate. This sampling rate is sufficiently high to track even
the fastest fluctuations of the signals. BERs, Radio-Frequency
(RF) power of the THz signal, and optical power of the IR sig-
nal are recorded via GPIB. We set the LabView time controller
at a clock rate of 500 ms to synchronize all recordings.
Our dust chamber is shaped like a cylinder with 30 cm dia-
meter, 30 cm height, with top and bottom plates. Figure
2(c)
shows the schematic diagram of the dust chamber design. A
known total mass of dust particles is placed inside the hopper,
which sits above the dust holder. An electronic pulse switch
controls an air valve and the released air removes the dust
from the dust holder and ejects it at a high speed into the
chamber from the top plate of the dust chamber. The hopper
can be easily refilled and it produces a fairly constant and re-
producible feed of dust. Air flows into the chamber with con-
stant volume speed through the holes from the bottom plate.
The holes, all with the same 0.18 cm diameter, are placed on 5
concentric set of circles with diameters 2.22, 4.76, 7.30, 9.84,
and 12.38 cm. The constant air volume flow rate is regulated
by an air flow controller to achieve specific flow rates. Two
Picarin windows which are highly transparent to both the
THz and IR beam are mounted at the input and output of the
chamber. After experiments are completed, the dusty air can
be extracted by the exhaust pump which is installed on the
top plate.
Bentonite powder is a mixture of clay formed from volcanic
ash decomposition and largely composed of montmorillonite
and beidellite. We use bentonite as dust particles for loading
the chamber. The average particle radius is 4.3 μm.
3. EXPERIMENTAL RESULTS
A. Signal Attenuation by Dust
After we launched constant IR power into the dust chamber, a
typical evolution of the attenuations in both THz and IR chan-
nels is shown in Fig.
3(a). When 0.08 g of dust is launched in
the propagation path of the beams, the maximum attenuation
(around 4.5 dB) of the IR light is clearly evident while the
impact on THz could be neglected (only 0.045 dB). As will
be shown by simulation (Section
4), the particle concentration
is estimated to be roughly 4 × 10
9
∕m
3
. The attenuation de-
creases with time since dust particles start falling down to the
bottom of the chamber. After 400 seconds, the IR transmitted
power recovers and approaches its original performance. Our
RF power meter is only accurate to within 0.01 dB, which
explains the rough quantization of the recorded THz signal
in Fig.
3(a). For the IR channel, a relationship between the at-
tenuation and the recorded BERs [Fig.
3(b)] is clearly visible.
The recorded BERs for the THz link in Fig.
3(b) verify again
that dust particles at our concentration levels (4 × 10
9
∕m
3
)
have little impact on the THz signal.
B. Comparison of Different Concentrations of Dust
Three different amounts of dust (0.05, 0.08, and 0.13 g) are
loaded in the dust chamber. Assuming injected dust particles
are dispersed equally over the volume of the chamber, the
dust concentrations can be estimated using
N
s
N
V
chamber
m
p
V
p
ρ
p
∕V
chamber
m
p
4
3
πr
3
ρ
p
∕V
chamber;
(1)
where N is number of particles, V
chamber
is the volume of the
chamber, m
p
is total mass of the dust that is loaded into the
chamber, ρ
p
is the mass density of the dust (2.7 g∕cm
2
), V
p
is
the volume of a dust particle, and r is the radius of the particle.
The corresponding concentrations for different amounts
of loaded dust (0.05, 0.08, and 0.13 g) are 2.5 × 10
9
∕m
3
,
0 100 200 300 400
0
0.01
0.02
0.03
0.04
0.05
Time (s)
Attenuation of THz Link (dB)
0 100 200 300 400
0
1
2
3
4
5
Attenuation of IR Link (dB)
THz Link
IR Link
(a)
0 100 200 300 400
-5.2
-5.1
Time (s)
log (BER) of THz link
0 100 200 300 400
-4
-2
log (BER) of IR link
THz link
IR link
(b)
Fig. 3. (Color online) (a) Attenuation of THz link and IR link as the
function of time, (b) Log(BER) of THz link and IR link as the function
of time.
2362 J. Opt. Soc. Am. A / Vol. 29, No. 11 / November 2012 Su et al.
4 × 10
9
∕m
3
, and 6.5 × 10
9
∕m
3
. Attenuation of the IR signal
and the corresponding BER as a function of time are mea-
sured for various dust amounts [Figs.
4(a) and 4(b)]. As ex-
pected, the attenuations vary significantly with changing
particle concentration. The attenuation of the THz link and
corresponding BER as a function of time are recorded as well
[Figs.
4(c)–4(d)]. In order to better visualize the dynamic re-
sponse, an offset has been added for the measurements cor-
responding to dust particle concentration of 4 × 10
9
∕m
3
and
6.5 × 10
9
∕m
3
. The THz signal exhibits only a minor decrease
in power.
C. Scintillation Impact on IR Link
In addition to the attenuation effects discussed above, scintil-
lation effects could also cause an IR power variation on the
receiver side. Because of the constantly changing dust
pattern, scintillation appears to be more significant when the
receiver has a small aperture. If scintillation effects were to
occur, we would expect more pronounced variations in the
power coupled into SSMF with small aperture (PD_3) com-
pared to the output of the free space large area detector
(PD_2) and MM fiber detector.
The variances of the detector signal can be further asso-
ciated with three different sources: the electronic noise as
well as scintillation effects caused by dust particles, and air
flow inside the chamber. Air flow generates fluctuations of
the optical detector output due to refraction index changes
even when no dust particles are launched. This noise contri-
bution we assess by the variance σ
2
air
. Fluctuations caused
by scintillation effects stemming from the dust particles we
account for by σ
2
dust-scintillation
. Since all noise contributions
are independent random processes, the total variance of the
voltage fluctuation σ
2
total
at the detector output can be ob-
tained by adding individual variances. However, except for
the electronic noise, the individual contributions to variance
depend on the optical power and detector features like its re-
sponsivity. To allow a fair comparison of the output fluctua-
tions from the three different detectors, we normalize their
variances with the low pass filtered and squared voltage value
V
2
t of their outputs. Thus, the normalized variance of total
detector output noise is given by
σ
2
total
∕V
2
tσ
2
electronics
∕V
2
tσ
2
air
∕V
2
initial
σ
2
dust-scintillation
∕V
2
initial
; (2)
where V
initial
stands for the detector output voltage at the
beginning of the experiment, when no dust particles were
launched but air is flowing through the chamber. The low
pass filtering of the detector output is chosen such that rela-
tively fast fluctuations stemming from electronic noise are
suppressed whereas slow processes like the decay of the dust
particle concentration and scintillations can be well captured
[Figs.
5(a) and 5(b)]. The filtering is performed by convoluting
the detector output with a rectangular impulse response
which leads to an averaging of 100 adjacent sampling points
from DAQ board output.
In our experience, at high sampling rates (∼10 kHz), DAQ
boards collect sufficient data for averaging to track the fastest
scintillation effects in the signals. We performed a cubic poly-
nomial fit and calculated the standard deviation for both out-
put signals from SSMF [Fig.
5(a)] and large area detector
[Fig.
5(b)] within 15 s time intervals. Fig. 5(c) shows the cubic
polynomial fit and calculated the standard deviation for signal
from the MM fiber detector, which is collected via GPIB with a
500 ms sampling rate. The cubic polynomial fit provides a rea-
sonable average trend for the data. The normalized variances
σ
2
dust-scintillation
∕V
2
initial
for different concentrations of the dust
0 100 200 300 400
-2
0
2
4
6
8
Time (s)
Attenuation of IR (dB)
2.5X10
9
/m
3
4X10
9
/m
3
6.5X10
9
/m
3
0 100 200 300 400
-6
-5
-4
-3
-2
-1
0
Time (s)
Log (BER) of IR Link
2.5x10
9
/m
3
4x10
9
/m
3
6.5x10
9
/m
3
0 100 200 300 400
0
0.05
0.1
Time (s)
Attenuation of THz Link (dB)
2.5x10
9
/m
3
4x10
9
/m
3
(+0.02 dB)
6.5x10
9
/m
3
(+0.06 dB)
0 100 200 300 400
-5.2
-5.1
-5
-4.9
-4.8
-4.7
-4.6
Time (s)
Log (BER) of THz Link
2.5x10
9
/m
3
4x10
9
/m
3
(+0.15)
6.5x10
9
/m
3
(+0.3)
(a) (b)
(c) (d)
Fig. 4. (Color online) (a) Attenuation of IR link for different dust concentrations, (b) Log(BER) of IR link for different dust concentrations,
(c) Attenuation of THz link for different dust concentrations, (d) Log(BER) of THz link for different dust concentrations (offset with the values
mentioned in the legend).
Su et al. Vol. 29, No. 11 / November 2012 / J. Opt. Soc. Am. A 2363
are calculated according to Eq. (2) and plotted in Fig. 5(d).
The brown triangles show as a reference the impact of air
turbulence in the chamber on the detector signal when no
dust is launched. They are comparably small but indicate that
the scintillation effects observable in dust clouds are due to
local, time-dependent variations in the dust particle density as
opposed to air turbulence. Apparently, the variations in the
power coupled into SSMF with small aperture (PD_3) are
more pronounced compared to the IR power detected by large
area free space detector (PD_2) and MM fiber detector, which
indicates that scintillation effects due to the presence of dust
particles are observable on the IR link. For the single mode
fiber detector, higher dust amounts produce a higher var-
iance. For the free space detector and MM fiber detector, the
normalized variance of the output power is smaller than the
one of the SSMF detector and it seems to be related to the dust
concentration, although our measurement results do not
clearly support such a statement. The reason is that scintilla-
tion effects are more pronounced for the small aperture de-
tector PD_3 than for the large area detector PD_2 and MM
fiber detector.
4. SIMULATION RESULTS OF THZ
ATTENUATION
In this section, the particle concentration as the function of
time is estimated using the evolution of the attenuation in
the IR channel according to Beers–Lambert law [
19] and Mie
scattering theory [
20]. Using the inferred time-dependent par-
ticle concentration, the THz attenuation as a function of time
is simulated by Mie scattering theory. A comparison of experi-
mental and simulation results shows good agreement.
According to the Beers–Lambert Law, the attenuation of
laser radiation through atmosphere can be expressed as
τλ;R
Pλ;R
Pλ; 0
e
−γλR
e
−σ
s
N
s
R
; (3)
where Pλ;R is optical power at the distance R, Pλ; 0 is the
initially emitted optical power, γλ is the extinction coeffi-
cient (per unit of length), σ
s
is the total extinction cross
section, and N
s
is particle concentration.
25 30 35 40
0.21
0.22
0.23
0.24
0.25
Time (s)
Voltage (V)
Polynomial Fit
SM Fiber Detector (PD
3
)
25 30 35 40
2.2
2.3
2.4
2.5
2.6
2.7
Time(s)
Voltage (V)
Polynomial Fit
Free Space (PD
2
)
25 30 35 40
0.035
0.036
0.037
0.038
0.039
0.04
Time (s)
Optical Power (uW)
MM Fiber Detetor
Polynomial Fit
(a) (b) (c)
0 50 100 150 200 250
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
x 10
-4
Time (s)
Normalizaed Standard Deviation
SM (low)
MM (low)
FS (low)
SM (medium)
MM (medium)
FS (medium)
SM (high)
MM (high)
FS (high)
Air Turbulance
(d)
Fig. 5. (Color online) (a) Output signal of single mode fiber detector with cubic polynomial fit. (b) Output signal of free space detector with cubic
polynomial fit. (c) Output signal of MM fiber detector with cubic polynomial fit. (d) Normalized standard deviation of different amount of dust for
single mode fiber detector, MM fiber detector, and free space detector.
2364 J. Opt. Soc. Am. A / Vol. 29, No. 11 / November 2012 Su et al.
