Fifth Order versus Third Order Intermodulation Distortion
Lisa Yong1, Mohamad Kadim Suaidi2, Anderson Kho Ngap Chai1, Awangku Abdul Rahman3
1
School of Engineering,
Swinburne University of Technology (Sarawak Campus),
93576 Kuching, Sarawak, Malaysia.
2
3
Faculty of Electronics and Computer Engineering,
Kolej Universiti Teknikal Kebangsaan Malaysia,
Locked Bag 1200, Ayer Keroh,
75450 Melaka, Malaysia.
Research and Innovation Management Centre (RIMC),
Universiti Malaysia Sarawak,
94300 Kota Samarahan, Sarawak, Malaysia.
Abstract – The third order intermodulation distortion
(IMD3) rather than the fifth order intermodulation
distortion (IMD5) is often considered when examining
the performance of an analogue fibre optic system.
This paper aims at looking into the possibility of the
IMD5 being comparable to the IMD3. The IMD5
generated when the laser is intensity-modulated by the
multiple carriers signal is simulated and compared to
the IMD3. Individual IMD5 level is found to be lower
than the IMD3 when the RF powers of the individual
carriers are low or the modulation index per channel is
low. However the composite IMD5 level at channel
position could exceed the composite IMD3 especially
when modulation index increases. This is contributed
by the number of IMD5 being greater than that of the
IMD3 based on counting. Hence indicating the IMD5
should not be neglected in examining the performance
of the analogue fibre optic system driven by multiple
carriers.
Keywords: IMD3 ; IMD5
1. Introduction
The third order intermodulation distortion (IMD3)
is often considered as the limitation of the analogue
optical fibre systems such as the subcarrier
multiplexed systems and hybrid radio fibre systems as
seen in Figure 1 [1], [2]. This distortion is produced
when two or three carriers of different frequencies are
supplied to a nonlinear system. The electrical-tooptical block in Figure 1 is a nonlinear device.
The input to the electrical-to-optical converter
(E/O) can consist of a single carrier or several equally
spaced carriers. The IMD3 tends to fall in-band and
might reduce the dynamic range [3]. Therefore it is
1-4244-1435-0/07/$25.00©2007 IEEE
often taken into consideration when examining the
performance of the analogue optical fibre systems [1].
Figure 1. Basic configuration of the hybrid radio fibre
systems.
The fifth order intermodulation distortion is the
next odd order distortion after IMD3 which also has the
tendency to fall in-band.
The fifth order
intermodulation distortion is often considered to be
negligible as their individual levels might be a great
deal lower than the second and third orders distortions.
Individual IMD5s might be insignificant as compared
to the IMD3s but their levels might escalate if their
quantities increase and the input magnitude intensifies.
Therefore the paper aims at comparing the IMD5 and
IMD3 generated when the laser diode is intensity
modulated by multiple carriers that are equally spaced
to determine the possibility of the individual or
composite IMD5 produced is comparable to the IMD3
This paper is organized as follows. The first part
introduces the third and fifth order intermodulation
distortions; and the analogue fibre optic system. The
second part provides the methods used to model the
laser diode in order to quantify the levels of the IMD5
and IMD3 produced. The individual and composite
IMD5 and IMD3 produced by intensity modulating the
laser diode are shown, compared and discussed in the
next part. The conclusion is given the last part.
im (t ) ⎡
ds(t )
d 2 s (t ) ⎤ ⎡
ds (t )
2
= ⎢ D1 s(t ) + E1
+ F1
+
⎥ − Ls(t ) + Ms(t )
qV
dt
dt 2 ⎦ ⎢⎣
dt
⎣
2. Laser Model
The laser diode is modeled using Volterra Series,
meaning that this device is considered as a weakly
non-linear system with memory.
The model
formulation begins by rearranging and combining the
single mode laser rate equation pair as given in
equations (1) and (2) [4] to give an output to input
relation as given in equation (3):
dN (t )
I
N (t )
=
−
− g 0 (N (t ) − N 0 )(1 − εS (t ))S (t )
τs
dt
qV
dS (t )
N (t ) S (t )
= Γg 0 ( N (t ) − N 0 )(1 − εS (t ) )S (t ) + Γβ
−
dt
τs
τp
(1)
(2)
where qV is the product of electron charge and active
region volume (Am3s), I is the injected current (A), τp
is the photon lifetime (s), β is the spontaneous
coupling coefficient, Γ is the optical confinement
factor, ε is power gain compression parameter (m3), τs
is the carrier lifetime (s), N0 is the transparent carrier
density (m-3), g0 is the optical power gain (m3s-1), N(t)
is the carrier density (m-3) and S(t) is the photon
density (m-3).
I 0 + ΔI N 0 ⎧ d (Sb + s (t ) ) (Sb + s(t ) )
=
+⎨
+
−
qV
dt
τs ⎩
τp
N ⎫⎛ 1 1
Γβ 0 ⎬⎜ +
τ s ⎭⎝ Γ τ s
⎡ d (Sb + s(t ) ) (Sb + s(t ) )
N ⎤ ⎪⎫
+
− Γβ 0 ⎥ ⎬
⎢
τp
τ s ⎦⎥ ⎪⎭
dt
⎣⎢
b
When the laser diode is represented by Volterra
Series, its output s(t) is expressed as [4], [5], [6]:
s (t ) = ∑ ∫ L ∫ hn (u1 Kun ) ⋅ ∏ im (t − ul )du1 K dun
∞
∞
n =1
∞
−∞
n
−∞
l =1
(5)
where hn(u1, . . .,un) is the n-th order Volterra Series
kernel. The exponential growth or probing method is
used to solve equation (5) with ΔI in equations (4) and
(5) is substituted with equation (6) [6], [7]:
ΔI = ∑ e jωi t = e jω1t + e jω 2t + L + e jω n t
n
i =1
1
D1 + j ω 1 E1 − ω12 F1
H 2 (ω 1 , ω 2 ) = −
⎤
⎥
⎥×
⎥
⎥⎦
(6)
(7)
1
H 1 (ω1 ) ⋅ H 1 (ω 2 ) ⋅ H 1 (ω 1 + ω 2 ) ⋅ G 2 (ω1 , ω 2 )
2
Where
2
G2 (ω1 , ω2 ) = −2 L − jM ⋅ (ω1 + ω2 ) + N TF ⋅ (ω1 + ω 2 )
(8)
(9)
where I0 and ΔI are the laser bias and modulating
current due to the intercepted mobile signal. Sb and
s(t) are the laser steady-state photon density and
photon density due to the modulating current. Then,
terms with Γg (1 − ε (S + s (t ) ))(S + s (t ) ) + Γ β as the
b
(4)
H 1 (ω 1 ) =
(3)
0
2
⎡
d 2 s (t ) ⎤
⎛ ds(t ) ⎞
4 ds (t )
+ ⎢Us(t ) 5 + Vs(t ) 3 ⎜
+ Xs(t ) 4
⎟ + Ws(t )
⎥ +L
dt
dt 2 ⎦
⎝ dt ⎠
⎣⎢
where ωn is the n-th angular frequency of the input.
The Fourier Transform of the first to fifth orders
Volterra Series kernels obtained are:
⎡
⎤
⎢
⎥
1− β
⎢
⎥
β
⎢ Γg (1 − ε (S + s (t ) ))(S + s(t ) ) + Γ ⎥
b
b
0
⎢⎣
τ s ⎥⎦
⎧⎡
d ⎪⎪⎢
1
+ ⎨⎢
dt ⎪⎢ Γg (1 − ε (S + s(t ) ))(S + s(t ) ) + Γ β
b
b
⎢ 0
τs
⎩⎪⎣
2
2
d 2 s (t )
⎛ ds(t ) ⎞
⎛ ds(t ) ⎞ ⎤ ⎡
3
2 ds (t )
+ 2Gs(t )⎜
+ N TF ⎜
⎟
⎟ ⎥ + ⎢ Rs(t ) + σs (t )
2
dt
dt
⎝ dt ⎠
⎝ dt ⎠ ⎥⎦ ⎢⎣
2
2
2
d s(t ) ⎤ ⎡
ds(t ) ⎤
4
2 ⎛ ds (t ) ⎞
3 d s (t )
+ Gs(t ) 2
+ P2 s(t ) 3
⎟ + P1s(t )
⎥ + ⎢Os(t ) + Qs(t ) ⎜
⎥
dt 2 ⎦ ⎢⎣
dt 2
dt ⎦
⎝ dt ⎠
N TF s (t )
τs
denominator in equation (3) are expanded using Taylor
Series. Collecting the s(t) terms leads to equation (4)
that relates the modulating input with its output [4],[5]:
⎧
⎪⎪ 3
1
H 3 (ω 1 , ω 2 , ω 3 ) = − H 1 (ω 1 + ω 2 + ω 3 ) ⋅ ⎨ ∑ 2H 1 (ω k ) ⋅
6
⎪ kk ≠, mm, n≠=n1
⎪⎩ m < n
H 2 (ω m , ω n ) ⋅ G 2 (ω k , ω m + ω n ) + H 1 (ω 1 )H 1 (ω 2 )H 1 (ω 3 ) ⋅
G 3 (ω 1 , ω 2 , ω 3 )
⎫
⎪
⎬
⎪
⎭
G3 (ω1 , ω2 , ω3 ) = 6 R + 2 jσ ⋅ (ω1 + ω2 + ω3 )
(10)
where
− 2G ⋅ (ω1 + ω2 + ω3 )
2
(11)
H 4 (ω 1 , ω 2 , ω 3 , ω 4 ) =
−1
H 1 (ω 1 + ω 2 + ω 3 + ω 4 ) ⋅
24
⎧
⎪⎪ 4
⎨ ∑ 6 H 1 (ω k )H 3 (ω l , ω m , ω n )G 2 (ω k , ω l + ω m + ω n )
⎪ kk ≠,l l, m≠ m, n≠=n1
⎩⎪ l < m < n
+
∑ [4 H (ω
4
k , l , m , n =1
k ≠l ≠m ≠ n
k < l ,< m < n
2
k
, ω l )H 2 (ω m , ω n )G 2 (ω k + ω l , ω m + ω n )
+ 2 H 1 (ω k )H 1 (ω l )H 2 (ω m , ω n )G 3 (ω k , ω l , ω m + ω n )] +
H 1 (ω 1 )H 1 (ω 2 )H 1 (ω 3 )H 1 (ω 4 )G 4 (ω 1 , ω 2 , ω 3 , ω 4 )
⎫
⎪
⎬
⎪
⎭
where
G 4 (ω1 , ω 2 , ω 3 , ω 4 ) = 24O + j 6 P2 ⋅ (ω1 + ω 2 + ω 3 + ω 4 )
H 5 (ω1 , ω 2 , ω3 , ω 4 , ω5 ) =
2
(13)
−1
H 1 (ω1 + ω 2 + ω3 + ω 4 + ω5 ) ⋅
120
⎡
5
⎢
⎢ ∑ [24 H 1 (ω k ) ⋅ H 4 (ωl , ω m , ω n , ω p )G2 (ω k , ωl + ω m + ω n + ω p )
⎢ k ,l , m , n , p =1
⎢⎣lk<≠ml ≠<mn <≠ np ≠ p
+
∑ [12H (ω
5
k ,l , m , n , p =1
k ≠l ≠ m≠ n ≠ p
k <l ,m< n< p
2
k
]
]
∑ [4 H (ω )⋅ H (ω , ω )⋅ H (ω
5
k , l , m , n , p =1
k ≠l ≠ m ≠ n ≠ p
l < m, n< p
+
1
k
2
l
m
2
n
]
, ω p )⋅ G3 (ω k , ωl + ω m , ω n + ω p )
∑ [2 H1 (ωk )⋅ H1 (ωl )⋅ H1 (ωm )⋅ H 2 (ωn , ω p )⋅ G4 (ωk , ωl , ωm , ωn + ω p )]
5
k ,l , m , n , p =1
k ≠l ≠ m≠ n ≠ p
k <l < m,n< p
+ H 1 (ω1 ) ⋅ H 1 (ω 2 ) ⋅ H 1 (ω3 ) ⋅ H 1 (ω 4 ) ⋅ H 1 (ω5 ) ⋅ G5 (ω1 , ω 2 , ω3 , ω 4 , ω5 )
⎤
⎥
⎥
⎥
⎦
(14)
where
G5 (ω1 ,ω2 ,ω3 ,ω4 ,ω5 ) = 120U + j 24W ⋅ (ω1 + ω2 + ω3 + ω4 + ω5 )
− 24 X ⋅ (ω1 + ω2 + ω3 + ω4 + ω5 )
2
(15)
The RF input power is related to the modulating input
current (ΔI) according to equation (16) [8]:
ΔI = 2 Pi / Ri
(16)
where Pi is the input power and Ri is the input
resistance of 50Ω. ΔI is related to the optical
modulation per carrier by [9]:
mopt = ΔI / (I 0 − I th )
∑m
N
n =1
2
(18)
opt
The mtot should be less than 1 to avoid overmodulation,
which can lead to higher levels of distortions [1].
The IMD5 and IMD3 generated based on two-tone
IM test are looked into first, where the intersection
between the IMD23 and carrier or third order input
intercept point (IIP3) as well as the intersection
between the IMD25 and carrier or fifth order input
intercept point (IIP5) are found. Then the composite
IMD5 and IMD3 produced at carrier position due to
intensity modulation of the laser by six equally spaced
carriers are compared.
3.1 Two-tone IM Test
, ωl )H 3 (ω m , ω n , ω p )G2 (ω k + ωl , ω m + ω n + ω p )
+ 6 H 1 (ω k ) ⋅ H 1 (ωl ) ⋅ H 3 (ω m , ω n , ω p )⋅ G3 (ω k , ωl , ω m + ω n + ω p ) +
mtot =
3. Results and Discussion
(12)
− 6 P1 ⋅ (ω1 + ω 2 + ω 3 + ω 4 )
where Ith is the bias and threshold currents. When the
input to the laser diode is made up of several carriers, the
total modulation index (mtot) needs to be considered. The
mopt is related to mtot by [1]:
(17)
The laser diode parameters used in this paper refers to
those in [6]. The carrier frequencies, bias current and
optical modulation index per carrier used for the twotone IM tests are indicated in Table 1. The lower
optical modulation index translates to -37.5 dBm in
Test 1. The mopt of 0.5 translates to -2.13dBm of RF
input power.
Table 1. Parameters used in the first two tone IM
measurements simulation.
Parameters
First carrier frequency, f1 (MHz)
Second carrier frequency, f2 (MHz)
Bias current, I0
Optical modulation index per carrier, mopt
Test 1
890.2
890.4
25mA
0.008519
Test 2
890.2
890.4
25mA
0.5
The IMD3 and IMD5 generated due to the twotone tests are referred to as IMD23 and IMD25, shown
in Figures 2 and 3. The IMD23s appeared at 890 and
890.6 MHz, adjacent to the carriers in both figures.
890 MHz is equivalent to 2f1-f2 while 890.6 MHz, 2f2f1. The IMD23 at 890.6MHz was found to be slightly
higher than that at 890MHz for both tests. The IMD23
due to mopt of 0.008519 was 133.34dB lower than the
first carrier or -133.34dBc. The IMD23 increased to 62.6dBc when mopt of 0.5 was used.
The IMD25 appeared at 889.8 and 890.8 MHz,
adjacent to the IMD23. 889.8 MHz is equivalent to
3f1-2f2 while 890.8 MHz, 3f2-2f1. The IMD25 in the
first two tone IM measurement was -174.75 dBc,
41.41dB lower than the IMD23. However, the second
test indicated that the IMD25 was -33.34 dBc and 29.34
dB higher than the IMD23. As such, the IMD5 should
not be totally ignored.
1 .10
18
1 .10
17
1 .10
Photon density (1/meter cube)
16
15
1 .10
1 .10
14
1 .10
13
1 .10
12
11
1 .10
1 .10
10
1 .10
9
1 .10
8
8.898 .10
8.9 .10
8
8.902 .10
8.904 .10
Frequency (Hz)
8
8
8
8.906 .10
8
8.908 .10
8
Carriers
Third order IMD
Fifth order IMD
Photon density (1/meter cube)
Figure 2. The outcome of the two-tone IM test using mopt of
0.008519.
1 .10
20
1 .10
19
1 .10
18
1 .10
17
than the IMD3.
Both IMDs intersected at
approximately -16.75dBm, meaning the IMD23 and
IMD25 are equal at -16.75dBm or when mopt is 0.093.
The IMD23 was found to be greater than the IMD25 for
RF input power less than -16.75dBm. The opposite
occurs when the RF input power exceeds -16.75dBm.
Therefore, the IMD5 tends to be more significant than
the IMD3 as the RF input power increases.
The IMD25 was at the same level as the first
carrier when the carriers’ RF power was 6.25dBm.
Hence, the IIP5 is 6.25dBm and translates to mopt of
1.312. The IIP3 is 29.25dBm. Beyond these RF input
powers, the IMD23 and IMD25 exceeded the carriers.
Hence, the IIP3 alone is not sufficient to characterize
the analogue fibre optic system’s nonlinearity.
3.3 Composite IMD3 and IMD5
The six equally spaced carriers from 890.2 MHz
to 910.2MHz were employed in simulating the
composite IMD3 and IMD5. The IMD3 of type 2f1-f2
and f1+f2-f3, and the IMD5 of type f1+f2+f3-f4-f5, 3f1-f2f3, 3f1-2f2, 2f1+f2-2f3, f1+f2+f3-2f4, 2f1+f2-f3-f4
appearing at carrier position were counted. The
number of carrier position IMD5 are greater than the
IMD3 as shown in Figure 5.
80
1 .10
8.898 .10
8.9 .10
8
8.902 .10
8.904 .10
Frequency (Hz)
8
8
8
8.906 .10
8
8.908 .10
Num ber of IMD
70
16
8
Carriers
Third order IMD
Fifth order IMD
Figure 3. The two-tone IM when the mopt is 0. 5.
60
50
40
30
20
10
0
1
3.2 IMD23 and IMD25 versus RF Power
The IMD23 and IMD25 for a bias current of 25mA,
and RF input power ranging between -120 to 40dBm is
shown in Figure 4.
3
4
5
6
Carrier Number
Fifth order intermodulation distortion
Third order intermodulation distortion
Figure 5. The sum of IMD3 and IMD5 at carrier positions.
The composite IMD5 and composite IMD3 levels
at carrier positions at 25mA bias current and low mtot
are given in Figure 6.
400
7.00E+11
200
6.00E+11
Photon Densities (m -3)
Log(Photon density [1/meter cube])
2
0
5.00E+11
4.00E+11
3.00E+11
2.00E+11
200
1.00E+11
120
100
80
60
40
20
RF input power (dBm)
0
20
40
First Carrier
Third order IMD
Fifth Order IMD
Figure 4. Carrier, IMD23 and IMD25 versus input power for
25mA bias current.
Both distortions in log scale increased
proportionally with the RF input power in dBm. The
IMD25 trend is twice steeper than that of the IMD23,
signifying the IMD5 tends to increase more rapidly
0.00E+00
1
2
3
4
5
6
Carrier Number
Composite fifth order intermodulation distortion level
Composite third order intermodulation distortion level
Figure 6. The composite IMD3 and IMD5 at carrier positions
with low mtot.
The composite IMD5 levels were very much lower
than the composite IMD3 levels, signifying that IMD3
60.00
50.00
40.00
dB
was more dominant than the IMD5 at low mtot. The
composite IMD3 maximized at the middle occupied
channels as more IMD3s fell onto these channels, as
compared to the end channels. The composite IMD5
level maximized at the fifth carrier rather than the
fourth since the frequency of the fifth carrier was
higher than the fourth.
The composite IMD5s and IMD3s appearing at
carrier positions for the mtot of one with 25mA bias
current are given in Figure 7. Both distortions
increased in magnitude with the composite IMD5
levels exceeding the composite IMD3 levels. This
signifies that the composite IMD5 was more dominant
than the composite IMD3 at the mtot of one with low
bias current.
30.00
20.00
10.00
0.00
Carrier
Number
Carrier-to-composite
IMD3
Carrier-to-composite IMD5
Carrier Number
Carrier-to-IMD
Figure 9. Carrier-to-composite intermodulation distortion
ratios for low mtot.
4. Conclusion
1.80E+17
1.60E+17
-3
Photon Density (m )
1.40E+17
1.20E+17
1.00E+17
8.00E+16
6.00E+16
4.00E+16
2.00E+16
0.00E+00
1
2
3
4
5
6
Carrier Number
Composite fifth order intermodulation distortion level
Composite third order intermodulation distortion level
Figure 7. The composite IMD3 and IMD5 at carrier positions
with low mtot.
The composite IMD3 and IMD5 are both lower
than the carrier magnitudes for both low and high mtot.
Thus resulting in positive carrier-to-composite IMD3
and carrier-to-composite IMD5 ratios as in Figures 8
and 9.
190.00
180.00
170.00
160.00
dB
150.00
140.00
130.00
120.00
110.00
100.00
1
2
Carrier-to-composite IMD3
3
4
Carrier Number IMD5
Carrier-to-composite
5
6
Carrier-to-IMD
Figure 8. Carrier-to-composite intermodulation distortion
ratios for low mtot.
The
carrier-to-composite
intermodulation
distortion curve seemed superimposed with the carrierto-composite IMD3 curve for low mtot. The carrier-tocomposite IMD5 curve appeared below the carrier-tocomposite IMD3 curve as given in Figure 9 as the
composite IMD5 at carrier position were more
dominant than the composite IMD3 for high mtot.
The IMD5 has been found to be more dominant than
the IMD3 when the mopt is high or RF input power
increases for low bias current when the laser diode is
modulated by six equally carriers. However, the
opposite occurs when the mopt is low. Nonetheless, the
IMD5 should not be ignored when examining the
performance analogue fibre optic system. Further
simulations can be performed to investigate the IMD5
generated when more carriers are employed with
different frequency arrangement and RF power levels.
References
[1] B. Wilson, Z. Ghamssemlooy and I. Darwazeh, Ed.
Analogue
Optical
Fibre
Communications,
Manchester : IEE, 1995.
[2] “Radio-frequency signal transport through optical
fibres”, International Telecommunications Union
Recommendation ITU-RF.1332-1* , 1997-1999.
[3] H. Al-Raweshidy and S. Komaki, Ed. Radio over
fiber technologies for mobile communications
networks, Boston: Artech House, 2002.
[4] N. Tayebi, and M. Kavehrad, “Laser nonlinearity
compensation for radio subcarrier multiplexed fiber
optic transmission systems,” IEICE Trans.
Commun., vol. E76-B, pp.1103-1114, Sept.1993.
[5] T. K. Biswas, and W. F. McGee, “Volterra Series
analysis of semiconductor laser diode,” IEEE
Photonics Technology Letters, vol.3, pp. 706 –708,
Aug. 1991.
[6] H. M. Salgado and J. J. O’Reilly, “Experimental
validation of Volterra series nonlinear modelling for
microwave subcarrier optical systems”, IEE Proc.
Optoelectron., vol. 143, pp. 209-213, August 1996.
[7] S. A. Maas, Nonlinear microwave circuits. New
York: IEEE Press, 1997.
[8] R. Ohmoto, H. Ohtsuka and H. Ichikawa, “Fiberoptic microcell radio systems with a spectrum
delivery scheme,” IEEE Journal on Selected Areas
in Communications, vol. 11, pp. 1108-1117, Sept.
1993.
[9] G. Keiser, Optical fiber communications. 3rd ed.,
Boston: McGraw Hill, 2000.