Sei sulla pagina 1di 6

Modeling and Analysis of the Effect of Impulsive Noise on

Broadband PLC Networks


Y. H. Ma, *P. L. So, E. Gunawan and Y. L. Guan
Network Technology Research Centre
School of Electrical and Electronic Engineering
Nanyang Technological University
Research TechnoPlaza, 4th Storey, Slab Block, 50 Nanyang Drive
Singapore 637553
*Email: eplso@ntu.edu.sg
Abstract
The impulsive noise is the main cause of packet errors in
power line communication networks. In this paper, the bit
error rate (BER) and medium access control (MAC)
protocol analytical models under impulsive noise
interference are developed. The accuracy of the developed
models is verified by computer simulations. In the
computer simulations, the Poisson arrival of the impulsive
noise is simulated. Through the analysis, it is shown that
OFDM outperforms QPSK for most of the impulsive
noise. OFDM can achieve good BER performance and
MAC protocol performance under impulsive noise
interference.
1. Introduction
Power line communication (PLC) networks can be
divided into home network and last mile access
network according to their applications. The applications
of PLC home network have been increasing recently due
to the development of HomePlug Powerline Alliance,
which has published its standard, HomePlug Specification
1.0 [1]. However, there are some challenges for
communications over power lines, such as attenuation,
multi-path effect and noise.
Noise in a power line channel is Non-additive White
Gaussian Noise (NAWGN) [2]-[6]. The noise is
categoried into four different types in [2] and extended
into five types in [3]. The five types of noise are colored
background noise, narrowband noise, periodic impulsive
noise asynchronous to the mains frequency, periodic
impulsive noise synchronous to the mains frequency and
asynchronous impulsive noise. The first three types of
noise usually remain stationary and are summarized as
background noise. The last two noise types are timevariant and are named as impulsive noise. The impulsive
noise has a short duration and a high power spectral
density. It may cause bit or burst errors in data
transmission. The arrival rate and power spectral density
of the impulsive noise are measured and a partitioned
Markov chain is developed to describe the impulsive
noise in [3]. Results in [2] reveal that the arrival of the
impulsive noise follows the Poisson distribution, and give
the distribution of the background noise. The
classification and characterization of impulsive noise on

indoor power lines used for data communications is


reported in [5]. The noise characteristics in Singapore
power line channels are measured and simulated in [6].
Multi-path effect is also a serious problem for PLC
because the distribution of power line is complicated. The
PLC multi-path is studied and the echo model is
developed in [10] and [11].
The impulsive noise and multi-path effects push to
consider communications techniques that can effectively
face such a hostile environment. OFDM is a good
candidate for broadband PLC since it permits to separate
overall transmitted data in many parallel independent substreams. The long symbol duration time makes OFDM
perform better than single carrier scheme for multi-path
channel. OFDM may perform better than single carrier
when the channel is interfered by impulsive noise, since it
can spread the impulsive noise over multiple symbols due
to inverse fast Fourier transform (IFFT).
For the early study of narrowband PLC network below
frequency of 1 MHz, some researches are involved in the
impulsive noise and its effect [7]-[12]. For the broadband
PLC, the BER performance of OFDM are simulated in
[13] and [14]. The impulsive noise effect on MAC
protocol performance for the PLC network is analyzed in
[15] and [16]. In this paper, the effect of the impulsive
noise on PLC networks is studied. Closed form formulas
for the BER performance of QPSK and OFDM under
impulsive noise interference are given. The closed form
formulas of MAC protocol performance based on packet
error rate (PER) are studied.
The paper is organized as follows. In Section 2, the
impulsive noise model is introduced. Section 3 gives the
analytical model of BER performance of QPSK and
OFDM modulation. Also, the performance of MAC
protocol npCSMA (non-persistent Carrier Sense Multiple
Access) based on BER and PER is discussed. In Section
4, analytical and simulation results are given and
compared. Conclusions are discussed in Section 5.
2. Impulsive Noise Model
In the power line channel, there are additive white
Guassian noise (AWGN) wk and impulsive noise ik . ik
is given by

ik = bk g k

(1)

Tnoise

where bk is Poisson process, which is the arrival of the


impulsive noise and g k is white Gaussian noise with
mean zero and variance 2 w 2 . This model can be
regarded as each transmitted data symbol being hit
independently by an impulsive noise with probability
distribution bk and with random amplitude g k .
Let ak be the transmitted signal, then the received
signal can be expressed as
rk = ak + nk

(2)

where
nk = wk + ik = wk + bk g k

(3)

The power distribution function (PDF) of the noise nk


is
pnk ( nkR , nkI ) = (1 bk )G ( nkR , 0, w2 )G ( nkI , 0, w2 )
+bk G (nkR , 0, w2 + i2 )G (nkI , 0, w2 + i2 )

(4)

T
White noise

Impulsive noise

Fig. 1: Impulsive noise and white Gaussian noise in a


time unit.
3.1. BER of QPSK under impulsive noise model
The total BER is the average of BER that under
impulsive noise interference and that under AWGN noise
as given in (8)
pb = pi pbi + p0 pbw = Tnoise pbi + (1 Tnoise ) pbw

G ( x, mx , x2 )

1
x 2

( x mx ) 2
2 x2

2 Eb
pbi = Q
N +N
0
i

k = 0,1, 2,...

(6)

The duration time of the impulsive noise Tnoise in a


given time period of T is shown in Fig. 1. The average
impulsive noise rate pi is defined as the result of the
average impulsive noise duration time in T. The AWGN
ratio p0 is the result of average AWGN time in T. Its
value is 1 pi . According to (6),

pi = [ eT
k =0

(T ) k
(kTnoise )] / T
k!

= Tnoise [ e T
k =1

= Tnoise

k 1

(9)

(10)

where Eb is the signal energy per bit, N 0 is AWGN

The impulsive noise has approximately a Poisson


distribution, which means the arrival of impulsive noise
follows a Poisson process with a rate of units per
second, so that the event of k arrivals in t seconds has the
probability distribution [4]:
k

2 Eb
pbw = Q
N
0

(5)

3. Performance Analysis of QPSK and OFDM under


Impulsive Noise Interference

Pk (t ) = e t ( t ) / k !

(8)

where pbi and pbw are the BER under the impulsive
noise and white Gaussian noise respectively.
According to the BER of QPSK in [17],

where nkR and nkI are the real and imaginary parts of
nk respectively and

Data

( T )
( T )
] = Tnoise [ e T
] (7)
(k 1)!
k!
k =0

power density, Ni is impulsive noise power density, and


Ni + N 0 is the power density of the impulsive noise
power density plus AWGN power density. Hence, the
BER of QPSK under impulsive noise interference is:
pb = Tnoise pbi + (1 Tnoise ) pbw
= Tnoise Q(

2 Eb
2 Eb
) + (1 Tnoise )Q(
)
Ni + N0
N0

(11)

For the packet with packet length L, PER is


p per = 1 (1 pb ) L

(12)

3.2. BER of OFDM under impulsive noise model


When the transmitted signal is an OFDM symbol, the
received symbol after front end filtering and sampling,
assuming perfect synchronization, timing and an ideal
channel, is given by [18]
rk =

N 1

n =1

an

j 2 nk
e N

+ wk + ik

k = 0, 1, 2,...., N 1
(13)

where an is QPSK or BPSK symbol and N is the number


of carriers. The transmitted symbols

{an }nN=01

recovered from the received sequence

{rn }nN=01

are
by

performing an N points discrete Fourier transform (DFT):


1

Rk =

N 1

rn e

j 2 nk
N

N n =1
= ak + Wk + I k

(14)

k = 0, 1, 2,...., N 1

where Wk is once again AWGN with variance 2 w2 , I k


is given by the DFT of impulsive noise as follows:
Ik =

N 1

n =1

in e

j 2 nk
N

k = 0, 1, 2,...., N 1

(15)

The impulsive noise is spread over N data symbols due


to the DFT operation, which is different from single
carrier system, in which the impulsive noise will affect
only one symbol. Hence the total noise is
N m = N w + pi Ni

(16)

where N w is the AWGN noise and Ni is the impulsive


noise. The BER is
2 Eb
pb = Q
N
m

S = Ge aG

(17)

where
Eb
Eb
ebno
=
=
N m N 0 + Tnoise Ni 1 + Tnoise

(18)

where is ratio of impulsive noise power to AGWN


power,
=

i2

w2
The signal-to-noise (SNR) ratio ebno is defined as
ebno =

a2

(19)

(20)

2 w2

Then
2 Eb
pb = Q
N
m

ebno
= Q 2
Tnoise
1
+

series of bits. When the packet is transmitted through the


power line channel, it may not be correctly transmitted
because of bit errors or burst errors. Hence, only when the
BER and PER of PLC networks are obtained, it can
analyze the MAC protocol performance.
The name of carrier sense in npCSMA stems from the
fact that nodes sense the existence of the carrier wave on
the communication channel. If a carrier wave is sensed on
the communication channel, the condition is called
busy; otherwise, it is called idle. CSMA is a protocol
that decides whether packet transmission should start as
the result of a carrier sense. When the result is busy, the
next action to avoid collision is needed. There are several
kinds of CSMA according to different actions taken after
the busy listening. In this paper, CSMA means nonpersistent CSMA. When packets are generated in a node,
the node starts the carrier sense. If the result of a carrier
sense is idle, the packet is transmitted to the channel
immediately. However, if the result of a carrier sense is
busy, the node stops the carrier sense, waits for a while,
and then starts the carrier sense again.
Throughput is defined as the average number of
successful transmitted packets in a packet transmission
time T seconds. The throughput of CSMA without
impulsive noise is

(21)

3.3. Performance of npCSMA


MAC protocols play important role in PLC networks.
In order to design a good MAC protocol, the physical
layer characteristics of PLC networks are important. Data
are transmitted in term of packets, which comprise a

1
G (1 + 2a ) + e aG

(22)

where G is the traffic offered to the PLC network, a is


the normalized propagation delay.
When the impulsive noise is considered, a successful
transmitted packet requires this packet not destroyed by the
impulsive noise, hence the throughput of npCSMA under
impulsive noise is
S = (1 pb ) L Ge aG

1
G (1 + 2a ) + e aG

(23)

where p b is the BER under a certain modulation scheme,


L is the length of packet.
4. Analytical and Simulation Results
The impulsive noise scenarios to be considered is
obtained from [3], which is measured and categorized into
three kinds of scenarios, namely heavily disturbed,
medium disturbed and weakly disturbed. For the
heavily disturbed impulsive noise, it was captured during
the evening hours in a transformer substation in an
industrial area. The medium disturbed was recorded in a
transformer substation in a residential area with detached
and terrace houses. The weakly disturbed was recorded
during nighttime in an apartment located in a large
building. The three impulsive noise scenarios are
tabulated in Table I. In the table, IAT is the Inter-Arrival

Time of the impulsive noise, which is the reciprocal of


arrival rate . TNoise is the average noise duration time.
TABLE I
PARAMETERS OF THE IMPULSIVE NOISE
Impulsive Noise Scenario

IAT

TNoise

I: Heavily disturbed
II: Medium disturbed
III: Weakly disturbed

0.0196s
0.9600s
8.1967s

0.0641ms
0.0607ms
0.1107ms

Fig. 2 gives the BER for QPSK and Fig. 3 shows the
BER for 512 carriers OFDM with QPSK on each carrier
for three kinds of impulsive noise. For a typical PLC
network, is determined. Hence, in the simulation, by
fixing and changing SNR, it can obtain the BER under
different signal power.
In Fig. 2, the increase of the SNR can improve the BER
performance when SNR is less than around 10dB. After
10dB, continuing to increase the SNR will not improve
the performance effectively. Furthermore, higher SNR
will introduce serious electromagnetic radiation problem
for PLC, which may interfere the shortwave radio.
For impulsive noise Scenario I, the best BER is only
1e-3. According to (12), for a 1kbits length packet, when
BER is 1e-3, PER is 0.632, which is not applicable for a
reliable communication system. For Scenarios II and III,
the best BERs are around 1e-5 and 1e-6, corresponding to
PERs of 0.01 and 0.001 respectively, which can be
regarded as reliable communication systems. Hence,
under heavily disturbed impulsive noise, QPSK cannot be
used. For medium and weakly disturbed impulsive noise,
QPSK can be used as modulation scheme.

Fig. 2: Analytical BER of QPSK under three scenarios of


impulsive noise interference.
In Fig. 3, BERs decrease smoothly with the increase of

SNR. When SNR is larger than 10dB, BERs will decrease


rapidly, which means for OFDM, increasing SNR will
improve the performance effectively. From this point of
view, OFDM performs better than QPSK. Comparing Fig.
2 with Fig. 3, it is obvious that for impulsive noise
Scenarios II and III, OFDM outperforms QPSK. For
impulsive noise Scenario I when SNR is larger than 8dB,
OFDM is better than QPSK. However, QPSK
outperforms OFDM when SNR is less than 8dB.

Fig. 3: Analytical BER of OFDM under three scenarios of


impulsive noise interference.
When the impulsive noise is medium or weakly
disturbed, OFDM performs better because the same
interference is spread out and causes less interference
over all N symbols due to DFT operation as in (14). For
the heavily disturbed impulsive noise Scenario I, when
SNR is low, the power of impulsive noise is higher than
signal. For QPSK, one impulsive noise will affect one or
several symbols, whereas in OFDM the high power
impulsive noise is spread into all N symbols and interferes
all symbols. For large SNR, the power of impulsive noise
is small compared with signal power, after spreading, its
effect on each symbol is small and leads to good BER
performance.
Fig. 4 gives the comparison between OFDM and
QPSK under impulsive noise Scenario I. The impulsive
noise to AWGN ratio is 10. From the figure, it is clear
that single carrier QPSK hits an error floor due to the
impulsive noise whereas OFDM does not exhibit the error
floor. At lower SNR, the performances of both OFDM
and QPSK are similar because in this region the system is
basically white noise limited and not impulsive noise
limited. Also, from this figure, the agreement of
simulation and analytical results is also demonstrated.
The arrival of the impulsive noise is simulated as a
Poisson process with the arrival rate of . The amplitude
of the impulsive noise is obtained according to the

impulsive noise to AWGN ratio . Fig. 5 is the impulsive


noise wave generated during the simulation of OFDM
BER performance in Fig. 4. In Fig. 5, it can be observed
that the impulsive noise and its amplitude are randomly
generated and its high amplitude compared with the white
Gaussian noise.

Fig. 4: Comparison of analytical and simulation results


for BER performance of QPSK and OFDM under
impulsive noise Scenario I.

By using OFDM, the throughput of PLC network drops


very small by impulsive noise interference, whereas using
single carrier QPSK, the throughput is only 0.1. Hence,
OFDM is a good choice to combat impulsive noise and
improve PLC network performance.

Fig. 6: Simulation and analytical results of throughput of


npCSMA under the heavily disturbed impulsive noise
interference.
5. Conclusions

Impulsive Noise
White Gaussian
Noise

Through the analysis, it is shown that single carrier


QPSK hits an error floor due to the impulsive noise
whereas OFDM does not exhibit the error floor. At lower
SNR, the performances of both OFDM and QPSK are
similar because in this region the system is basically
limited by white noise and not impulsive noise. The MAC
protocol performance based on the BER of QPSK and
OFDM has been analyzed. It comes to the conclusions
that under the heavily disturbed impulsive noise
interference, the network throughput of QPSK drops
significantly compared with that without impulsive noise
interference. However, OFDM can perform well with
impulsive noise interference. By using OFDM
modulation, it can effectively decrease the interference of
impulsive noise and achieve good network performance.

Fig. 5: Impulsive noise wave generated during simulation


of OFDM BER performance.
The simulation and analytical results of the throughput
analysis of npCSMA is given in Fig. 6. The network is
interfered by the impulsive noise Scenario I. The results
consist of the comparison of the throughput under QPSK
and OFDM modulation. In the analysis, the packet length
L is set as 1kbits. Bits error rate is obtained from Fig. 4.
In Fig. 4, when SNR=10dB, BERs for QPSK and OFDM
are 2e-4 and 4e-6 respectively. According to (12), PERs
for QPSK and OFDM are 0.18 and 0.0008 respectively.

Acknowledgements
The first author gratefully acknowledges the financial
support provided by Nanyang Technological University,
Singapore.
References
[1]

S. Gardner, The HomePlug standard for


powerline home networking, Proc. ISPLC 2001,
pp. 67-72, 2001.

[2]

O.G. Hooijen, A channel model for the


residential power circuit used as a digital
communication
medium,
IEEE
Trans.
Electromagn. Compat., vol. 40, no. 4, pp. 331336, 1998.

[12]

M.H.L. Chan, D. Friedman and R.W.


Donaldson, Performance enhancement using
forward error correction on power line
communication channels, IEEE Trans. on
Power Delivery, vol. 9, no. 2, pp. 645-653, 1994.

[3]

M. Zimmermann and K. Dostert, Analysis and


modeling of impulsive noise in broadband
powerline communications, IEEE Trans.
Electromagn. Compat., vol. 44, no. 1, pp. 249258, 2002.

[13]

K. Kuri, Y. Hase, S. Ohmori, F. Takahashi and


R. Kohno, Powerline channel coding and
modulation considering frequency domain error
characteristics, Proc. ISPLC 2003, pp. 221-225,
2003.

[4]

O.G. Hooijen, On the channel capacity of the


residential power circuit used as a digital
communications
medium,
IEEE
Communications Letter, vol. 2, no. 10, pp. 267198, Oct 1998.

[14]

E. Del Re, R. Fantacci, S. Morosi and R.


Seravalle, Comparison of CDMA and OFDM
techniques
for
downstream
power-line
communications on low voltage grid, IEEE
Trans. Power Delivery, vol. 18, no. 4, pp. 11041109, Oct 2003.

[5]

V. Degardin, M, Lienard, A. Zeddam, F.


Gauthier and P. Degauque, Classification and
characterization of impulsive noise on indoor
power line used for data communications, IEEE
Trans. Consumer Electron., vol. 48, no. 4, pp.
913-918, 2002.

[15]

H. Hrasnica, A. Haidine, and R. Lehnert,


Reservation MAC protocols for powerline
communications, Proc. ISPLC 2001, pp. 121128, 2001.

[16]

G. Bianchi and G. Conigliaro, An hybrid


reservation-polling MAC protocol for powerline
communications, Proc. ISPLC 2002.

[17]

B. Sklar, Digital Communications, Prentice-Hall


International, 2001, pp. 167-242.

[18]

M. Ghosh, Analysis of the effect of impulsive


noise on multicarrier and single carrier QAM
systems, IEEE Trans. Commun., vol. 44, no. 2,
pp. 145-147, Feb 1996.

[6]

L.T. Tang, P.L. So, E. Gunawan, Y.L. Guan, S.


Chen and T.T. Lie, Characterization and
modeling of in-building power line for highspeed data transmission, IEEE Trans. Power
Delivery, vol. 18, no. 1, pp. 69-77, Jan 2003.

[7]

M. Zimmer and K. Dostert, A multi-path signal


propagation model for the powerline channel in
the high frequency range, Proc. ISPLC 1999,
pp. 45-51, 1999.

[8]

M. Zimmer and K. Dostert, A multipath model


for powerline channel, IEEE Trans. Commun.,
vol. 50, no. 4, pp. 553-559, April 2002.

[9]

M.H.L. Chan and R.W. Donaldson, Amplitude,


width, and interarrival distribution for noise
impulses
on
intrabuilding
power
line
communication networks, IEEE Trans.
Electromagn. Compat., vol. 31, no. 3, pp. 320323, 1989.

[10]

J. Onunga and R.W. Donaldson, Distribution


line communications using CSMA access control
with priority acknowledgements, IEEE Trans.
Power Delivery, vol. 4, no. 2, pp. 878-886, 1989.

[11]

J. Onunga and R.W. Donaldson, A simple


packet retransmission strategy for throughput
and delay enhancement on power line
communication channels, IEEE Trans. Power
Delivery, vol. 8, no. 3, pp. 818-826, 1993.

Potrebbero piacerti anche