Sei sulla pagina 1di 18

Electrical Power and Energy Systems 95 (2018) 617–634

Contents lists available at ScienceDirect

Electrical Power and Energy Systems


journal homepage: www.elsevier.com/locate/ijepes

A channel model for power line communication using 4PSK technology for MARK
diagnosis: Some lessons learned

Madhav Mishraa,b, , Maarten van Rietc
a
SKF-University Technology Centre, Luleå University of Technology, 971 87 Luleå, Sweden
b
Division of Operation and Maintenance Engineering, Luleå University of Technology, 971 87 Luleå, Sweden
c
Alliander N.V. Postbus 50, 6920 AB Duiven, The Netherlands

A R T I C L E I N F O A B S T R A C T

Keywords: Modern smart grids and smart metering concepts are based on reliable digital communications. The absence of
Power line communication dedicated communications media, such as telephone lines or fibre optics within a power line network, can make
Medium voltage transmission challenging. Electrical power companies are interested in implementing an overall communicating
BER power line network. The power line communication (PLC) system uses the electric power distribution grid as a
PSK
data transmission medium. The data transmission problem resulted due to poorly developed Medium Voltage
Coupling
Network of PLC Channel Model and challenges in data transmission technology, so this hampers better per-
formance. This paper studies PLC over a medium voltage network with a goal of achieving greater bit rates and
communication that is more reliable over power lines. It presents a complete channel model of a PLC system and
evaluation of Bit Error Rate (BER) of Phase Shift Keying (PSK) when corrupted with noise. It calculates the
number of sections between two substations to determine signal loss. The PSK modulation scheme in simulation
is experimentally found to be more robust against such power line distortions as noise for point-to-point
transmission. The model and calculations use Matlab and QUCS.

1. Introduction failure, bad weather, and noise. It is a challenge for power companies to
control all these factors. Network management systems must provide
Modern smart grids and smart metering concepts require reliable sufficient information and control capabilities to operate the power
digital communications. The absence of dedicated communications system in a safe, secure, and economical way.
media, such as telephone lines or fibre optics within the power line The electric power industry needs to be transformed to meet these
distribution network, as well as transmission difficulties (e.g. time-fre- challenges and cope with the needs of a modern digital society.
quency selectivity, high signal attenuation, (self-)interference, im- Customers want to pay less but still demand high energy quality, re-
pulsive noise, etc.) in other media (e.g. power lines or wireless com- liability, and a wide choice of extra services. The power line commu-
munications) can challenge the stability and availability of the nication (PLC) system uses the electric power distribution grid as a data
communications system. A power line communication (PLC) system is a transmission medium. The main idea of PLC is to reduce cost while
low cost solution; it provides communication services through an ex- creating new telecommunications networks that are readily available
isting infrastructure and can be owned by the power suppliers. worldwide. These networks can be used to overcome the tele-
Electricity is produced and delivered to customers through the fol- communications “last mile” [2,3].
lowing steps; generation, transmission and distribution. Fig. 1 shows a PLC dates back to the mid-19th century [4] when ripple carrier
transportation network. signalling (RCS) was applied in medium (MV) and low voltage (LV)
Both the general public and industrial customers are increasingly networks to distribute the load and prevent extreme load peaks. The
dependent on electricity; for industry, the consequences of electricity applied carrier frequencies enabled information to flow over transfor-
failure can be especially devastating. This means the delivery of elec- mers on MV and LV networks with only minor attenuation. However,
tricity must be as reliable as possible. Enough electric power must be because of the large number of branches, MV and LV were poor media
generated and delivered to customers (public or industry) without [5]. MV and LV also had low data rates and supplied only unidirectional
disturbances or voltage losses, despite obstacles, including equipment data transmission from the power supply company to the consumers.


Corresponding author at: SKF-University Technology Centre, Luleå University of Technology, 971 87 Luleå, Sweden.
E-mail address: madhav.mishra@ltu.se (M. Mishra).

http://dx.doi.org/10.1016/j.ijepes.2017.09.020
Received 12 April 2017; Received in revised form 29 August 2017; Accepted 18 September 2017
0142-0615/ © 2017 Elsevier Ltd. All rights reserved.
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Fig. 1. Electric transportation network [1].

Amplitude shift keying (ASK) and frequency shift keying (FSK) specifies five bandwidths from 3 kHz to 148 kHz. Orthogonal frequency
methods were later used to transmit digital information with RCS. division multiplexing (OFDM) and 4PSK technology comply with the
Higher data transmission rates and decreased transmission power re- standard. Recently, OFDM has been used in high data rate modems
quirement resulted from the development of modulation methods and transposed to lower frequencies, particularly for smart metre reading.
the use of higher frequencies in the carrier signal. Bidirectional data OFDM is good to use in different carrier situations [1,13,14], but co-
transmission was the general form of transmission. loured noise can cause imprecise synchronization, reducing the modem
By the late 1980s and early 1990s, more complex error control performance [15]. Phase shift keying (PSK) technology is widely used
coding techniques and their subsequent implementation into low-cost in point-to-point transmissions and considered a better solution than
microcontrollers were proposed for the hardware of PLC modems [5]. others for this scenario but it is more expensive than multi-carrier
The Enermet Melko system appeared in 1984. This system uses phase technology. With OFDM, different carries can be selected for commu-
shift keying (PSK) modulation and a frequency band of 3025–4825 Hz nication, but with 4PSK only a single carrier is allowed.
for data transmission. Because of the low frequency band, the carrier This paper proposes a model of a power line communication (PLC)
signal can pass through distribution transformers. Melko is capable of system over a medium voltage (MV) network using 4PSK (phase shift
bidirectional data transmission; it also facilitates remote metre reading keying) technology. It focuses on finding the bit error rate (BER) in the
and load management. A local operation networks (LonWorks) platform transmission medium.
was created by Echelon in 1990 [6]. It is a standard-based control-
networking platform, robust, flexible, and expandable. Physical layer 2. Power line communication over medium voltage networks
(PHY) signalling can be implemented over such media as a twisted pair,
a power line, fibre optics, and radio frequency (RF). The basic architecture of electricity networks has not changed sig-
With recent developments in technology, the applied frequency nificantly in past decades, making it difficult to meet the increasing
bands have been upgraded from kilohertz to megahertz frequencies. demands of people who are highly reliant on electricity for day-to-day
More complicated mechanisms are being used to apply more advanced life. At some point, the vision of a shared energy and communications
modulation and error control coding methods in embedded systems. infrastructure will be realized, but the smart grid is only halfway there.
This is a direct result of the development of new processors, digital There is a need for a common infrastructure that will cost-effectively
signal processing, and algorithm techniques. Moreover, the advance- and efficiently deliver real-time data to meet the needs of a dynamic
ment of Internet has enabled new methods of information transmission energy market [8].
in all households. Power line communication (PLC) is a relatively modern electricity
Power demands are increasing, and electricity is the lifeblood of all network, with the capability of communicating data bits over an elec-
developed nations. The efficiency of power distribution matters for the tricity network. PLC can be divided into low voltage (LV), medium
profitability of all power companies. They need energy management voltage (MV), and high voltage (HV). LV-PLC has been using a smart
systems to manage global power consumption and improve energy ef- metering concept [2,3]. LV-PLC also includes a utility oriented low
ficiency. The issue of minimizing and eradicating any form of power frequency and low speed applications, like automatic metre reading
wastage is of prime concern. Because of the stiff competition, power (AMR), or smart metres, dynamic billing, and so on. The development
companies are looking for technologies that will perform well and re- of digital equipment and the standardization of efforts are making LV-
duce costs. PLC more lucrative for electrical utilities’ telecommunications services.
The solution is a smart power grid. A smart grid will take advantage MV-PLC and HV-PLC, historically oriented to tele-protection, are re-
of multiple kinds of communications technologies, ranging from fibre garded as reliable communication channels [2,4,8,15]. In addition, MV
optics, to wireless, to wire-line (PLCs, xDSL, coax, etc.) [7]. An element is low cost, causing many companies to adopt it.
common to all electrical loads from a grid perspective is the connection
between the provider (primary station) and the consumer (secondary
2.1. Coupling methods
station). It is necessary to develop low powered communication devices
gathering consumption data at the consumer end. One major objective
A PLC modem is able to communicate using a commercially avail-
of the PLC community is to attain a realistic power line channel model.
able indicative/capacitive coupling device, an electrical circuit con-
In practice, the performance of PLC modems fluctuates widely, de-
necting two other circuits, often with different characteristics. In what
pending on where they are deployed, making it difficult to estimate
follows, coupling network refers to the circuit connecting the commu-
(i.e., to model) data communication [8].
nications system to the power line. There are two kinds of coupling in
PLC has recently been developed for low voltage (LV) systems, and
medium voltage networks: inductive and capacitive.
the application of PLC to medium voltage (MV) power networks is of
Inductive coupling: Inductive coupling devices are usually easy to
increasing interest [9,10]. Network management optimization and the
construct, economical, and safe to use because they are not galvanically
monitoring of the system and operational services are the most im-
connected to the MV cable during or after installation. A great ad-
portant PLC applications for MV networks [11]. In Europe, available
vantage of inductive coupling is the quick and easy installation of the
frequency intervals for communication systems on LV and MV power
coupling device. A retractable magnetic coil is split into two halves and
networks are given in CENELEC EN 50065-1 [12]. The standard
is positioned around the MV cable or its shield without switching down

618
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

as an inductive coupler, because a great advantage of inductive cou-


pling is quick and easy installaon of the coupling device, and this is not
the case for capacitive coupling [16]. See Fig. 4b for a picture of a
capacitive coupler.
Table 1 compares inductive and capacitive coupling. We have al-
ready mentioned the advantage of installability for inductive coupling
but there are other differences as well. As Table 1 makes clear (see
Table 2),

2.1.1. Explanation of terms


Frequency Band: In Europe, the allowed bandwidth of medium
voltage networks can be from 9 kHz to 148.5 kHz, as specified by the
CENELEC standard [12].
Fig. 2. Inductive couplers in medium voltage networks [1].
Modulation: The technique of superimposing the message signal on
the carrier is known as modulation. In modulation, a property or
the MV-LV transformation and thus without interrupting the 50 Hz parameter of one signal (in this case the carrier) is varied in proportion
current flow in the sub-network. The high frequency sinusoid signals to the second signal (modulating signal). Types of modulation include
flowing through the windings cause a changing magnetic field, which, amplitude, frequency, frequency-shift, amplitude shift keying (AKS),
in turn, creates a sinusoid electrical field in the MV cable. The current among others. Modulation is performed at the transmitting end, as
flow closes over the capacitor to the earth, so the coil can be positioned given in Eq. (1).
in the three phases or in the shield connected to the earth [10,13].
Carrier, v (t ) = Input I × cos (2πfc t + θ) (1)
Inductive couplers are shown in Figs. 2–4.
Capacitive coupling: Capacitive couplers are connected in parallel where Input I = amplitude, fc = carrier frequency, and θ = phase
with the phase conductors. A coupling capacitor cannot be as reliable angle.
The process of converting analog signals back to digital bits is

Fig. 3. (a) and (b) inductive signal coupling:


practical case in a secondary station.

Fig. 4. Inductive coupling element (a) and capacity coupling element


(b).

619
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Table 1 Signal to Noise Ratio: Signal-to-noise ratio (SNR or S/N) is defined as


Coupling methods [17,18]. the power ratio between a signal (meaningful information) and the
background noise (unwanted signal) and is expressed as
Capacitive Inductive
S Signal Power ⎞
Cost +++ ++ = 10. log10 ⎛
Effect on system reliability ++ +++ N ⎝ Noise Power ⎠ (3)
Installability + +++
Galvanic isolation from phase voltage ++ +++
Note that the signal-to-noise ratio (S/N) shown in Eq. (3) is usually
Over-voltage protection of modem ++ ++ expressed in decibels (dB), derived as shown in Eq. (4):
Attenuation without output filter of inverter + +++
V
Attenuation with output filter of inverter +++ ++ Power (P ) = Voltage (V ) × Current (I ) = V × = V × V /1 Ω
R (4)

Note that in real systems, line impedance may be 5–25 Ω, Therefore,


Table 2 SignalPower and NoisePower are not correct in real systems. However,
Comparison of different BER [19,25].
(
SignalPower
)
their ratio = SNR = NoisePower is correct in any line impedance
BPSK QPSK 16-PSK 16-QAM 64-QAM Signal Power ⎞
because ⎛
SNR BER BER BER BER BER ⎝ Noise Power ⎠
8 0.0001 0.0076 0.245 0.012 0.010 Signal Voltage × Signal Voltage / lineimpedance
22 – – 0.011 0 0
=
Noise Voltage × Noise Voltage/ lineimpedance
= (Signal Voltage )2 /(Noise Voltage )2 (5)

known as demodulation (or detection). This reverse operation is per- Therefore, there is no direct relation between SNR and line impedance.
formed at the receiving end. Bandwidth capacity: To calculate bandwidth capacity, in this work,
PSK modulation technique: In phase shift keying (PSK) modulation we use Shannon’s theorem. It gives an upper bound to the capacity of a
data transmission (the method used here), the phase of the carrier link in bits per second (bps) as a function of the available bandwidth
signal is switched between two or more values in response to the and the signal-to-noise ratio of the link [21]. The theorem can be stated
baseband digital data. The phase of the sinusoid is altered according to as channel capacity (Cch ):
the information sequence. If only two phases are used, a binary 1 can be S
transmitted using, for instance, cos(2πfc t ) , and a binary 0 would be Cch = B·log 2 ⎛1 + ⎞
⎝ N⎠ (6)
carried by, see Eq. (2).
where Cch is the achievable channel capacity, B is the bandwidth of the
cos(2πfc t + π ) = −cos(2πfc t ) (2) line, S is the average signal power, and N is the average noise power.
Following practical experience in a work environment, we select the
The demodulator must be able to determine the phase of the received values for the parameters as: B = 100 kHz, S/N = 7. Substituting the
sinusoid with respect to some reference phase. values into Eq. (7), we get the following:
PSK generation: In this experiment, PSK generation, as shown in
Fig. 5, was modelled in Matlab, as explained in the Appendices. PSK is S
Cch = B·log 2 ⎛1 + ⎞ = 300 kbps.
among the most popular carrier schemes. ⎝ N⎠ (7)
Channel: The channel is a physical medium, such as wired, wireless Note that in our experiment, the theoretical capacity of the channel is
etc. It is important to know the characteristics of the channel, such as obtained as 300 kbps but, in practice, only 8–128 kbps throughput is
the attenuation and the noise level, because these parameters directly achievable. The reasons behind of this because of there are several
affect the performance of the communication system. factors can be influenced, such as noise, uncertainty of measurement,
Noise and disturbance: A common feature of a power line commu- uncertainty of model parameters, EMC effect, etc., However, in the
nication channel is the presence of noise. In conventional radio com- theoretical we are not always able to cover these kinds of effects.
munication systems, the dominant source of receiver noise is thermal
noise. The power line noise at an outlet is the sum of noise waveforms
3. Modelling and method
produced and emitted to the line from appliances connected to the
power line network. Noise and disturbances on the electrical power
3.1. Transmission line model
network can be classified as superimposed disturbances (e.g. outages,
frequency variations, harmonic distortions, etc.) or waveshape dis-
For this work, we calculated the cable length of each section be-
turbances (e.g. transient disturbances, persistent oscillations) [19,20].
tween two substations (K4 and K5), depending on the carrier frequency.
cos_wt This resulted in the use of eight sections between the substations. We
did this to correct any form of signal loss and to more easily detect any
Input_I defects in the cable between the sections. It was also easier to model the
X transmission line section.
The transmission line plays a vital role in the quality of the signal
being transferred and received. A good model of the transmission line
PSK Output
Out
+ helps to approximate an acceptable signal to noise ratio (SNR) and to
simulate the effect of noise. The transmission line model shown in Fig. 6
was developed for this work (see Fig. 7).
Input_Q The parameters used in the model as programmed and simulated in
X QUCS are given below:
Length of section = d, Velocity of waves = ν, Characteristic im-
sin_wt pedance unknown for cable (Zkar ) , given Zkar = 22 Ω, Quality factor of
impedance = QL , Quality factor of capacitance = Qc ,
Fig. 5. PSK Generation.
Assumption, QL = Qc = 1000

620
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

d
L = Zkar ·
v (14)
This means d ≪ λ; experience: d ≈ λ/8; total length (dtot)
= number of sections (Ns) ⋅d Ns = dtot/d = 2000/2000⋅8 = 8.
From the above calculations, we know the length of a section ≤1/
8 ⋅ λ, so eight sections are required between the two substations K4 and
K5. Fig. 8 represents a section cable length denoted by SUB (1)(8). SUB
is also used in QUCS simulation to denote a cable section. A distributed
parameter medium voltage cable model is considered together with the
power transformers, signal-coupling networks for the receiving and
Fig. 6. Transmission line Pi-model.
transmitting line coupling.
The simulation equations used for Fig. 8 are given below:
Executing code in QUCS: Case 1
1 ωL Q
C= · d , Rs = , RP = c where ω = 2πfc , ω
Zkar ·v ωC ωC
= 100 kHz·2·π = 0.62 Mrad/s
where Rs is the series resistance, and Rp is parallel resistance.
1 ωL1 Q
C1 = length· · v , Rs = , RP = c , L1 = length·Zkar / v
Zkar QL ωC1
Fig. 7. Model of the cable per section. Executing code in QUSC: Case 2

Zkar ω = 0.62 Mrad/s, length = 2000 m, QL = 1000, Qc = 1000, Zkar


L=
v (8) = 22Ω, v = 0.25 G

1
C= ·d 3.2. Complete model of PLC
Zkar ·v (9)

ωL Fig. 9 shows a digital communication system. In this system, a


Rs = power line has been used as a communication channel. The transmitter
ωC (10)
is shown on the left of the diagram and the receiver to the right.
Qc A coupling circuit shown in Fig. 10 functions as a connection be-
RP =
ωC (11) tween the communication system and the power line. It has a dual
purpose. First, it stops the 50 Hz signal, which can damage the equip-
where = 2πfc ; distance between the MV substations K4 and K5 is ap- ment, from entering the equipment. Second, it verifies that the main
proximately = 2 km = 2000 m; frequency range fc = 0–100 kHz =
part of the received/transmitted signal lies within the frequency band
100⋅1000 = 1 ⋅ 105 Hz; and velocity of wave (v) = 2⋅108 m/s.
used for communication. Increasing the dynamic range of the receiver,
1 2·108 it ensures that no interfering signal on the channel is introduced by the
(λ ) = v· = = 2000 m transmitter. A general simulation model is shown in Fig. 10.
fc 1·105 (12)
With the calculations shown above, we identified the need for eight
However, (13) is only valid if d ≪ λ. Therefore, as shown in Eq. (14), sections or substations lying between the two stations labelled K4 and
K5. These eight sections form the complete channel model of power line
d
C = 1/2· communication. To get realistic and practical output from the model
v·Zkar (13) itself, we introduced noise in the input to simulate real noise. The

Fig. 8. Model of SUB.

Fig. 9. General power line communication


system.

621
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Fig. 10. General power line communication simulation model.

Fig. 11. Complete PLC channel model.

L1 L2

L1

L2

Fig. 13. Case of coupling.

Fig. 12. Result of PLC model using QUCs.


I
or H = ,
2πr (17)
experiment with noise was carried out in a different model using
Matlab; the code of this model as shown in Fig. 11 is included in Bb = μH (18)
Appendix.
Φ = Nn × Bb × Aa , (Aa = area) (19)
For the experiment, we used a carrier frequency of 100 kHz, which
results in 1.1 μV. The result is shown in Fig. 12. Aa
Φ = μr ·Nn· ·I
2πr (20)
3.3. Inductive coupling model Φ A
M= = μ·Nn· a
I 2πr (21)
A coupler is a device that helps a communication signal to be su-
perimposed onto and/or extracted from a power waveform. It also stops Aa
L1 = Nw · ·μ
the power waveform from entering the modem. Inductive couplers have 2πr (22)
been used in the medium voltage cables of transmitter and receiver L2 = length·Zkar / v (23)
side. These couplers have only two physical connections to the modem
and require a power cable (live or neutral) to be positioned through the M = k L1 L2 < 1, (24)
centre of some magnetic core, completing the magnetic circuit. An
example of an inductance model that can be used in power line com- where Nn = number of turns of the coil; U1 and U2 are the voltage of
munication is shown in Fig. 13. the inductive coupling calculations.
In Fig. 13, L1 is the length of tightly turned coil on the magnetic
kernel, and L2 is the length of remaining coil after being turned on the 3.4. A communication system and its model
magnetic kernel.
The coupling case can be expressed as: Digital communications techniques are at the core of any commu-
nication systems, including the power line communication system. A
U1 = jωL1 I1 + JωMI1 (15) model of a communication system is shown in Fig. 14. The objective of
the communication system is to communicate digital information (a
U2 = JωMI1 + jωL2 I2 (16)
sequence of binary information digits) over a noisy channel at as high a
M = kc L1 L2 , where, kc = 0.1 bit rate as possible.

622
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Tx_mod +
tx_mod rx_dem

Input ModulaƟon PL-Channel DemodulaƟon Output

Fig. 14. Communications system model.

Modelling Binary Data: Binary data are defined as Rww (t1 t2) = E {w (t1) w (t1 t2)} = (N0/2) δ (t1−t2) (28)
tx data = round (rand (1,nbaud∗2)) . Rand is the random generator
function in Matlab and refers to uniformly distributed pseudorandom where the standard deviation is σ and Im is the identity matrix.
numbers. The corresponding code for this calculation is attached in Assuming that the additive noise n is the Gaussian probability dis-
Appendix. tribution function P(x) [22] then
Modelling Noise: The signal that contains equal power within a fixed
bandwidth at any centre frequency is called white noise. A random 1 −(x −μ w )2
P (x ) = e
vector w is a white random vector if and only if its mean vector and 2πσ 2 2σ 2 (29)
autocorrelation matrix are the following:
where μ w = 0 and σ 2 = N0/2 . A model of white noise as programmed in
μ w = E {w} = 0 (25) Matlab is attached in Appendix.
Modelling Filters: In modulation systems, low pass filters are used for
Rww = E {wwT } = σIm (26) FIR type filters; an IIR type filter like butter-worth is not used, because
That is, zero is a random vector, and its autocorrelation matrix is a these filters are not linear-phase. FIR type filters may be designed as
multiple of the identity matrix. When the autocorrelation matrix is a linear in all frequencies. A TX FIR low-pass-filter block diagram is given
multiple of the identity, we say that it has spherical correlation. A in Fig. 15.
continuous time random process w (t) where tn ∈ R is a white noise We simulated the filter shown in Fig. 15 in Matlab. To give a sense
process if and only if its mean function and autocorrelation function of how the filter works, we provide the following snippet of code and
satisfy the following: show the result of the TX FIR filter input and output with FIR tap as
achieved from the Matlab simulation in Fig. 16.
μ w (tn ) = E {w (tn )} = 0 (27)

Fig. 15. TX FIR low-pass-filter.


I_Q_data
FIR_reg (64 complex)
(1 complex)

FIR_tap
(64 complex)

FIR_out (1 complex)

Fig. 16. TX FIR filter input output and FIR_tap.

623
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

50 100 150 200 250

-20

-40

-60
0 5 10 15 20 25

-1

101.5 102 102.5 103

-20

-40

-60
20 25 30 35 40 45

Fig. 17. Spectrum of binary data and analog signal using 4PSK technique.

IQ data is a complex number , where I = real (I Qdata), Q


= imag (I Qdata)

Before modulation I Qdata are passed through the FIR filter. FIR_out is
filtered output data, which is also a complex number. Finally, analo-
g_out = real(FIR_out).cos(wt) + imag (FIR_out). sin(wt). 1 baud is time
unit for 1 constellation transmit, 1 spck is time unit 1 sample for DA-
converter. Suppose in this system, baud rate = 1200 Hz,
spck = 9600 Hz. I_Q_encode generate 1(one) I_Q_data in 1 baud. How-
ever, the FIR filter output is 8 samples in 1 baud. The TX FIR filter may
Fig. 18. Example of a model of modulation.
be up sample and low-pass filtering. The whole function code is at-
tached in Appendix.
An FIR filter is necessary to perform PSK modulation and demo- shown below the binary data and the analog signal respectively. The
dulation on data. As filters all reset to zero before beginning the in- spectrum is measured in dBs.
tegration for the next symbol interval, this permits the use of PSK u 0 = K ·u1·u2 (30)
modulation and demodulation. For the above model, a 64-tap Full-cos-
roll-off-Low-pass-filter is employed. Further references to the filter u1 = A·cos(2πf1 t ) (31)
spectrum used for this work appear in Appendix.
u2 = A·cos(2πf2 t ) (32)

3.5. Modelling modulator with multiplier


3.6. Estimation of bit error rate for 4PSK
The multiplier has its main applications in modulators and demo-
dulators. This means that the output of the multiplier can be found near 4PSK modulation consists of two binary PSKs on in-phase and
the carrier frequency and the information is in the baseband. The quadrature components of the signal. Bit error rate (BER) plays an
modulation we use in what follows is called double side modulation important role in efficiency of the model using 4PSK.
because the information can be found twice [23]. Using the process of 1 −(y− Es /2 )2
modulation, the binary data are converted into an analog signal, as P (x /11) = e
πN0 N0 (33)
shown in Fig. 17 (see Fig. 18). The corresponding spectrum of the di-
gital bits and the analog signal achieved using the 4PSK technique is The scaling factor of Es /2 ) is for normalizing the average energy of the

624
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

tx_data Fig. 19. BER calculation at TX and RX.


rx_data
Inputdata BER by bit Outputdata
10000baud=20000bits compare 10000baud=20000bits

PSK4_TXRX
TX_mod RX_dem
analog_out analog_in
TransmiƩer + Receiver

SNR
White noise
ngain*randn

transmitted symbols to 1, assuming all constellation points are equal. Process of simulation:
The conditional probability distribution function (PDF) of y is 11. The A 4-PSK system 1 baud is 1/1200 s. TX may generate 1 constellation
theoretical limit is calculated by the following constellation: signal- by 2 bit input data.
power = 1 ∗ 1 = 1, Noise-power = (1/1.414) ∗ (1/1.414) = 1/2. In this Note that in this experiment, 1 baud = 2 bits. tx_data = [1 0 1 1 1 0
condition, raff BER estimation is (S + N)/N = 3 → = 5 dB → 0..]. txd for TX_mod is 2 bits data = [1 0].
error = 1/2(error distribute is about 50%). Training sequence-1: Normal modem systems employ a training
A model for comparing BER based on the effect of SNR on the input sequence (Preamble signal) before data transmission. This system also
data appears in Fig. 19. In our experiment data, a transfer speed of employs a 17-baud training sequence. This sequence is equivalent to
64 kbps is considered, with acceptable white noise of 0.04 amplitude 32 bits “1” and 2 bits “0” data transmission before real data.
and SNR of 7. In Fig. 19, the red denotes variable names and green Training sequence-2: Purpose of this training sequence is for
denotes function names. For this experiment, the input data are Receiver (Rx) convenience. In RX, FIR_out is 8 complex, but required
20,000 bits, which equates to 10,000 baud. This has been calculated as data is 1 complex for I_Q_decoder. In the initial stage, RX selects the best
shown below: sample from FIR_out. RX decides the best sample from the 8 sample of
FIR_out in the first 16 baud sequence. After the pi-phase detects some
Baud_rate = 1200 Hz period (possibly 10), RX select the best point from 8 FIR_out, usually
Bit_rate =2400bps = Baud_rate∗2 bits(4-pase) max amplitude FIR_out.
Carrier_frequency = 2400 Hz
Sampling_clock = 9600 Hz 3.7. Modelling transmitter and receiver
Simulation length = number of baud 10,000 baud = 20,000 bit
Fig. 20 shows the workings of the modulator in the transmitter. As
BER is also known in the terms of the probability of error (POE) and we are using the 4PSK modulation technique, all the incoming bits are
can be expressed as: multiplied by 2 bits and then sent for the regular process of converting
1 to analog signal.
POE = (1−erf ) Eb/ No Using the transmitter, the bits are changed into relevant analog
2 (34)
signals after performing modulation with 250 samples. The results are
where erf is the function of error, Eb is energy in one bit and No is the shown in Fig. 21.
noise power spectral density. As per the common sequence in data communication, before data
For this simulation, we did not consider a large bit rate and high bits are transformed to analog signal, they go through an encoding
noise with high frequency because the program is very big and resource process. For this process, we used an I_Q_Encoder. The block diagram of
intensive. A model of the experiment with above parameters is given in this type of encoder is shown in Fig. 22.
Fig. 19, and the code for the simulation is attached in Appendix. Calculation of TX analog_out spectrum:

Fig. 20. Transmitter block.


TX_mod

cos_wt (8 real)

Real(FIR_out)
X
(8 real)
txd TX FIR
2bit I-Q I_Q_data Low-
+ DA
Encoder (1 complex) Pass
Filter PSK Output
img(FIR_out)
X
(8 real)
sin_wt (8 real)

625
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

2 The analog_out is ready to be transmitted using the PLC. The


spectrum of the respective analog out and the analog signal in the re-
1 ceiver after adding noise in constellation.
The receiver receives analog signals via the PLC, which has noise; it
0
transforms the analog signals back to digital bits using the demodula-
-1 tion process. A receiver block diagram is shown in Fig. 23. The model
implementation of the receiver was done in Matlab and the related code
-2 of this implementation is attached in Appendix.
0 50 100 150 200 250 At this point in the process, an I_Q_Decoder decodes the digital data,
Fig. 21. Result of modulation signal. which matches the input data given in the transmitter. A block diagram
of an I_Q_Decoder is given in Fig. 24.
Our models use parameters that are close to a real-life environment;
In this experiment, the analog_out is calculated by using the formula
therefore, the demodulation technique, filter, noise, coupling and bit
given below:
error rate give a convincing output.

analog_out = real(FIR_out). ∗ cos_wt + imag(FIR_out). ∗ sin_wt


4. Results and discussion
where w, carrier frequency, is 2400 Hz at 9600 Hz spck, as proved
by Shannon’s sampling theorem; cos_wt = [1 0 −1 0 1 0 −1 0]; and We have proposed a model for complete power line communication;
sin_wt = [0 1 0 −1 0 1 0 −1]. the model is shown in Fig. 25. In our experiment, we introduced noise

Fig. 22. I_Q_Encoder block diagram.


I_Q_encoder & diīerenƟal encoder

I_Q_data(1complex)
IniƟal value=1
I_Q_data

txd txd phc value


(2 bits) 00 0 1
01 1 j Value
X
11 2 -1 (1 complex)
10 3 -j *complex mulƟply
phc = phase_change
value=exp(phase_change*1j*pi/2)

RX_dem Fig. 23. Receiver block.

cos_wt (8 real)

real(FIR_in) real(FIR_out)
X

RX FIR rxd(2 bits)


I-Q
AD Low-Pass
Decoder
Filter

imag(FIR_in) imag(FIR_out)
X

sin_wt (8 real)

Fig. 24. Decoder.

I_Q_data (1 complex)
IniƟal value = 1
I_Q_data
X Y rxd rxd
phd=[x,y] X>Y&X>-Y 00 (2 bits)
/ X<Y&X>-Y 01
1 complex X<Y&X<-Y 11
X>Y&X<-Y 10
*complex divide

626
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Fig. 25. Complete PLC channel model for signal coupling.

4 in the transmission medium, thus altering the output. We used a 4PSK


modulation technique for data transmission; it is considered one of the
best techniques for single carrier communication systems. The focal
3
point of the experiment was calculating the bit error rate (BER) in a
controlled environment, as close to real life as possible. The outcome of
2 the experiment was a 1.1 microvolt current in the inductive coupling
signal at 100 kHz carrier frequency.
To simulate a real-life scenario where noise is pervasive, we only
1 considered white noise. White noise of up to 0.04 amplitude was in-
jected into the transmission line with noise injectors to see the effect on
noise in the BER. Up to 0.04 amplitude of white noise, the data sent
0
from the transmitter was received without any loss. The results are
shown in Fig. 26.
-1 We used a single carrier 4PSK modulation and demodulation
system. To simulate noise injection, our BER calculation was composed
of a 16-pattern noise iteration, as shown in Fig. 26, in the form of RX
-2 constellations. Initially, only low noise was introduced; after each
iteration, the noise increased until it reached the 16th iteration. Fig. 26.
depicts the noise at the last iteration.
-3 In a theoretically ideal situation where no noise exists, there can be
no data loss; this is explained in Appendix. In fact, noise always exists in
-4 the communication channel. Therefore, our experiment includes a
-4 -2 0 2 4 reasonable amount of noise. As soon as noise is encountered in the
communication channel, the BER must be calculated to access the im-
Fig. 26. After adding white noise constellation in 4PSK.
pact of noise in the communication process and deal with it accord-
ingly.
Fig. 27 shows the relation between SNR and BER. The X-axis

BER(10**x) vs SNR(db) with White-Noise


BER(10**x) vs SNR(db) with White-Noise
0
-1.5
-0.5

-2 -1
X: it Error Rate ;ERͿ с ;10XͿ

X: it Error Rate ;ERͿ с ;10XͿ

-1.5
-2.5
-2

-3 -2.5

-3
-3.5
-3.5

-4 -4
4 5 6 7 8 9 10 11 12 13 2 4 6 8 10 12 14 16
^iŐnaů to Eoise RaƟo͕ ^ER ;ĚͿ ^iŐnĂůtoEoisĞZĂƟo͕^EZ;ĚͿ

(a) (b)
Fig. 27. Bit error rate (BER) simulation when noise is 0.04 (a) and 0.07 (b).

627
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Table 3 for single carrier power line communication systems. PSK can increase
SNR and BER values. the system’s robustness. Moreover, it is practically implementable and
gives good results when the SNR is greater than 7 dB at 100 kHz. With
SNR BER
the use of FIR filters [23,24] and an appropriate modulation technique,
26 dB 20/1 the system is able to attain acceptable output in the receiver even with a
20 dB 10/1 low SNR. Using FIR filters also helps to minimize the BER, giving better
14 dB 5/1
results.
10 dB 3/1
6 dB 2/1 For this work, we calculated the cable length of each section be-
tween two substations (K4 and K5), depending on the carrier frequency.
This resulted in the use of eight sections between the substations. We
Table 4 did this to correct any form of signal loss and to more easily detect any
Y-axis represent BER X. X means BER = 10^X. defects in the cable between the sections. It was also easier to model the
transmission line section. Work done by Komboj et al. [25] has com-
X = −1 10^(−1) = 1/10
parable results, but we built a complete model of PLC over medium
X = −2 10^(−2) = 1/100 voltage using 4PSK technique, while the previous experiment was
X = −3 10^(−3) = 1/1000 carried out for coupling, not the complete system. We have advanced
X = −4 10^(−4) = 1/10,000
the research by showing that for single carrier communication, 4PSK is
X = −5 10^(−5) = 1/100,000
X = −6 10^(−6) = 1/1,000,000 a viable technology. The use of power line communication eliminates
the need to install different cables for power and data. Electric power
and data transmission speeds of 8–128 kb/s can be achieved with PLC,
represents SNR and the Y-axis is BER. When the noise is less, SNR has a making this a more practical and feasible solution than other technol-
greater value, and when the noise is greater, SNR has a smaller value; ogies like GPRS (max speed 15 kb/s and expensive). Finally, the use of
this ratio is expressed in dB. Tables 3 and 4 show BER vs SNR. inductive coupling saves costs in the installation of inner stations by
In our results, BER data values are based on 10,000 reducing the cost of the magnetic kernel. This, in turn, reduces the need
baud = 20,000 bit, so when X = under -4.3 is equal to BER = 0, X may for human resources and simplifies the management of the power line
be infinite. When the SNR is 7 db, the BER is recorded at -2.2 (10∗∗x). communication system.
This is an acceptable result using 4PSK and adding white noise and does Based on our results, we have the following recommendations for
not cause any data loss. The code of this simulation is attached in future study. We focused on narrow band; all the power substations,
Appendix. The same experiment with noise of 0.07 amplitude gives regardless of distance, use narrow band. If broadband can be used for
N = 0.07. Therefore, with more noise, the signal to noise ratio de- substations closer than 400 m, this would enable faster data transmis-
creases; this is bad for communication, as it will result in more data sion and improve the average data transfer speed of the whole network.
loss. Other results of this experiment with noise of 0.07 amplitude are The maximum data rate is 128 kb/s; it can be improved up to at least
attached in appendix. 256 kb/s with the use of 16 PSK in future. Calculations can also be
We modelled the complete channel model of power line commu- considered for longer cable lengths, for example, cable fault detection
nication over medium voltage. The result of the simulation was 1.1 at 10 kV for an MV network.
microvolt. In this simulation, power consumption is lower; this will
save money for the company. Acknowledgements

5. Conclusion The authors would like to acknowledge Dutch Power Company


Allinader NV-Liandon BV for their financial support and for providing
Based on our results, we can conclude 4PSK is a suitable technology access to the measurement resources and locations.

628
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Appendix A

MATLAB CODE
BER calculation in presence of Noise

629
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

630
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

631
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

632
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

Appendix B. Supplementary material

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijepes.2017.09.020.

References protection to prevent dg islanding: an investigating procedure. In: Power tech


conference proceedings, 2003 IEEE Bologna, vol. 3. IEEE; 2003, June (7pp).
[12] EN 50065-1:1991—Signaling on Low-Voltage Electrical Installations in the
[1] Fliege NJ. IP over Medium Voltage Networks New Methods, internal documents, Frequency Range 3 to 148.5 kHz—Part 1: General Requirements, Frequency Bands
new applications, Arnhem; 2010. and Electromagnetic Disturbances Amendment A1:1992 to EN 50065–1:1991;
[2] Halid H, Abdelfatten H, Ralf Lehnert et al. Broadband power line communication Amendment A2:1995 to EN 50065–1:1991; Amendment A3:1996 to EN
network design. West Sussex: John Wiley and Sons Ltd; 2004 (0-470-85741-2). 50065–1:1991.
[3] Nandi S, Thota S, Nag A, Divyasukhananda S, Goswami P, Aravindakshan A, et al. [13] Trachanas, I. (2009). Robust Multiuser OFDM Power Line Communications Over
Computing for rural empowerment: enabled by last-mile telecommunications. IEEE Medium Voltage Networks: Robustes Mehrbenutzer-OFDM-System Für Die
Commun Mag 2016;54(6):102–9. Datenübertragung in Mittelspannungsnetzen. Sierke.
[4] Ferreira HC, Grové HM, Hooijen O, Han Vinck AJ. Power line communication [14] Trachanas I, Gaida M. Adaptive BER enhancement using frequency selective power
theory and application for narrowband and broadband communication over power distribution in DMT systems.
lines. West Sussex: John Wiley and Sons Ltd; 2010 (978-0-470-74030-9). [15] Aquilue R, Ribo M, Regue JR, Pijoan JL, Sanchez G. Scattering parameters-based
[5] Konate C, Kosonen A, Ahola J, Machmoum M, Diouris JF. Power line communica- channel characterization and modeling for underground medium-voltage power-
tion in motor cables of inverter-fed electric drives. IEEE Trans Power Delivery line communications. IEEE Trans Power Delivery 2009;24(3):1122–31.
2010;25(1):125–31. [16] Wade ER, Asada HH. Design of a broadcasting modem for a DC PLC scheme. IEEE/
[6] Echelon. Echelon Corporation, embedded control networking technology [Online] ASME Trans Mechatron 2006;11(5):533–40.
[Cited: April 1, 2017] < http://www.echelon.com/ > . [17] Kosonen A, Ahola J. Comparison of signal coupling methods for power line com-
[7] Galli S, Logvinov O. Recent developments in the standardization of power line munication between a motor and an inverter. IET Electric Power Appl
communications within the IEEE. IEEE Commun Mag 2008;46(7). 2010;4(6):431–40.
[8] Sharma K, Saini LM. Power-line communications for smart grid: Progress, chal- [18] da Silva Costa LG, de Queiroz ACM, Adebisi B, da Costa VLR, Ribeiro MV. Coupling
lenges, opportunities and status. Renew Sustain Energy Rev 2017;67:704–51. for power line communications: a survey. J Commun Inform Syst 2017;32(1).
[9] Dostert, KM. Power lines as high speed data transmission channels-modelling the [19] Sharma V, Singh G. On BER assessment of conventional-and wavelet-OFDM over
physical limits. In: Spread Spectrum Techniques and Applications, 1998. AWGN channel. Optik-Int J or Light Electron Opt 2014;125(20):6071–3.
Proceedings., 1998 IEEE 5th International Symposium on, vol. 2. IEEE, 1998, [20] Babu AS, Rao DKS. Evaluation of BER for AWGN, Rayleigh and Rician fading
September, p. 585–9. channels under various modulation schemes. Int J Comput Appl 2011;26(9).
[10] Hardy ME, Ardalan S, O'Neal JB, Gale LJ, Shuey KC. A model for communication [21] Price E, Woodruff DP. Applications of the Shannon-Hartley theorem to data streams
signal propagation on three phase power distribution lines. IEEE Trans Power Del and sparse recovery. In: 2012 IEEE international symposium on information theory
1991;6(3):966–72. proceedings (ISIT). IEEE; 2012, pp. 2446–50, July.
[11] Benato R, Caldon R, Cesena F. Application of distribution line carrier-based

633
M. Mishra, M. van Riet Electrical Power and Energy Systems 95 (2018) 617–634

[22] Ribeiro MI. Gaussian probability density functions: properties and error char- power-line communication systems noise levels. IEEE Trans Power Delivery
acterization. Lisboa, Portugal: Institute for Systems and Robotics; 2004. 2013;28(4):2004–13.
[23] Zhong L, Lichun L, Huiqi L. Application research on sparse fast fourier transform [25] Kamboj R, Verma S, Malik S. Comparison of ber and number of errors with different
algorithm in white gaussian noise. Proc Comput Sci 2017;107:802–7. modulation techniques in mimo-ofdm wireless communication system. Int J Adv
[24] Milioudis AN, Syranidis KN, Andreou GT, Labridis DP. Modeling of medium-voltage Eng Technol 2013;6(3):1169.

634

Potrebbero piacerti anche