May

Must reads
Internet of Things in the Grid
Intelligence from the Data Center to the Edge
Sponsored by
 Must reads
Internet of Things in the Grid
27 Reducing Cost and Complexity with Industrial
System Consolidation
By Intel

Intelligence from the Data Center to the Edge

INSIDE 39 Narrowband Powerline Communication

Applications and Challenges Part I
By Bogdan Baraboi, Ariane Controls Inc

3 Product How-to: Making Sense of electrical power

smart grid data
By Steve Taranovich, EDN
48 Narrowband Powerline Communication
Applications and Challenges Part II
By Bogdan Baraboi, Ariane Controls Inc

9 Smart meters can be fully customizable

By Ludek Holoubek, STMicroelectronics
54 An overview of ZigBees Smart Energy Profile
2.0 standard
By Srinath Balaraman and Anil Khanna, Mentor Graphics

19 The balance of power:

Developing a smart meter
By Khurram Waheed, Freescale Semiconductor
59 Closing the Industrial Communications GAP
With a Secure, Virtualized Solution
By Intel

2 Sponsored by
Product How-to:
Making Sense of electrical
power SMART GRID
data
By Steve Taranovich, EDN

S
ince I lived many years on Long Island,
NY, I remember in the 80s, when I was
invited to visit the Long Island Lighting
Company (Lilco) Hicksville central control center
for all of Long Island. There were multiple walls
of huge color displays with discrete lights every-
where to signal operators of a fault or problem
at one of the many, many substations. To me
that was information overload similar to an air
traffic controller had at that time. That was be-
fore the Smart Grid appeared on the scene.
In their older system, SMUD used a paper wall
map that monitoring crews could access via a
ladder. File cabinets full of blueprints and assorted
documents rounded out the monitoring system.
As we progress in the 21st century, the Smart
Grid is becoming a wonderful asset to power
companies as well as to power users to allow
levels of power efficiency through smart systems.
However, that same Smart Grid presents orders of

3 Sponsored by
magnitude larger volumes of data with infor-
mation coming into the utility control center
in virtually real time. The human brain can-
not possibly process this fast enough.
Figure 1a: Situation intelligence overview (Courtesy of Space-Time Insight)
4 Sponsored by
An organization like SMUD now has to
deal with these massive volumes of data
that present new challenges to control
center monitoring personnel. One ex-
ample is data such as widely distributed
generation sources that present a new
challenge to monitor and analyze each
one of these sources. The data was in
a legacy data format not conducive to
real-time analysis. With the Smart Grid,
changes in some part of the grid, even
in a remote area, could affect assets one
might not expect intuitively. Furthermore,
adding new assets to the system pose
another tough challenge to integrate into
the control and monitoring system.
Enter Space-Time Insights software
solution to help with SMUDs new Situ-
ational Awareness and Visualization In-
telligence (SAVI) initiative. Now SMUD
would be able to analyze and correlate
huge volumes of data from the many field
resources in their region. SMUDs efforts
implementing a Wide Area Situational
Awareness1 (WASA) system allowed them
to now handle microsecond measure-
ments recorded by synchrophasors (de-
Figure 1b: Situational Intelligence architecture (Courtesy of Space-Time Insight) Figure 2: A modern control center in Folsom, California
21st century high-tech control centers (Courtesy of California ISO) (Courtesy of Space-Time Insight)

vices used in the measurement of quality
and state of their power system). The
highly developed and intuitive software
by Space-Time Insight enables control
center operators to easily and quickly
access real-time information they need
to make quick decisions to keep the grid
stable and continuously available to their
customers.
Situational Intelligence (SI)
Steve Ehrlich, SVP, Marketing of Space-
Time Insight, wrote an insightful tutorial on
SI entitled Situational Intelligence melds
IT, OT and XT to improve decision mak-
ing. See Figure 1a and 1b.
High-tech visual displays in Folsom,
California are ubiquitous in 21st century
control centers. (See Figure 2)
These types of video display systems
allow utility control center operators to
process large volumes of information at
a glance with smart visual displays en-
hanced by software:
Advanced features include enhanced
visualization capabilities, high-tech work
consoles, a modern computerized en-
ergy management system and synchro-

5 Sponsored by
 phasor technology that takes the pulse serve a crucial need as the ISO in- ticipate consumer demand on the grid.
of the power grid in tens of milliseconds tegrates thousands of megawatts of
rather than using the former industry green power onto the grid. The en- The WASA framework
standard of every four seconds. hanced visibility improves wind and The WASA framework is composed of
High-tech tools focus on the fast pace solar performance forecasting as well two phases as can be seen in Figure 3.
of power plant start-ups, voltage stabil- as advance weather prediction to an- First is the setup and commissioning that
ity, renewable forecasting, congestion
management and reliability. Computer
systems provide this information with
pictures rather than thousands of indi-
vidual data points.
The Energy Management System (EMS)
is the most advanced in the industry,
providing the latest application software
on the newest server hardware. This
increases performance, capacity and
automation functions.
Ten visualization screens display infor-
mation from Google Earth and other
applications spanning across an 80
wide x 6.5 high video wall at the Fol-
Figure 3: WASA framework is based upon context awareness and hybrid perspective
som headquarters. These screens (Image courtesy of Reference 1)

6 Sponsored by
Figure 4: A weather monitoring system display shows local conditions such as storms or Figure 5: Smart meters enable the utility to meter status and conditions through Space-
wind that can affect power utility delivery to customers. Asset risk levels are also evaluated Time Insights software (Courtesy of Space-Time Insight)
and available to the central power monitoring crew. (Courtesy of Space-Time Insight)

initializes the context and configures the
entire system design. Second is the de-
velopment phase which is based upon
the high-level services that the utility
needs to be active throughout the sys-
tems life cycle.
SMUDs new Distributions Operation
Center
Michael Greenhaigh told us that 55
screens are used in their system and tied
into Google Earth. SMUD personnel now
have interactive visual access on the con-
trol center video wall as well as on their
desktops. Their real-time and historical
data from the smart grid, distribution, out-
age, fire and weather systems. See Fig-
ures 4, 5 and 6 for examples of a smarter
grid making smarter decisions.

7 Sponsored by
 brought to operators attention. Green
usually indicates normal conditions, yellow
can indicate some trend that is occurring
to which preventive maintenance can re-
spond and red is usually a fault of failure.
Going forward, there should be less and
less information overload to operators and
more and more system reliability with
smarter systems.
Please let us know what experiences you
have in this area.

References
1
Wide-Area Situational Awareness for Critical Infra-
structure Protection, Alcaraz and Lopez, University of
Malaga, Spain, 2013
Figure 6: Field service crews and their details can be located in real time for effective management of such utility field
assets for customer priorities when problems crop up (Courtesy of Space-Time Insight)

One of the best examples regarding tively respond to storms, wind, fire and
Situational Intelligence is the Northeast more to prevent unnecessary outages.
blackout of 2013. With a good SI system, With the color-coding in geospatial dis-
utilities will be far better able to proac- plays, problem areas are immediately

8 Sponsored by
Smart meters can be fully customizable
By Ludek Holoubek , STMicroelectronics

N
ew European regulations and the estab-
lishment of the SmartGrid have trans-
formed the electricity distribution market.
EU regulations demands AMR (automated meter
reading) that allows any user to see his energy
consumption every 15 minutes over the entire
day, while the SmartGrid allows us to achieve
this using PLM (power line modem) communica-
tion over various protocols, for example, Linky
G1, G3, and PRIME.
The old mechanical disc meters (Ferraris) are
no longer effective and digital e-meters are tak-
ing their place. This change not only brings
easier integration and more communication
possibilities but also more intelligence. What are
the benefits of a digital e-meter made using a
microcontroller?
Functions can be changed and extended by

9 Sponsored by
 simple firmware upgrades
Microcontrollers can drive switches
and/or other devices
Tamper detection can be implemented
to compare currents in phase and the
neutral line
Tamper date, a time log, and a real time
clock with a calendar can be set up
Tariffs can be managed via connection
into the SmartGrid
Prepaid e-meters can be introduced
and managed more efficiently
The designer of a fully digital e-meter will
have to choose from these possibilities:
a) A dedicated chip for energy measure-
ment - AFE (Analog Front End) that
drives mechanical register and pulse
LED
b) A dedicated AFE connected to a micro-
controller
c) A microcontroller with integrated AFE
10
d) A general purpose microcontroller with
external ADC
e) A general purpose microcontroller with
embedded ADC
Since the EN 50470-1:2006 standard
prescribes the maximum allowed errors at
every point of the entire meter current and
voltage range, reaching the limits is not a
very easy task. That is why there are so
many AFEs on the market that are tested
and fully compliant to EN 50470-1:2006,
which prescribes ISTART, IMIN, ITR, IREF,
IMAX [ITR = IREF/2 (alternatively called
INOM), IMIN=ITR/2, ISTART = 20 mA].
Class B power meters
Lets evaluate an example of the construc-
tion of a power meter in Class B, with
maximum current of 100 A. For such a
high maximum current, the typical request
for nominal current is 5 A. In this case, a
precision of 1.5% is prescribed for a range
from IMIN = 250 mA to ITR = 500 mA, and
a precision of 1% for range from ITR = 500
mA to IMAX. Below IMIN no precision is
prescribed, but an e-meter must detect
zero current (no load) conditions.
In order to achieve this precision, the
AFE contains 16/24 bit sigma delta ADCs,
as well as advanced decimation, filtering
and computational algorithms that reject
all kinds of possible noise and errors.
Since many AFEs are external, or if prop-
erly designed when embedded are isolat-
ed from the noise of the microcontroller,
they can reach the necessary precision.
The cost we pay when using AFEs in
combination with a microcontroller is sim-
ply a two-chip solution price (although,
as well see, it can be up to a seven-chip
solution price).
Wouldnt it be better to use one chip
Sponsored by

Figure 1: The latest AFE for e-meters from STM
without the embedded dedicated AFE?
Designing an e-meter without these
AFE chips is possible; however, there are
tradeoffs that need to be considered. For
11
example, an e-meter could be constructed
using a simple 8 or 10 bit ADC that is typi-
cal among MCUs on the market today.
This is not a single chip solution. You will
need to add additional circuitry in front
of the ADC inputs (signal conditioning) in
order to handle the current range, filtering,
and noise. These additional components
will add cost to the overall solution.
In a highly competitive market, adding
cost for additional functionality (algorithm
control) is not acceptable. In order to take
advantage of the MCU solution, a good
quality ADC (external or embedded) is
necessary to build a cost effective e-meter.
Using a high quality and higher resolution
external ADC has an advantage, allowing
the designer to control how injected noise
is handled with the conversion. This exter-
nal ADC is still a two-chip solution.
The embedded ADC allows a single
chip solution, knowing that the injected
noise management is mostly handled by
the chip manufacturer. The embedded
ADC solution does add risk to the design,
Sponsored by
which needs to be managed. On the other
hand, this is a very cost effective solution.
Again, a trade off arises - noise manage-
ment versus a single chip solution with
the ability of the designer to modify the
algorithm. Depending on the application,
some customers do favor the single chip
solution.
Some customers prefer a solution in
which they can use their own computa-
tional algorithms rather than an AFE chip
with an already proven hardwired algo-
rithm. Why would that be?
They simply want to control the compu-
tation process. They do not want to rely
on an algorithm and the expertise of an
AFE supplier. They prefer to use their own
expertise. Moreover, some AFEs have
limitations. They do not perform all the
computations the customer wants, or the
definition of the computation could differ
12
from customer to customer. The required
values are not only RMS voltage, RMS
current and active power, but also reactive
power, total harmonic distortion, distortion
power, power factor, phase shift, harmonic
analysis, coverage of certain harmonics.
For example, the reactive power could be
computed more than one way:
It could be computed from a vector
triangle of the apparent and the real
power
It could be computed as real power
from the signal, where all the harmonics
of the current were shifted by 90 degree
It could be computed from Fourier
computation as an imaginary part of the
power spectra
It could be computed by the very sim-
ple but popular method of shifting the
current waveform by 90 degrees of the
fundamental harmonic
Some customers want to be able to con-
trol the computation in case they need to
change the computation algorithm quickly.
Moreover, when using current transform-
ers as current sensors, because of the
galvanic isolation, the one-chip solution for
three phase meters could be very efficient
from the point of view of PCB design, as-
sembly, reliability, and overall cost. In the
US market, where the meters are often in
a circular housing using complete metal
shielding, any noise the lines coming from
sensors to the chip will be decidedly re-
jected. Using the shunts for three phase
measurements makes the use of a multi-
chip solution a must because of the insula-
tion implementation.
A powerful general-purpose microcon-
troller, such as that introduced by STMi-
croelectronics, has been used for our ex-
ample: the STM32F373 Cortex-M4F core
Sponsored by
running on 72 MHz with three embedded reduces only2 the differential non-linearity are able to manage the overall in-system
50 ksps 16-bit sigma-delta ADCs (SDADC) (DNL) of an ADC is the main presumption to noise3, we could get an increased number
and one 1 Msps 12-bit fast successive be able to build a precise metering system. of effective bits from the ADC.
approximation register ADC. The very suc- Although some ADC precision correc- Although single values coming from the
cessful value, performance and connectiv- tion is possible by software, it could cause ADC contain some noise, the situation is
ity line series from STM32 family, which difficulties in the manufacturing, certifica- different if we speak about the RMS value
includes the value-, performance- and tion and calibration processes. Since for computed from the aforementioned values.
connectivity-line series, continues with the non-sigma-delta ADC the hardware (or The noise drops with the number of the
Analog line series. The embedded analog software) linearity enhancement is needed, values used for all computation. To sum-
part makes this microcontroller suitable for when higher precision is expected, sigma- marize, attention must be paid to the over-
a 3-phase e-meter implementation giving delta ADC are naturally very linear. If we all system linearity, in-system correlated
the computational flexibility mentioned in
previous paragraphs. STM32F2 at STM32F4 at
FFT Data
Prior to the STM32F373 announcement, 120 MHz 168 MHz Gain
Algorithm format
ST had already introduced a success- units in (s) units in (s)
ful design1 of a one-phase e-meter using
Q15 (16-bit) 63.4 22.1 x2.9
a general-purpose microcontroller, the
STM32F100 with 12-bit ADC. A lot of the 64-points Q31 (32-bit) 69.6 40.6 x1.7
expertise learned through the STM32F100
Floating 443.5 30.9 x14.3
design was used in the STM32F373. For
example, the knowledge that oversampling Figure 2: Floating-point performance

13 Sponsored by
noise, and the overall system performance. ues, tariff specification and load profiles world, half of them preferred the oppo-
The STM32F373 brings this performance, Embedded RTC, internal oscillators, and site: chip on glass approach. Why?
and furthermore, it brings features that communication peripherals make this The chip on glass always contains a volt-
make it possible to build not only a cost chip easy to use standalone but also age booster that allows using less-ex-
effective e-meter, but one with a harmonic interconnect with other devices like IR pensive LCDs in order to cover the entire
analyzer. Since the STM32F373 delivers ports, PLM, MEMs, among others operational temperature range7 and still
these features, an e-meter implementation Although the most important output of the offer a quality, easy to read, display.
can be relatively simple and effective: e-meter used for certification and calibra- The chip on glass uses only SPI: it is
The standard, widely used Cortex-M4F tion purposes is the LED output, an LCD easier to maintain EMC for just three
32-bit core with floating point, changing display is an integral part of the e-meter. wires.
the bothering with fixed-point precision In the EU, ARM systems are still mostly Because only three wires for an LCD
to the pleasure of working in floating in pilots, and customers demand meters are used, this leaves many pins free
point4 with a display in order to control their for other uses, saving the cost of GPIO
An available DSP library using the DSP hour-to-hour or day-to-day consumption. expanders or extra solutions to extend
functions of the Cortex-M4F core makes The design of an e-meter display rightly functionality of the microcontroller.
an implementation of the all the math, receives a lot of attention since that is PCB layout is faster and easier.
filtering, and FFT, for example, very the primary consumer interface. Even When using a software LCD driver in
simple and easy 5
though the directly driven LCD segment microcontroller rather than a hardware
The STM32F373 on-chip flash can be display was thought to be the least LCD driver, using chip on glass reduces
used as an EEPROM for calibration
6
expensive and most popular solution, the microcontroller load and makes the
constants, tamper data, measured val- when we question customers over the software shorter and simpler.

14 Sponsored by
 With the huge number of STM32F373
GPIO pins (for the 100-pin package, there
are 84 GPIO pins), the SW LCD driver
could directly drive a large segment LCD8
in case this option is needed. The e-meter
reference design makes it possible to drive
LCDs from 4 x 14 segments to 4 x 31 or to
connect SPI B&W and color display.
Although some ADC precision correc-
tion is possible by software, it could cause
difficulties in the manufacturing, certifica-
tion and calibration processes. Since for
non-sigma-delta ADC the hardware (or
software) linearity enhancement is needed,
when higher precision is expected, sigma-
delta ADC are naturally very linear. If we
are able to manage the overall in-system
noise[3], we could get an increased num-
ber of effective bits from the ADC.
Although single values coming from the
ADC contain some noise, the situation is
15
different if we speak about the RMS value
computed from the aforementioned values.
The noise drops with the number of the
values used for all computation. To sum-
marize, attention must be paid to the over-
all system linearity, in-system correlated
noise, and the overall system performance.
The STM32F373 brings this performance,
and furthermore, it brings features that
make it possible to build not only a cost
effective e-meter, but one with a harmonic
analyzer. Since the STM32F373 delivers
these features, an e-meter implementation
can be relatively simple and effective:
The standard, widely used Cortex-
M4F 32-bit core with floating point,
changing the bothering with fixed-
point precision to the pleasure of
working in floating point[4]
An available DSP library using the
DSP functions of the Cortex-M4F
core makes an implementation of the
all the math, filtering, and FFT, for ex-
ample, very simple and easy5
The STM32F373 on-chip flash can be
used as an EEPROM6 for calibration
constants, tamper data, measured val-
ues, tariff specification and load profiles
Embedded RTC, internal oscillators,
and communication peripherals make
this chip easy to use standalone but
also interconnect with other devices like
IR ports, PLM, MEMs, among others
Although the most important output of
the e-meter used for certification and cali-
bration purposes is the LED output, an
LCD display is an integral part of the e-me-
ter. In the EU, ARM systems are still mostly
in pilots, and customers demand meters
with a display in order to control their hour-
to-hour or day-to-day consumption.
The design of an e-meter display rightly
Sponsored by
 USB-Stink LCD ZigBee

LED
optoisolated Segment LDC
SPI Color of b&w

Joystick Crypto
AUKEY

PLM STM32
ST75xx connector (1SAR+3SD) MEMS

SMPS
Dual
Eeprom
CT CT CT CT powerless
reading
product in box
rogowski rogowski rogowski rogowski SHUNT customisation

Figure 3: STEVAL-IPE022V1 e-meter reference design

receives a lot of attention since that is the
primary consumer interface. Even though
the directly driven LCD segment display
was thought to be the least expensive and
most popular solution, when we question
customers over the world, half of them pensive LCDs in order to cover the entire
preferred the opposite: chip on glass ap- operational temperature range7 and still
proach. Why? offer a quality, easy to read, display.
The chip on glass always contains a volt- The chip on glass uses only SPI: it is eas-
age booster that allows using less-ex- ier to maintain EMC for just three wires.

16 Sponsored by
 Because only three wires for an LCD
are used, this leaves many pins free
for other uses, saving the cost of GPIO
expanders or extra solutions to extend
functionality of the microcontroller.
PCB layout is faster and easier.
When using a software LCD driver in
microcontroller rather than a hardware
LCD driver, using chip on glass reduces
the microcontroller load and makes the
software shorter and simpler.
With the huge number of STM32F373
GPIO pins (for the 100-pin package, there
are 84 GPIO pins), the SW LCD driver
could directly drive a large segment LCD8
in case this option is needed. The e-meter
reference design makes it possible to drive
LCDs from 4 x 14 segments to 4 x 31 or to
connect SPI B&W and color display.
Once the sensors, the display, and the
microcontroller with ADC, are chosen, the
17
suitable algorithm for power computa-
tion should be implemented. Two ways of
power computation could be used - com-
putation in a time domain or in a frequency
domain. Both approaches have their ad-
vantages and disadvantages:
Time domain computation
Very easy implementation for sys-
tems where the phase shift (U vs. I)
can be easily compensated (by data
shift) before the data are given to the
computation algorithm. Otherwise, it
is not so easy to perform this com-
pensation: not trivial computation for
fundamental frequency and nearly
impossible (advanced filtering) for
higher frequencies. Selective phase
compensation of the higher harmon-
ics is nearly impossible.
The computational load of the micro-
controller is very low for this approach
when just RMS values and active
power are needed.
When additional data like the power
factor, reactive power, THD, higher
harmonic content, and cosines phi
are needed, the computation become
very complex.
Frequency domain computation
All the computation is done in one shot,
where FFT delivers all the information
about the signals: amplitudes and an-
gles of all harmonics.
Since 6 x 64-floating-point FFT on 72
MHz Cortex-M4F core takes 1 ms in total
(3 voltages and 3 currents), we can even
implement a frequency analyzer easily.
More computational power is taken by
the goniometric functions used further for
active and reactive power, PF, THD, com-
putation: tan-1, sin and cos. The compu-
tation load can be reduced by a strong
Sponsored by

focus on the important harmonics - for
example, on the first 10, and a weak focus
on the rest of the harmonics in the signal.
E-meter evaluation kits for STM32F373,
the STEVAL-IPE022V1, present a very
scalable power metering system designed
to measure electric energy in a 3-phase
network for the maximum current of 100 A
per phase while IREF is kept low at 5A us-
ing current transformer as a current sensor.
The board features the following:
Reaching Class B - EN 50470-3:2006 or
Class 1 - IEC 62053-21:2003 accuracy
Ready for industrial case, 16 x 16 cm
PCB including 3-phase SMPS power
supply from ST
Peripheral on board or to interconnect:
MEMS, PLM, ZigBee, Crypto, RFID,
Opto-head
High number of debugging possibilities:
Opto-isolated JTAG tool on board[9]
18
(isolation can be bridged and the
board could be powered from a USB
when debugging is needed)
Segment or color SPI graphic display, 4
joystick, user configuration jumpers
Demonstration software optimized demontration
for readability: structured variables,
modularity
The STEVAL-IPE022V1 board works ac-
cording to requirements specified in EN
50470-1:2006 with a current transformer
as a current sensor and is available to 9
customers. There is also an experimental
option on the board that allows designers
to interconnect one shunt or four Rogowski
coils directly or through an external opera-
tional amplifier.
References
1
One phase measurement with a shunt introducing very low
phase shift and moderate requirements (no THD, ...) allowed
for time domain computation to be used instead of frequen-
cy domain computation. See AN3322.
2
That brings better resolution, but does not affect precision
represented by integral non-linearity (INL).
3
White noise or noise not correlated with mains and system
processes if not required by algorighms.
Due to the core implementation, the 32-bit floating-point
computation is faster than 32-bit fixed-point, see figure 1.
5
Standard CMSIS libraries are available with a number of
examples.
6
ST provides C code that implement virtual EEPROM, tak-
ing care of flash rewrite cycle optimization.
7
In contrast to the directly driven LCD by microcontroller
using only VCC = 3.3V.
8
SW for microcontrollers without HW LCD driver are sup-
plied by ST.
ST-link connected and powered from USB since the JTAG
signals are opto-isolated.
Sponsored by
balance of power
The

Developing a smart meter

By Khurram Waheed, Freescale Semiconductor

E
lectric utilities face the challenging task of matching highly
variable demand to fixed supply. Peak demand can be a multiple
of average demand, making it hard to get full value from highly
capital intensive power generation plants. Offering consumers a
time-based incentivized tariff, or having the capability to remotely
manage the schedules of high consumption equipment, is a more
efficient and controllable way of balancing power supply and demand.
To offer a time-based tariff, the utility must know when the user is
consuming power, which means there must be a communication path
from the utility to the consumers meter.

19 Sponsored by
 Obviously, wireless standards are can-
didates for utility-meter communication,
but they have problems. Zigbee-based 2.4
GHz technologies have been proposed,
but in practice they suffer from a short
range of 20-50m which doesnt provide a
cost-efficient connectivity path back to the
utility. Adding mesh networking can extend
the range of a Zigbee network, however
large intra-meter distances in rural and
suburban settings, and intervening build-
ing structures, present a challenging en-
vironment for wireless communication in
the ISM 2.4GHz band. As an example, a
crowded multi-story apartment complex
presents intricate coverage and connectiv-
ity issues that can quickly make a wireless
network expensive to install and maintain.
The new IEEE 802.15.4g Smart Utility
Networks (SUN) standard is designed to
address these challenges using various
frequency bands, but its not ready for
20
mass deployment. The biggest drawback
with wireless is the need to use scarce and
valuable spectrum, which might not be
cheaply and/or readily available.
The obvious communications path is the
powerline itself. Why not piggyback a sig-
nal along the electricity supply?
For a vey long time, utilities have been
using powerlines to communicate with
generation equipment and provide voice
communication between sub-stations.
This involves adding a signal to high-
tension lines. These signals typically use
amplitude modulation or Frequency Shift
Keying and work at a modest data rate
of 1 kb/s or less, the low data rate allow-
ing the signal to propagate over long dis-
tances. This technology is used to switch
in and out distribution units, to check the
integrity of the grid, and to provide basic
voice communication to remote generation
and sub-stations. However, there are thou-
sands more electricity meters than dis-
tribution units and this approach doesnt
scale well for utility-meter links.
To communicate with meters, there is
need for a higher data rate and a technol-
ogy that supports multiple users in a semi-
autonomous and agile communication net-
work. The G3-PLC narrowband orthogonal
frequency-division multiplex (OFDM) pow-
erline communication standard has been
designed for this purpose and provides
the ability to overcome unique powerline
communication challenges. The standard
borrows many technologies from wire-
less, for example OFDM communication,
forward error correction mechanisms such
as Reed-Solomon, Viterbi convolutional
decoding, and time and frequency domain
interleaving to name just a few.
The powerline itself is a very challenging
environment, which in many ways is noisier
than wireless channels. The noise on a
Sponsored by
powerline is highly non-stationary and has
a Gaussian component (similar to wireless)
and a non-Gaussian, impulsive noise com-
ponent (very different from wireless) that can
be either periodic or aperiodic (Figure 1).
Interferers on a powerline can be ei-
Figure 1: Time domain capture of typical powerline noise in an
industrial building (a) raw power line noise samples, (b) Cenelec-A
band powerline noise
21
ther narrowband or wideband -emanat-
ing from other devices on the network or
other communication networks. For ex-
ample, induction motors introduce a lot of
cyclostationary noise and are common;
theyre present in industrial machinery,
washing machines, refrigerators
etc. Switched mode power sup-
plies also introduce noise, and
given their popularity for phone
and laptop chargers, their use is
widespread. Because they are a
physical medium, powerlines suf-
fer from branching; the impedance
at one point in the network can be
quite different from the impedance
elsewhere, even in the same house.
Turning on and off equipment can
instantaneously change the im-
pedance at a point in the network.
Typically, powerline impedances
range from 0.1-200 . All told, the
powerline channel is complex and noisy.
Figure 1 and Figure 2 show an example of
typical powerline noise in time and time-
frequency domain that a G3-PLC modem
has to endure.
At Freescale, our group develops tech-
nologies for smart metering in particular,
powerline communications. Our communi-
cation network is built using powerline mo-
dems (PLM) comprising a microcontroller
with embedded digital and analog signal
processing, an analog front-end, and a line
coupler (see Figure 3).
The modem implements the PHY from
the G3-PLC standard, plus proprietary
signal processing that we developed to
improve the performance of our PLM. The
analog front-end converts the digital out-
put of the modem signal processor to an
analog voltage that meets regulations for
signals added to powerlines. However,
the output of the analog front end is a low
Sponsored by
 our system problems in simulation before
building hardware. So the starting point
for our work is our extensive end-to-end
system-level model that weve written in
MATLAB and Simulink (see Figure 4).
We begin by simulating the digital parts of
the system. Like most standards, G3-PLC
specifies the transmission algorithms, see
Figure 4, but has less detail on the receive
side. Before our received signal even gets
(a) (b)
to the G3-PLC specified physical layer, we
Figure 2: Time-Frequency spectrograph of (a) Raw powerline noise showing the complex structure of powerline noise. need to be sure the algorithms weve writ-
Notice the various cyclostationary noise components and their spectral content; (b) Powerline noise within Cenelec-A
band as seen by a G3-PLC modem. Notice 50dB stronger 120/240Hz noise components in the frequency range of 62
ten for signal recovery are good enough
to 72 kHz. to extract the signal from the noise and
voltage signal that needs to be amplified support multiple users and will perform reflections on the line. We model those
by a line driver and added to the power- well under time-varying and harsh channel algorithms in MATLAB and use our system
line using the third element; the line cou- conditions all at the lowest cost. modeling to dynamically verify algorithm
pler. The powerline impedance can vary Obviously, building hardware is very ex- performance under a wide range of condi-
dramatically, so the line driver is typically pensive and our goal is to be foolproof in tions before deployment on actual hard-
configured to drive signals to a powerline our component choice and achieve first- ware. We also use the simulation as a tool
with a time-variant impedance. Our over- pass success. In turn, this means simula- to study the trade-offs involved in integra-
all challenge is to build a system that will tion is extremely important we want to fix tion of performance enhancement features

22 Sponsored by
Figure 3: Simulation/Modeling Strategy for PowerLine Model Specification and
Performance Evaluation.
into the front-end of the modem and reduce
the overall system bill of material and cost.
In addition to digital simulation, we need
to model the analog signal processing
embedded in the modem chip as well as
components for our external analog front-
end. Each analog component has a differ-
ent performance and price. Simulink simu-
lations can be used to identify the critical
23
Figure 4: G3-PLC Transmit Model realized in Matlab/Simulink.
specifications for the front-end compo-
nents external to the powerline modem
and examine their effect on system per-
formance. A system comprises of multiple
analog, mixed-signal and digital compo-
nents and we need to model and simulate
their interactions. Each system component
has a specific linearity, signal bandwidth
and frequency response. Non-linearities
can severely impact system performance
so we need to examine IP2, IP3, P1dB,
and spectral regrowth (see Figure 5).
Some component choices can impede
the system performance in undesirable
ways. If we cant change the component,
then we may have to correct for its non-
linearities somewhere in the system. At the
system-level, were interested in the perfor-
Sponsored by
 (a) (b)

Figure 5: Linearity Analysis of the Transmit Line Pre-Driver Stage (a) IP3 requirements, (b) IP2 requirements

mance of our gain control algorithms and
we want to evaluate BER and receiver sen-
sitivity under real powerline conditions. Of
course, our goal is to achieve our require-
ments at the lowest price, which means
running a large number of simulations, mak-
ing speed of simulation crucial. We use an
extensive end-to-end mixed-signal bit error
rate (BER) simulation in Simulink using a
simple Gaussian noise channel model to cut
the computational cost.
We use a similar process to specify com-
ponents for the line coupler. Depending on
the type of powerline communications we
want, this could couple to a low voltage

24 Sponsored by
110V or 220V powerline or even a medium
voltage powerline at 6kV or 22kV. Within
the line coupler, there are a number of pos-
sible trade-offs and we use simulation to
model and investigate their performance
impact. We also have an output constraint
imposed by regulation. CENELEC (the
European electrical engineering standard-
ization body) states that the maximum
signal we can add to a 220V powerline is
134dBV over the entire bandwidth. Part
of our component selection procedure is to
check that we remain within this limit and
meet the inband and out-of-band spectral
requirements.
Once we have our candidate analog com-
ponents, we need to understand how robust
our system will be under a range of operat-
ing condition. This means we need a more
sophisticated channel model than our simple
Gaussian model. Weve developed an exten-
sive powerline channel model that includes
25
impulsive noise, frequency selective noise,
Gaussian noise, and other effects.
The model is a great deal more sophisti-
cated and detailed than our simple Gauss-
ian channel model, but the cost is the add-
ed computational power needed. Because
the more detailed channel model takes lon-
ger to run, we reserve its use for the more
detailed system modeling that takes place
towards the end of system development.
We run this model under a range of condi-
tions to check that our system stays within
the bounds of our requirements. In prac-
tice, this means we run parameter sweeps
of our models using a MATLAB script.
Channel conditions determine the modu-
lation scheme. The G3-PLC standard in-
cludes different modulation types that can
be adaptively chosen based on the signal
to noise ratio of the received signal. For
example ROBO or DBPSK is chosen when
theres a lot of noise, DQPSK when there is
less noise, and D8PSK when the channel
has very little noise. Our system has to be
able to switch between different possible
modulation schemes dynamically depend-
ing on the powerline noise level, so we test
this behavior as part of our system-wide
evaluations (see Figure 6).
Our simulation needs vary through the
development process, so the sophistica-
tion of our models evolves too. At the
beginning, we need to simulate a direct
connection between the transmit and re-
ceive physical layers, then we need to add
in front-end models, channel models, line
coupler models etc. To make life easier,
we have created a configurable MATLAB
model that enables us to build the model
for the simulation architecture we need.
In the tool, we specify which components
we want to model and how to connect
them. For example, we might specify that
we want the transmit physical layer model
Sponsored by

Figure 6: Frame Error Rate of a G3-PLC narrowband OFDM powerline
communication system on an isolated clean powerline.
(written in MATLAB) connected to an ana-
log front end (written in Simulink) including ware should give the same answer at
the line coupler model. Our tool makes all
the necessary connections.
Eventually of course, design works
26
comes to an end and its
time to finalize components
and build the real hardware.
Once built, the hardware has
to be tested, so we use our
simulation environment again
to help verify the behavior
of the hardware. We use our
model to generate test vec-
tors for a wide range of op-
erating conditions. We also
do a step-by-step compari-
son of our hardware and our
MATLAB physical layer mod-
els and their performance in
the presence of noise. Our
MATLAB model is designed
to be bit-exact, so our model and hard-
each stage in the processing chain and
for all data input.
Smart meters are an important technol-
ogy for the near future, but the mix of ana-
log, digital, and powerline systems creates
unique challenges. We have to get the bal-
ance of power right between these compo-
nents and speed of simulation is critical
to building our system in a timely way.
Smart meters using G3-PLC communica-
tion technology are currently being evaluat-
ed in pilot projects. Our goal is to contrib-
ute to this technology for mass adoption.
In a few years time, this technology will be
coming to a home near you.
References
G3-PLC Alliance Homepage; Website:
http://www.g3-plc.com/
Cyclic Spectral Analysis of Power Line Noise in the
3-200 kHz Band, Karl Nieman, Jing Lin, Marcel Nassar,
Khurram Waheed, and Brian L. Evans; IEEE International
Symposium on Power Line Communications and its Ap-
plications, 2013.
G3-PLC Specification for Powerline Communication:
Overview, System Simulation and Field Trial Results,
Kaveh Razazian, Maher Umari, Amir Kamalizad, Victor
Loginov, Michael Navid; IEEE International Symposium on
Power Line Communications and its Applications, 2010.
Sponsored by
SPONSORED CONTENT
Reducing Cost and
Complexity with Industrial
System Consolidation
V
irtualization on multi-core Intel Core
vPro processors helps lower overall
solution cost and reduce factory footprint
and integration effort through hardware consoli-
dation.
Summary
In todays highly competitive manufactur-
ing environment, success requires a constant
focus on cost cutting while maintaining pro-
27
duction throughput and employee safety. For
manufacturers, this includes finding new ways
to lower operating expenses, a large part of
which are the purchase and support of indus-
trial systems. A significant cost stems from the
inefficiencies created by the growing numbers
and varieties of systems on the factory floor.
For instance, system proliferation is consum-
ing precious space and straining IT resources,
especially when systems have unique support
Sponsored by
SPONSORED CONTENT
requirements for configuration, backups,
spares and software patching.
Efficiency can be improved when multi-
ple factory functions are consolidated onto
a single hardware platform, thus decreas-
ing operating expense, factory footprint,
energy consumption, and integration and
support effort. This can be done using ad-
vanced multi-core processors along with
proven virtualization technology, which has
been around since the 1960s1 and is most
notably used in data centers where many
applications are consolidated onto a single
server. Still, virtualization tools and meth-
ods used in the server environment are
different from what is appropriate for the
embedded environment.
This white paper describes how virtual-
ization technology running on multi-core
Intel Core vPro processors can be
used in industrial automation to consoli-
date computing devices for motion con-
28
trol, programmable logic control (PLC), devices), so each VM runs as if it had its
human machine interface (HMI), machine own machine.
vision, data acquisition, functional safety As a result, applications run on their na-
and so forth. This approach can help man- tive OSes (referred to as guest OSes in
ufacturers reduce cost and complexity on virtualization parlance), allowing them to
the factory floor. easily migrate to a new system often with
only minor or no changes. To illustrate this
Virtualization Basics capability, Figure 1 shows that four ap-
In traditional industrial automation sys- plications running on their own OSes and
tems, application software, the operating boards can be consolidated onto a single
system (OS) and the physical hardware are board with a multi-core processor and a
tightly coupled. Virtualization breaks this hypervisor. The hypervisor manages the
link and provides the ability to run multiple execution of guest OSes in much the same
OSes and their associated applications on way an OS manages the execution of the
the same physical board. This is achieved applications it hosts.
by executing software in individual parti-
tions, called virtual machines (VMs), that Virtualization in Industrial Automation
are managed by a new software layer, Some industrial control systems are de-
known as the hypervisor or virtual machine signed with multiple boards because they
monitor (VMM). The hypervisor abstracts run applications like PLC, motion control,
the boards underlying hardware resources and HMI with different sets of require-
(e.g., processor cores, memory and I/O ments. PLC and motion control are time-
Sponsored by
SPONSORED CONTENT

time tasks that would otherwise compete

for CPU resources.

Benefits from Consolidation

By consolidating devices using virtualiza-
tion technology, original equipment manu-
Hypervisor
facturers (OEMs) developing industrial au-
tomation solutions can provide substantial
benefits to their customers, such as:
Single-Board Multi-core Processor

Multiple Single-Processor Boards Lower overall solution cost: Although

Figure 1. General Virtualization Example
critical applications, which are best served
by a real-time operating system (RTOS)
that delivers deterministic performance. In
contrast, developers of HMI applications
may prefer a general-purpose operating
system (GPOS) supported by tools that
ease the development of touch screen dis-
plays, rich graphics and multimedia.
Figure 2 shows how a single board with
virtualization technology can address all
these requirements, as well as others dis-
cussed later. Multi-core processors with
virtualization technology allow systems to
simultaneously run RTOSes and GPOSes,
each on dedicated processor cores. This
configuration can increase the determin-
ism of time-critical applications, because
they operate unencumbered by non-real-
a consolidated device may cost more
than any of the individual subsystems,
it should cost less to manufacture than
the combined subsystems because it
has a smaller bill of materials (BOM). In
addition, virtualization makes it easier
for OEMs to add new functionality to a
system and expand their offerings.
Smaller factory footprint: Consoli-
dated equipment takes up less factory

29 Sponsored by
SPONSORED CONTENT

floor space than the individual systems require security software and may elimi- Virtual Machine 1 Virtual Machine 2 Virtual Machine 3

it replaces. nate some varieties of security solutions Soft PLC Data Acquisition
Other applications
(e.g., HMI)
the factory must support. In addition,
Real-time General-purpose General-purpose
Reduced overall energy there are fewer devices for hackers to operating system operation system operation system

consumption: The power efficiency of attempt to infiltrate, thus reducing the

Intel Core vPro processors, combined attack surface of the factory floor. Hypervisor

with system consolidation, can yield a

solution that consumes less power than Easier system management: Intel Core vPro Processor with Intel Virtualization Technology

the individual systems combined. When subsystems are consolidated,

Reduced integration cost:
By consolidating subsystems, OEMs
effectively eliminate many integration
tasks for their customers. For instance,
the networking, cabling, shielding and
configuration that connect multiple
subsystems together are handled with-
in the system.
Simpler to secure:
The consolidated approach decreases
the number of computing devices that
factory IT personnel have a smaller
number of devices to install, provision
and manage. Also, a consolidated sys-
tem is likely to have more capable hard-
ware and software than the subsystems
it replaces, allowing for additional man-
ageability options and capabilities.
Higher reliability:
The greater the number of systems, the
larger the number of devices that can fail.
Consequently, a consolidated system
should have a better mean time between
Figure 2. Industrial System Consolidation Example
failures (MTBF) than the combination of
subsystems it replaces. Furthermore,
there are fewer spares for factories to
carry, and maintenance and repair proce-
dures are simpler all ultimately leading
to shorter downtimes.
Consolidating Systems on Multi-Core
Processors
Multi-core architectures, such as Intel
Core vPro processors, provide the com-

30 Sponsored by
SPONSORED CONTENT
puting power needed to consolidate in-
dustrial systems and deliver real-time,
deterministic performance. Multi-core
processor architecture allows OEMs to
dedicate hardware-level computing re-
sources to specific VMs, thereby enabling
an RTOS to behave deterministically re-
gardless of the applications running in the
other VMs. In addition, developers can
more easily reallocate system resources
across cores as system needs change.
One of the key benefits of consolidation
is improved resource efficiency, which is
achieved through a multi-core architec-
ture-based platform. An industrial solution
that combines multiple subsystems on
one platform requires just one comput-
ing system and power supply, which re-
sults in a smaller footprint, higher density,
lower power consumption and a simpler
design compared to multiple subsystems
with their own hardware.
31
Hardware-Assisted Virtualization hardware to prevent the software running
Technologies in one VM from interfering with the soft-
Although virtualization is generally viewed ware running in another VM. Along these
as a software technology, hardware fea- lines, virtualization helps avoid unintended
tures have been added to processors to interactions between applications by pre-
improve the performance and security venting one from accessing anothers
of virtualization. For instance, Intel has memory space. Some of the key benefits
enhanced the capabilities of virtualiza- of virtualization in industrial automation
tion technology with a complementary and other embedded applications are
hardware-assist technology called Intel listed in Table 1.
Virtualization Technology (Intel VT), an
2
With respect to performance, Intel has
ingredient of Intel vPro technology. developed three different, yet comple-
It performs various virtualization tasks in mentary, virtualization acceleration tech-
hardware, like memory address translation, nologies that span multiple platform com-
which reduces the overhead and footprint ponents, including the processor, chipset
of virtualization software, and improves and NICs:
its performance. For instance, VM to VM Intel Virtualization Technology
switching time is significantly faster when (Intel VT) for IA-32, Intel 64 and
memory address translation is performed Intel Architecture (Intel VT-x) speeds
in hardware instead of by software. up the transfer of platform control be-
In addition, Intel VT increases the robust- tween the guest OSes and the hypervisor.
ness of virtualized environments by using In Intel processors, it reduces virtualiza-
Sponsored by
SPONSORED CONTENT
Capabilities
Isolates applications in secure partitions
Runs RTOS on a dedicated processor core
Performs virtualization tasks in hardware
Table 1. Intel Virtualization Technology Capabilities and Benefits
tion overhead by eliminating the need for
the hypervisor to listen, trap and execute
certain instructions on behalf of each
guest OS. When hypervisor interventions
are required, it provides hardware support
so handoffs between the hypervisor and
guest OSes are faster and more secure.
Intel Virtualization Technology
(Intel VT) for Directed I/O (Intel VT-d)
accelerates data movement by enabling
the hypervisor to directly and securely as-
32
Benefits
Increases system reliability and stability
Eases software migration and consolidtaion
Decreases loop jitter
Improves determinism
Decreases hypervisor load on the processor
Reduces VM to VM switching time
sign I/O devices to specific guest OSes.
Each device is given a dedicated area
in system memory so data can travel
directly and without hypervisor involve-
ment. I/O traffic flows more quickly, with
more processor cycles available to run
applications. Security and availability are
also improved, since I/O data intended for
a specific device or guest OS cannot be
accessed by any other hardware or guest
software component.
Intel Virtualization Technology
(Intel VT) for Connectivity (Intel
VT-c) performs PCI-SIG* Single Root I/O
Virtualization (SR-IOV) functions that al-
low the partitioning of a single Intel
Ethernet Server Adapter port into multiple
virtual functions. These virtual functions
may be allocated to VMs, each with their
own bandwidth allocation. They offer a
high-performance, low-latency path for
data packets to get into the VM. Intel
VT-c, integrated in Intel Ethernet NICs,
enables improved networking throughput
with lower CPU utilization and reduced
system latency.
Improving Virtualization Performance
Its possible to ensure the real-time per-
formance necessary for consolidated
factory automation solutions using Intel
VT and an RTOS when several main is-
sues are addressed. Foremost, its neces-
Sponsored by
SPONSORED CONTENT
sary to minimize the interrupt latency and
the overhead associated with general-
purpose processors. A major source of
performance loss is from VM enters and
exits, which typically occur when the hy-
pervisor must service an interrupt or han-
dle a special event. These transitions are
expensive operations because execution
contexts must be saved and retrieved,
and during this time the guest is stalled.
Figure 3 depicts the VM/Host enters
and exits that could result from an exter-
nal interrupt. In this case, the guest OS
runs until an external interrupt arrives.
Subsequently, there are a total of eight
exits and enters before the guest OS is
allowed to restart its stalled process. This
overhead can become substantial since
its not uncommon for I/O-intensive ap-
plications to have hundreds or thousands
of interrupts arriving in a second. These
constant disruptions cannot be tolerated
33
with time-critical control applications
because of the resulting degradation in
performance, latency and determinism.
Intel has worked together with operating
system vendors to reduce the typical inter-
rupt latency from between 300 and 700 uS
to sub 20 uS,3,4 achieving near-native per-
formance (i.e., similar to non-virtualized) in
a virtualized environment. This is possible
through the implementation of hardware
and software Exitmechanisms that mini-
mize the interrupt overhead inherent in a
Guest Running
Vm Exit
Host Enter Host Exit
Host Running
Interrupt
Figure 3. Interrupt Impact
virtualized environment, some of which are
described in the following:
Intel Virtualization Technology
FlexPriority: When a processor is per-
forming a control task, it often receives
interrupts from other devices or appli-
cations. To minimize the impact on per-
formance, a special register, called the
APIC Task Priority Register (TPR), in the
processor monitors the priority of tasks
to prevent the interruption of one task
Running Running
VM Enter Vm Exit VM Enter
Host Enter Host Exit
Running
Sponsored by
SPONSORED CONTENT
by another with lower priority. Intel Vir-
tualization Technology FlexPriority (Fig-
ure 4) creates a virtual copy of the TPR
that can be read, and in some cases
changed, by guest OSes without hyper-
visor intervention. This eliminates most
VM exits due to guests accessing task
Without Intel Virtualizatoin Technology FlexPriority
Virtual Machine (VM)
Guest
Operating System
VM
Exits
APIC-TPR access in
software
Virtual Machine Monitor (VMM)
Fetch/decode instruction
Emulate APIC-TPR behavior
Thousands of cycles per exit
Figure 4. Depiction of Intel Virtualization Technology FlexPriority
34
priority registers and thereby provides a
major performance improvement.
Virtual Processor IDs (VPID): Previ-
ously, every time the hypervisor per-
formed content switching between
VMs, the active VM and its data struc-
With Intel Virtualizatoin Technology FlexPriority
Virtual Machine (VM)
Guest
Operating System
No VM
Exits
APIC-TPR access in
hardware
configure
Virtual Machine Monitor (VMM)
Instruction executes directly
Hardware emulates APIC-TPR access
No VM exits
ture had to be flushed out of the transi-
tion look-aside buffers (TLB) associated
with the CPU caches. As a result, there
was performance loss on all VM exits
because the hypervisor did not know
which cache line was associated with
any particular VM.
With Virtual Processor IDs (VPID),
the virtual machine control structure
(VMCS) contains a VM ID tag that as-
sociates cache lines with each actively
running VM on the CPU. This permits
the CPU to flush only the cache lines
associated with a particular VM when
it is flushed from the CPU, avoiding the
need to reload cache lines for a VM that
was not migrated and resulting in lower
overhead.
Guest Preemption Timer: Program-
mable by hypervisor, this timer pro-
vides a mechanism to enable a hy-
Sponsored by
SPONSORED CONTENT
pervisor to preempt (i.e., halt) the
execution of a guest OS by causing a
VM exit when the timer expires. This
feature makes it easier to switch tasks,
fulfill real-time control requirements
or allocate a certain amount of CPU
cycles to a task.
Descriptor Table Exiting: This feature
enables a hypervisor to protect a guest
OS from internal attack by preventing
the relocation of key system data struc-
tures. This mechanism helps to better
protect safety-critical applications.
Pause-Loop Exiting: Spin-locking
code typically uses PAUSE instruc-
tions in a loop. This feature detects
when the duration of a loop is longer
than normal (a sign of lock-holder
preemption) and forces an exit into the
hypervisor. After the hypervisor takes
35
control, it can schedule another VM.
Spin locks are often used in control
applications for inter-process synchro-
nization.
Virtual Advanced Programmable
Interrupt Controller (vAPIC): The hy-
Without Virtual APIC (vAPIC)
Virtual Machine (VM)
Guest
Operating System
VM
Exits
vAPIC model
in software
Virtual Machine Monitor (VMM)
Fetch/decode instruction
Emulate APIC behavior
Approximately 15,000 cycle per exit
Figure 5. The vAPIC Implemented in Hardware
pervisor previously had to maintain
a virtual APIC model in software for
handling interrupts. This functionality
is now implemented with microcode,
called the vAPCI, which the guest can
access without triggering a VM exit, as
shown in Figure 5.
With Virtual APIC (vAPIC)
Virtual Machine (VM)
Guest
Operating System
No VM
Exits
vAPIC in CPU
(hardware microcode)
configure
Virtual Machine Monitor (VMM)
Instruction executes directly
Hardware and microcode emulate APIC
No VM exits
Sponsored by
SPONSORED CONTENT

Single-Root I/O Virtualization

(SR-IOV): The PCI Special Interest
Group (PCI-SIG) specification, SR-IOV
allows one NIC to service multiple
VMs, as shown in Figure 6. The
specification provides a standard
mechanism for devices to advertise
their ability to be simultaneously
shared among multiple virtual
machines. It also allows for the par-
titioning of a PCI function into many
virtual interfaces for the purpose of
sharing the resources of a PCI Express*
device in a virtual environment.
Figure 6. Technologies for Improving Virtualized I/O

Each virtual function can support a

unique and separate data path for I/O-re-
lated functions within the PCI Express hi-
erarchy. Use of SR-IOV in factory automa-
tion, for example, allows the bandwidth of
a NIC to be partitioned into smaller slices
that may be allocated to specific virtual
machines, or guests, via a standard inter-
face. This resource sharing can increase
the total utilization of any given resource
presented on an SR-IOV-capable PCI
Express device, potentially reducing the
cost of a virtual system.
Deploying Intel Virtualization
Technology
Intel Virtualization Technology is enabled
by a number of hardware and software
components, including Intel VT-enabled
Intel processors and chipsets, which are

36 Sponsored by
SPONSORED CONTENT
Platform Components
Intel Core vPro Processor Intel Virtualization Technology (Intel VT)-enabled
Intel Chipset
Virtual Machine Monitor Software
BIOS
Table 2. Required Intel Virtualization Technology Components visit
listed in Table 2. Intel VT requires virtual
machine monitor software and Intel VT-
enabled BIOS software.
The Path to Industrial System
Consolidation
For product manufacturers, factory floor
37
Required Capability
Intel VT-enabled
Available from software vendors such as Green Hills*,
LynuxWorks*, TenAsys*, Real-Time Systerms* and
Wind River*
Intel VT-enabled, available from AMI*, Phoenix* and
Insyde*
equipment plays a critical role in creating
a competitive advantage with respect to
manufacturing precision, throughput and
cost. On the cost side, industrial system
consolidation can have a large impact
in reducing overall equipment cost, fac-
tory footprint, energy consumption, in-
tegration effort, security complexity and
system support. This is achievable with
virtualization technology, a powerful capa-
bility in Intel Core vPro processor-based
platforms that allows OEMs to com-
bine multiple devices found in todays
manufacturing environment into a single
system. The path to lower cost and com-
plexity is being blazed by Intels continual
innovations in multi-core computing and
virtualization technologies that help OEMs
deliver higher performing and more robust
industrial systems.
For more information about Intel
solutions for industrial automation, visit
www.intel.com/industrial
References
1
Source: http://www.ibm.com/developerworks/
library/l-linuxvirt/
2
Intel Virtualization Technology (Intel VT) requires
a computer system with an enabled Intel processor,
Sponsored by
SPONSORED CONTENT

BIOS, virtual machine monitor (VMM) and, for some

uses, certain platform software enabled for it. Func-
Industrial System Consolidation
tionality, performance, or other benefits will vary Reduce complexity and improve factory efficiency through real-time virtualization
depending on hardware and software configurations
and may require a BIOS update. Software applica-
tions may not be compatible with all operating sys-
tems. Please check with your application vendor.
3
Performance estimates are based on internal Intel
analysis and are provided for informational purpos-
es only.
4
Performance tests and ratings are measured using
specific computer systems and/or components and
reflect the approximate performance of Intel prod-
ucts as measured by those tests. Any difference in
system hardware or software design or configura-
tion may affect actual performance. Buyers should
consult other sources of information to evaluate the
performance of systems or components they are
considering purchasing. For more information on
performance tests and on the performance of Intel
products, visit http://www.intel.com/performance/
resources/limits.htm.

38 Sponsored by
Narrowband Powerline
Communication
Applications and Challenges Part I
By Bogdan Baraboi, Ariane Controls Inc
T
ransmitting data between remote devices
has a long history and is becoming sine
qua non in the modern world. Thanks to
technological evolution and new emerging mar-
kets, powerline communication (PLC) is today a
viable technology for both low speed and high
speed networking. Broadband PLC represents
an interesting alternative to cable, DSL or Wi-Fi
as a last-mile solution for Internet distribution
39
and home multimedia networking. For narrow-
band PLC, there is today a potentially vast mar-
ket of command and control applications.
This article intends to give an overview of the
state of the art and trends in narrowband PLC.
The topics include basic modulation principles,
characteristics of power lines, applicable regu-
lations, typical applications and implementa-
tion challenges. Also, market perspectives and
Sponsored by
promising opportunities are discussed to developments date from end of the 19th ciations. The emerging PLC technologies
assess the viability of the narrowband PLC century. In the beginning of the 20th cen- became promising for both consumers
in the Smart Energy world. tury, carrier frequency systems began to and energy providers.
operate over high-voltage lines, then over
No new wires medium and low voltage distribution sys- Broad vs. Narrow
The key advantage of PLC is the use of tems for telemetry purposes. We distinguish today between two classes
existing electrical lines as communication The term Power Line Carriers emerged of PLC systems: narrowband and broad-
medium, which provides the major benefit in the 1940s. In the 1970s and 1980s band. Usually, narrowband PLC (NB-PLC)
of eliminating considerable costs of install- home-control PLC devices became com- refers to low bandwidth communication,
ing networking infrastructure, like dedicat- mercially available, and at the same time utilizing the frequency band below 500kHz
ed cables or antennas. the first standards were developed. The and providing data rates of tens of kpbs.
Data is sent on the power lines by super- interest increased during the 1990s, with Broadband PLC (BPL) utilizes a much
posing a modulated high-frequency carrier popular technologies like X-10, CEBus wider frequency band, typically between
signal on the line voltage, being high, me- and LonWorks. The drive was to produce 2MHz and 30MHz, and allows for data
dium or low, AC or DC. The carrier signal a reliable system which is cheap enough rates of hundreds of Mbps.
is then decoupled and demodulated at the to be widely installed and able to com- BPL technologies are used today for
receiving end to recover the information. pete cost effectively with wireless solu- high-speed data transfer applications like
tions. Over the last 20 years, research Internet, HDTV, and audio. The optimism in
Once upon a time in PLC has intensified; new modulation this market is reflected by a number of so-
Powerline communications have a long and error control coding techniques were lutions from companies like Atheros, Max-
history; actually, it is almost as old as the proposed, as well as new standards from im, Sigma, all in the Home Area Network-
power grid itself. The first research and industry alliances and professional asso- ing. The NB-PLC finds many applications

40 Sponsored by
 Narrowband PLC Broadband PLC
based on single-carrier modulation, while
broadband technologies use multi-carrier
Data rate Up to 200kbps Over 1Mbps
techniques. Lets look at the features of
Frequency Up to 500kHz Over 2MHz the main types of modulation used in nar-
rowband PLC systems.
Modulation FSK, S-FSK, BPSK, SS, OFDM OFDM

Building Automation
Renewable Energy Internet Single-carrier modulation
Advanced Metering HDTV
Applications
Street Lighting Audio Single-carrier modulations, like Ampli-
Electric Vehicle Gaming tude-Shift Keying (ASK), Frequency-Shift
Smart Grid
Keying (FSK) and Phase-Shift Keying
Ariane Controls
Atheros
Cyrpress
Broadcom (PSK), use a number of distinct signals
Echelon
Lantiq to represent digital data. With these
Providers Maxim
Marvell
ST Microelectronics
Texas Instruments
Maxim schemes, data is transmitted by chang-
Sigma
Yitran
ing, or modulating, a characteristic of a
reference signal, for example the ampli-
in the command, control and monitoring ing Echelon, Maxim, ST Microelectronics, tude, the frequency or the phase. In the
markets, like smart building automation, Texas Instruments and Ariane Controls. simplest form of these modulations, two
renewable energy generation, advanced discrete values of these parameters are
metering, street lighting, plug-in electric ve- Modulation schemes used to represent binary data. For ex-
hicles, etc. Consequently, the narrowband Different modulation schemes are used ample, Binary Frequency-Shift Keying
PLC market is seeing healthy competition, by the two categories of PLC technolo- (BFSK) uses a pair of discrete frequen-
with a large number of suppliers, includ- gies. Narrowband systems are mostly cies to transmit the 1 (called the mark

41 Sponsored by
frequency) and the 0 (called the space
frequency).
The performance of single-carrier PLC
technologies has been proven in many
implementations. X-10 is based on On-Off
Keying modulation, the simplest form of
ASK that represents data as the presence
or absence of a carrier wave, synchronized
with the zero-crossing of the power line
voltage. Echelons transceivers use BPSK,
STs ST7537 to ST7540 transceivers are
based on BFSK, as well as Arianes PLM-1
modem. These solutions have the advan-
tage of providing reliable communication
at reduced power consumption and low
cost. The low data rate, typically limited to
10kbps, makes these solutions appropri-
ate for command and control, metering
and monitoring applications.
The main drawback of single-carrier PLC
systems is their sensitivity to narrowband
noise and signal distortion. The robustness
42
can be increased by error detection and cor- Multi-carrier modulation
rection mechanisms, which combined with
message repetition can be very efficient.
Spread-spectrum modulation
Another modulation scheme used in NB-
PLC systems is spread-spectrum (SS).
This technique operates by spreading the
original narrowband information over a
wider band of frequencies. Consequently,
the transmitted signal occupies a band-
width much greater than the minimum
necessary to send the information. In this
way, SS modulation allows overcoming
frequency-selective perturbations and
burst noises from the power lines. How-
ever, the redundancy in data transmission bit rate transmission by transmitting dif-
also makes the spectral efficiency low.
There are few implementations of SS
techniques in PLC systems, the most suc-
cessful being Yitrans DCSK (Differential
Code Shift Keying) modems.
While mainly used in broadband PLC sys-
tems, multi-carrier modulation schemes
have been recently applied in narrowband
PLC. One of the most popular multi-carrier
modulation techniques is OFDM (Orthogo-
nal Frequency-Division Multiplexing). Sev-
eral OFDM-based NB-PLC solutions that
address mainly the smart grid market have
been proposed by industrial alliances like
PRIME and G3-PLC.
With OFDM, data is split into sub-carri-
ers of different frequencies, modulated in
parallel with conventional techniques, like
BPSK, QPSK or QAM. Hence, a low bit
rate transmission is converted into a high
ferent parts of the data on many different
sub-carriers.
The main advantage of multi-carrier
modulation over single-carrier schemes
is its robustness to narrowband interfer-
Sponsored by
ence and frequency-selective perturba- The power grid electrical substations located near demand
tions. Also, multi-carrier systems allow Lets take a look at the power grid and its centers. Medium voltage lines (typically
higher data rates than single-carrier and characteristics as communication channel. below 50kV) and low voltage lines (less
SS modulation schemes. An important Electrical power lines are usually classified than 1kV) form the distribution network,
feature of OFDM is its adaptability to the into high, medium and low voltage lines. which carries electricity from the transmis-
channel conditions; sub-carriers within the High voltage lines form the transmission sion system and delivers it to consumers.
OFDM waveform can be selected in order network that transports energy at over Various wiring topologies exist and differ
to avoid transmitting at frequencies where 100kV from generating power plants to from country to country, and also within
the signal-to-noise ratio is too low.
However, these features come at a price. Transmission Grid
OFDM-based systems are more complex,
High Voltage Lines
meaning higher cost and increased power
vvv
consumption compared to single-carrier
vvv
Medium Voltage Lines
solutions. Also, because of their large fre-
quency bandwidth, the maximum effective
data rate is considerably limited by the re-
Distribution Grid vvv
vvv vvv
vvv vvv
vvv vvv
vvv Substations
strictions in narrowband frequency bands
- to about 30kbps in European CENEL- Low Voltage Lines
EC bands (frequencies up to 148.5kHz)
and about 128kbps at frequencies up to
500kHz. Solar Wind Large Small Commercial MDU Homes
Farms Farms Industry Industy Buildings

43 Sponsored by
each country, for example radial and inter- lenging environment for communication. for example) can also cause extremely
connected distribution networks, delta and The transmission grid and even more the low impedance values. The impedance
wye three-phase services, single-phase distribution grid have the characteristics of is highly varying with frequency, but also
and split-phase low-voltage services, dif- a very complicated wired network, which changing in time, as devices are contin-
ferent earthing systems. vary with time, location and are frequency- uously plugged in or out, and switched
Powerline communications have been dependent. High signal attenuation and on or off
used at all levels of the power grid; on high considerable noise are the main factors Multiple phases - In residential split-
voltage lines for telemetry, protection and that affect the communication perfor- phase power lines and commercial or
control by utility companies; on medium mance. Also, severe electromagnetic com- industrial three-phase systems, PLC
voltage for advanced metering and grid patibility restrictions impose non-negligible transceivers can be connected on dif-
optimization by electricity providers, but challenges to PLC systems. ferent phases. The PLC signals that
also for high-speed Internet distribution; are typically transmitted between one
finally, the vast majority of PLC applica- Signal attenuation phase and neutral have to travel from
tions are on low voltage lines, both indoorThere are many sources of attenuation one phase to the other through the
and outdoor (between transformers and and distortion of PLC signals on the power distribution transformer or some phase-
buildings), for both low-speed command lines. Here is a summary of the most com- to-phase loads, and are generally highly
and control applications and high-speed mon factors: attenuated.
networking. Impedance of loads connected to low Transformers - Power distribution trans-
voltage power line - In most frequency formers cause high signal attenuation
PLC challenges ranges, the load impedance shows when PLC devices are located on pri-
The power lines were not designed for data inductive or capacitive behavior, but mary and secondary sides, for example
transmission and they constitute a chal- resistive loads (like heating elements, in communication between meters on

44 Sponsored by
 the low voltage line and a concentrator
connected on medium voltage side.
Line losses - The characteristic imped-
ance of the wiring between transmit and
receive locations can also add signifi-
cant signal attenuation, particularly in
outdoor long power lines (for example
in street lighting networks or metering
systems).
Multipath propagation - Signal propaga-
tion on multiple paths and signal reflec-
tion due to impedance mismatches in
branching points can generate attenua-
tion distortion of PLC signals. This phe-
nomenon is less observed at low com-
munication frequencies (below 150kHz),
since the signal wavelength is quite long
(over 1km).
Measurements carried out in indoor and
outdoor environments reported in the
literature have shown that the signal at-
tenuation can range from 15dBs to 60dB
45
or more. While the attenuation depends
on the network topology, location of PLC
transceivers and connected loads, it is also
time- and frequency-dependent. Attenua-
tion variations of 20dBs over time are not
unusual. Also, it has been observed that
signal attenuation tends to increase with
frequency, although the increase is not
always monotonic.
Noise
When the PLC signals are severely attenu-
ated, the noise becomes a significant con-
cern for data reception. Generally, channel
noise is generated by electrical loads and
varies with time of day, location and fre-
quency. The noise power level at a certain
location is the sum of noise waveforms
from different sources and depends on the
distance to noise sources. Also, noise has
greater power in the lower frequency range
and decreases at higher frequency, which
is particularly challenging for narrowband
PLC operating below 150kHz.
As one would expect, there are many
sources of noise on the power lines. They
can be roughly classified in several classes:
Narrowband noise, that is mainly due to
amplitude modulated signals (for exam-
ple from broadcast stations).
Continuous background noise, which
can be:
Time-invariant, having constant en-
velope for a long period (for example,
thermal noise generated by internal
circuitry).
Time-variant, whose envelope changes
with line voltage (for example, noise
from inverter-drive fluorescent lamps).
Impulsive noise, consisting of abrupt
impulses with short duration and high
amplitude. There are at least three cat-
egories of impulsive noise:
Impulses synchronous with AC line volt-
Sponsored by
 age, at the same frequency or double
(for example, noise generated by light
dimmers or brush motors).
Impulses asynchronous to AC line volt-
age, whose frequency is much higher
than that of mains voltage (for example,
noise from switching regulators).
Single-event impulses, that occur at ran-
dom times, less frequently, and that are
typically caused by switching operations
(for example, wall switches or capacitor
banks for power factor correction).
While noise is mostly generated by the
devices connected to the power line, one
possible source of noise that can disturb
the reception of low communication sig-
nals is PLC transceivers circuitry, mainly
its power supply. Switching-mode power
supplies (SMPS) and regulators can be
particularly problematic and their switch-
ing frequency should be carefully consid-
46
ered. SMPS may couple significant noise These normative specifications are gen-
at the switching frequency or its harmon- erally produced by regulatory agencies,
ics back onto the mains power line. Also, independent or governmental, and apply
a SMPS or a DC/DC switching regulator for specific geographical areas. For exam-
with improperly filtered output can pro- ple, Federal Communications Commission
duce noise and ripple on the transceiver (FCC) in US, Industry Canada, European
DC supply voltages. Committee for Electrotechnical Standard-
ization (CENELEC) in the European Union,
EMC regulations Association of Radio Industries and Busi-
Besides the harsh channel conditions, nesses (ARIB) in Japan.
another challenge for PLC is the compli- The above diagram summarizes the allo-
ance with the standards and regulations for cation of frequency bands for narrowband
electromagnetic compatibility. These stan- PLC by the main international standards.
dards do not define particular modulation We can observe that while US FCC Part
methods, but typically specifies: 15, Industry Canada ICES-006 and Ja-
Frequency bands allocated for differ- pan ARIB provide larger frequency range,
ent applications the European CENELEC standard is more
Limits for the transmitted signals in the restrictive, with a frequency band from
operating bands 3kHz to 95kHz reserved for utility applica-
Limits for electromagnetic emissions tions and only from 95kHz to 148.5kHz for
(conducted and radiated). consumer applications. Also, the maximum
Sponsored by
signal levels and electromagnetic emis-
sions accepted by CENELEC are tighter.

About the Author

Bogdan Baraboi is an Application Engineer
at Ariane Controls, a leading developer
of narrowband powerline communica-
tion solutions. He has a Master degree in
Electrical Engineering from Universit La-
val (Qubec, Canada) and over 10 years
of experience in electronics, analog signal
processing, power systems and powerline
communications. He can be reached at:
bbaraboi@arianecontrols.com. Allocation of frequency bands for narrowband PLC

47 Sponsored by
Narrowband Powerline
Communication
Applications and Challenges Part II
By Bogdan Baraboi, Ariane Controls Inc
PLC Transceivers
Whatever the type of modulation used, a PLC
transceiver includes a few basic blocks. Differ-
ent solutions use various levels of integration
of these components, from full-digital modems
with external discrete components to highly in-
tegrated systems-on-chip.
The core of each transceiver is the modem
itself, which usually implements the Physi-
48
cal and Data Link layers, according to the OSI
model. Modulation-demodulation, error correc-
tion, media access control are among the major
functions performed at this level. A processing
unit is responsible for networking, protocol, and
application-specific functions. At the other side,
transmitted and received signals are processed,
in digital and analog form; the main operations
include filtering and amplification. The interface
Sponsored by
 tion with low signal-to-noise ratio. Media
Signal
MCU Modem Processing Coupling Circuit access control and adjustable communica-
Protocol PHY DAC/ADC Capacitive/Inductive Power Line tion frequency allow sharing the medium
Networking MAC Filtering Proctection AC/DC
Application Amplification Isolation and adapting to channel conditions.

Narrowband PLC: Where and how

NB-PLC finds applications wherever there
Power Supply
is electrical wiring, from utility power grid
Linear/ Switched-mode/ Capacitive
and distributed renewable energy systems
to homes and buildings, public lighting,
Figure 1: Typical block-diagram of PLC transceiver
and plug-in electric vehicles. There is also a
large variety of custom applications involv-
with the power lines is made via a capaci- with the channel electrical impairments ing AC, DC or un-powered lines, like fire-
tive or inductive coupling circuit, which and must meet the regional regulations for works control, emergency lighting, dive data
also provides galvanic isolation and pro- EMC. In most power line installations, high management, submersible water pumps
tection against line voltage disturbances. transmission power is required so that the to mention just a few. Each implementation
The design of the analog front-end is re- transmitter be able to drive an adequate has its own challenges that mainly depend
lated to the channel characteristics, such amount of signal into low impedance loads. on the channel characteristics and market
as amplitude and frequency of line voltage, Excellent receiving sensitivity is necessary requirements. Well review the main areas
wiring style, location, potential disturbanc- in order to receive highly attenuated signals. of application for NB-PLC and well discuss
es, applicable regulations, etc. High noise immunity, error detection and typical implementation challenges and spe-
PLC transceivers must be able to cope correction mechanisms enable data recep- cific solutions to consider.

49 Sponsored by
Home and building automation
Home and building automation is one of
the primaries markets for narrowband PLC.
By eliminating the need to install new wir-
ing, PLC transceivers allow easily creating
smart automation systems in homes, ho-
tels, offices, commercial and institutional
buildings. Energy savings, greater comfort
and safety are only a few benefits. Typical
implementations include lighting control,
load shedding, heating and cooling sys-
tems, energy monitoring, fire detection, etc.
The main PLC challenges in home and
building environment are the high density
of loads, which produce both signal attenu-
ation and noise, and split-phase or three-
phase wiring topology, which results in addi-
tional signal loss. Typical solutions include:
Line filters. Built into PLC devices or
as stand-alone equipment, they help
preventing noise and signal attenua-
tion from critical appliances (for exam-
50
ple, High Pressure Discharge lighting,
fluorescent light electronic ballasts,
refrigerators, computers, etc.).
Phase-couplers. Connected near the
breaker panel, they are usually used
to allow PLC signals travel between
phases.
Repeaters. Typically associated with
phase-couplers, these devices are
helpful in large locations. A coupler-
repeater detects data signals upon
one phase and repeats them on all
phases of the power line system.
Clean energy management
A more recent market for NB-PLC is in the
distributed renewable energy systems.
Generating electricity from many small
energy sources is a modern alternative to
traditional large generation plants. The use
of renewable distributed energy is grow-
ing rapidly, as it allows reducing distribu-
tion losses, improving security of supply,
lowering pollution, and cutting customer
electric bills. Data transfer between energy
production devices (like solar panels or
wind turbines), inverters and gateways is
required for the management of distributed
generation systems and for reliable inte-
gration to the power grid.
There are two approaches for managing
solar or wind energy today:
1. Using large single inverters that convert
the DC voltage from several PV solar
panels or wind turbines into AC energy.
2. Using small micro-inverters that per-
form the DC to AC conversion at each
individual source.
In single inverter topologies, the PLC
signals are sent on the DC line between
individual sources and inverter. The main
challenge here is typical high attenuation
from inverters input capacitors. A possible
solution is the use of inductive coupling
Sponsored by
instead of capacitive coupling. Inductive
couplers are based on specialized current
transformers that are connected in series
on the communication channel. They allow
injecting a PLC current signal instead of a
voltage on the line, and avoiding the signal
attenuation caused by low parallel imped-
ance of the inverters capacitors.
In micro-inverter topologies, the com-
munication is performed on the AC line
between each micro-inverter and in-home
gateways or displays. The challenges here
are typical to indoor power lines, i.e. mul-
tiple phases and possible high density of
loads. Consequently, similar techniques to
those find in home and building automa-
tion applications can be applied, includ-
ing line filters, phase couplers, and signal
repeaters in larger locations.
Public lighting
Municipalities are more and more con-
51
cerned to manage their public lighting and
are looking for solutions to save energy,
reduce maintenance costs, while providing
more reliable and safer lighting. By add-
ing intelligent control of each individual
lamp, the street lights can be turned off or
dimmed as optimally as possible depend-
ing on time, traffic or weather. Lamps that
reach the end of their life cycle can be
replaced before they fail. The energy con-
sumption can be measured and optimized
by accommodating new types of lamps.
NB-PLC offers an efficient solution for
remote management of street lighting,
tunnel lighting, parking lot lighting and
traffic light controllers. The main PLC
challenge are the long power lines (usu-
ally several kilometers) and particularly
low access impedance (mainly due to the
large capacitors that are typically used for
power factor correction inside the street
lamps). Typical techniques to cope with
these challenges include increasing the
PLC transmitting power and using signal
repeaters at specific intervals.
Advanced metering
Powerline communication has long been
a favorite for utilities since it requires less
new infrastructure than other technologies.
Using NB-PLC, electricity providers can
remotely monitor and control energy usage
at the customer site. Smart meters can be
linked to concentrators on low or medium
voltage grid, allowing suppliers to have re-
mote access to each individual customer,
for transmitting information such as rates,
pre-paid amounts, current and cumula-
tive counts, etc. Utilities can also develop
Demand-Response mechanisms to man-
age customer consumption of electricity in
response to supply conditions. Operational
parameters can be monitored, including
tamper or power factor.
Sponsored by
 Obviously, the challenges for PLC are devices and utilities. Benefits associated scale renewable energy systems.
proportional to the system complexity: with the emerging Smart Grid include The omnipresence of the power lines
long power lines, distribution transformers, more efficient energy infrastructure, more renders NB-PLC a preferred solution for
low access impedance, split- and 3-phase reliable supply of electricity, reduced vul- data communication in many applications
power line topologies. This means severe nerability to natural disasters or attacks, in this market. However, considering the
signal attenuation and calls for appropri- easy integration of large-scale and low- system complexity and diversity of devic-
ate improving techniques, including phase
couplers, signal repeaters, increased trans- Renewable
Energy
mitting power, transformer bypass, etc.
Power Plants
Industrial
Market trends: Smart Grid Microgrids
The advanced metering needs to be re-
Distributed Distributed
garded today as part of the broader con- Automation Generation
Connected
cept of Smart Grid. Todays AC power Cities Energy
Management
grid is mostly based on centralized en-
Vehicle-to-Grid
ergy generation in large remote power
plants and unidirectional flow of energy Smart Metering
to the consumer. The Smart Grid repre- Vehicle-to-Home

sents an opportunity for modernization of

generation, distribution and utilization of Plug-in
Electric Vehicles
electrical energy based on bidirectional
Intelligent
communication between consumers, Homes

Smart Grid building blocks

52 Sponsored by
es, the Smart Grid requires a mix of tech-
nologies operating together in a meshed
network, including PLC, Ethernet, Wi-Fi,
ZigBee, GPRS, etc. The specific commu-
nications media over which they exchange
information is less important, as long as it
fulfills the conditions specific to each utili-
zation, such as reliability, throughput, elec-
tromagnetic emissions, security, etc.
A fundamental challenge in the Smart
Grid is to ensure balance of generation
and demand while integrating a large vari-
ety of technologies. This is based on two
requirements for all wired and wireless
communications:
Coexistence, implying that different
communication services share the me-
dium without affecting each others op-
eration. Coexistence is accomplished
by assigning specific frequency range,
associated bandwidth and output pow-
er for the data to be transmitted.
53
Inter-operability, meaning that devices munications over low voltage line. Also,
using various technologies can ex- companies forming industrial alliances
change information directly. The Smart develop specifications that aim to promote
Energy Profile 2.0 is an emerging stan- different PLC technologies for the same
dard that is being developed to create markets (e.g., PRIME, G3-PLC, Home-
an interoperable protocol that con- Plug Green PHY, Netricity PLC ). While this
nects devices to the Smart Grid. The standardization activity proves the increas-
standard itself is designed to run over ing interest in NB-PLC, it remains to be
the Internet protocol (TCP/IP) and is seen how these solutions will be able to
media access control (MAC) and phys- deal with co-existence and inter-operabili-
ical layer (PHY) agnostic. ty in the Smart Grid market. p
At the same time, one can observe a
lively activity in the development of nar-
rowband PLC standards that propose
various physical layer and medium access
solutions. International organizations like
IEEE and ITU-T create formal standards
(IEEE 1901.2, ITU-T G.hn) that try to as-
sure efficient use of the power line chan-
nel defining detailed mechanisms for
coexistence between different NB-PLC
technologies for indoor and outdoor com-
Sponsored by
An overview of
ZigBees Smart Energy
Profile 2.0 standard
By Srinath Balaraman and Anil Khanna, Mentor Graphics
T
he number of devices defined as part of
the Internet of Things continues to grow
quickly as individuals and industries alike
find innovative ways to utilize connected devic-
es and networks. Enabling machine-to-machine
(M2M) communication has endless possibili-
ties, and is one of the more prominent technol-
ogy areas in the emergence of smart energy. As
home meters, personal devices and appliances
54
begin to connect to each other, a bigger and far
more comprehensive picture emerges on how
to make smarter energy consumption decisions.
Connecting devices in the home and local Inter-
net networks to a smart grid enables two-way
communication between homeowner and the
power company and is becoming more of a real-
ity with each passing day.
This article provides an overview of the Smart
Sponsored by
Energy Profile 2.0 (SEP), an emerging stan- and intelligence which leads us towards the
dard in the smart energy market, being vision of a smart networked world.
developed by the ZigBee Alliance. With One of the applications of this set of tech-
the basic understanding of SEP 2.0, soft- nologies is to improve energy consumption,
ware developers will be better prepared to also termed smart energy. The concept be-
select embedded software suited for the hind smart energy is controlling energy use
development of smart energy applications. internally, within the home, and externally
from the home to outside connected devic-
A new frontier with the Internet of Things es, networks, and the smart grid itselfall Figure 1: Energy providers, the development of a nationwide smart
The Internet of Things is a popular buzz with the goal of optimizing energy produc- grid and energy-conscious consumers will usher in a new era of
smart energy use
phrase and one that conjures up exciting tion, distribution, and usage. Bi-directional
visions of the future a future where your communication between home networks 2). There are several products (smart ther-
refrigerator self-checks its contents and and the power grid opens up possibilitwies mostats, smart switches, smart refrigerator,
emails you a grocery shopping list just as for improved reliability and sustainability. and more), which are available today that
you are about to leave work (not that gro- Smart grids and smart homes (smart ap- offer some level of intelligence and wireless
cery shopping is a terribly exciting experi- pliances, gateways etc.) and smart meters connectivity. Some of the more advanced
ence), your house gets ready for your arrival (electricity, gas, water) are key elements of appliances include built-in Web servers to
by adjusting the temperature to an optimum the smart energy ecosystem. Smart home interact with other devices in the connected
level, and the oven can pre-heat for dinner. appliances are typically those devices home. Smart meters are the gateways
The rapid proliferation of devices embed- consumers interact with daily. By enabling into these homes (and offices) and collect
ded with a combination of powerful mi- these devices to talk to each other and be and measure resource usage before shar-
croprocessors, sensors, and wireless con- controllable by the consumer, a whole new ing some or all of this information with the
nectivity has resulted in more functionality dimension of convenience is added (Figure smart grid. The grid, in turn, acts upon this

55 Sponsored by
information by taking necessary steps such
as load adjustment, peak curtailment, and
even demand-side management.
Smart energy devices, apart from per-
forming their standard functions, must be
able to communicate with other smart en-
ergy devices within the local network and
be able to send and receive relevant in-
formation (pricing, usage, alerts, etc.). The
exchange of data not only improves the
overall efficiency and fault tolerance but op-
timizes the consumption of energy. Smart
meters collect and transmit usage data to
the energy providers and allow consumers
the ability to monitor and manage their own
energy consumption. In other words, usage
data flows from the consumer to the energy
provider and, at the same time, pricing data
flows from the energy provider to the con-
sumer. This bi-directional flow of informa-
tion allows consumers to make decisions
to manage consumption. This two-way,
real-time communication enables energy
56
providers to improve planning and improve
energy distribution.
Standardizing smart energy design
As multiple manufacturers design smart
energy systems, it is becoming increas-
ingly clear that all devices interoperate in
a network. The ZigBee Alliance is working
on a specification called the Smart En-
ergy Profile 2.0 (SEP 2.0) to help formalize
the requirements for many aspects of the
smart energy ecosystem including device
communication, connectivity and informa-
tion sharing requirements.
SEP 2.0 provides the guidelines in which
the devices should communicate with one
another. It defines various device properties
that can be manipulated. These properties
(also known as resources) work together
in logical groups to implement SEP 2.0
functionalities (called the function sets).
A metering system, or pricing system, is an
example of an application-specific func-
Figure 2: In a typical smart home, devices such as a washing
machine, an in-home display, and a power meter all work
together in tandem to make the home and grid smarter
tion set. Devices like smart meters imple-
ment one or more function sets to provide
value-added services such as usage sta-
tistics and trends. These pricing statistics
and trends can then be used by either the
energy provider or the consumer to further
manage services or usage, respectively.
Function sets and their resources on a
device are accessed through HTTP URLs.
These devices dynamically discover rel-
evant services on the network using
Sponsored by
technologies like mDNS and DNS-SD
and register themselves to further access
resources to implement SEP 2.0 function-
ality. To provide for a truly interoperable
ecosystem of interconnected smart energy
devices, use of TCP/UDP and IP-based
networking is necessary. Support for se-
curity features within a device is critical
because of vulnerabilities from exposure to
a broader network, and more importantly,
the access a device provides to the energy
grid. Since many smart devices serve up
continuous, reliable, and real-time data,
they must be Always-ON and Connect- 32-bit microcontrollers (MCUs) provide
ed which necessitates that all smart ener-
gy devices be power efficient themselves.
Lastly, they must also support both wired
and wireless networking capabilities.
A majority of existing home appliance de-
vices are not built to support the advanced
features of the emerging M2M wave, so
incorporating many different capabilities
into a single device means significant and
57
expensive hardware upgrades, resulting
in an increased bill of materials and cost.
Manufacturers must balance the benefits of
delivering smart energy-enabled appliances
with the additional cost.
Moving forward, appliance manufactur-
ers have more options to find cost effec-
tive solutions to design home appliances
that are smart energy enabled. The choice
of SoC hardware for these home appli-
ance devices should be made by striking
the right balance between functionality,
form factor, software support, and cost.
ful architecture, XML, and EXI encoding
a unique blend of processing power,
memory, and connectivity to make them a
strong candidate. The current generation
of microcontrollers, such as Freescale
Kinetis, STMicroelectronics STM32, or
the TI Stellaris (ARM Cortex-M core), offer
tremendous amounts of features and ca-
pabilities at a very compelling price point.
Selecting the right hardware is only the
beginning. The differentiator in this equa-
tion is the software choice.
The software technology requirements
laid out by the SEP 2.0 specifications
include: a rich TCP/IP stack with UDP
support; IPv6 services with dynamic ser-
vice discovery capabilities like mDNS
and DNS-SD; and a HTTP implementa-
tion with support for primitives like GET,
PUT, POST, and DELETE. SEP 2.0 also
mandates support for security implemen-
tations like SSL/TLS and several modern
day Internet technologies like the REST-
schemes etc. Such extensive support
for software technologies is readily avail-
able in Linux, but unfortunately, the use
of microcontrollers with RAM sizes in the
range of 96K to 128K eliminates Linux as
an option. Developing such technologies
in-house is expensive and time consum-
ing, which leads to the possible imple-
mentation of a real-time operating system
Sponsored by
(RTOS) for these devices.
RTOSes are not only fast, efficient and
robust, they typically include an extensive
networking stack, a solid support for secu-
rity using SSL or TLS, and most certainly
meet the heavily constrained footprint and
other memory requirements these devices
require. The Nucleus RTOS provided by
Mentor Graphics is an example of one
such solution (Figure 3). Nucleus is a wide-
ly deployed and scalable RTOS that meets
all smart grid device requirements. It has
both hard, real-time performance and inte-
grated power management services. Such
an RTOS can fit in a memory-constrained
MCU, yet still provide the large set of func-
tionality required by a connected, smart
grid device.
Conclusion
With the projected rapid growth in the
adoption of smart grid technologies, de-
signing fully compliant devices that keep
58
the bill of materials minimized will become
a major challenge for manufacturers. To
create a device compliant with the SEP 2.0
specification, a homegrown software de-
sign is likely not an option due to the high
number of functional requirements and the
expensive in-house development efforts. On
the other extreme, using a general purpose
operating system will result in unaccept-
able cost increases because of the need for
significantly upgraded hardware resources.
Device manufacturers need to find the right
balance when choosing both the software
design and hardware platform. The use of
a scalable, power-efficient, real-time op-
erating system with extensive networking
support (wired and wireless) along with
one of the 32-bit MCUs now available in
the marketis the closest to meeting all of
these requirements. Following this design
paradigm, designers will significantly reduce Figure 3: An example of a hardware design that can support a
their time-to-market and still realize all of wide range of peripherals, powered by a real-time operating
system such as Nucleus RTOS, which offers all the services
their smart grid application goals.p required by SEP 2.0
Sponsored by
SPONSORED CONTENT
Closing the Industrial
Communications GAP With
a Secure, Virtualized Solution
S
hanghai Baosight Software Co., Ltd. is one
of the largest system integrators serving the
industrial automation marketplace in the
Peoples Republic of China. Founded in 2001, Bao-
sight provides software solutions for Chinas metal-
lurgy, chemical, equipment manufacturing, mining,
and energy industries, with an expanding presence
in new areas such as the Internet of Things (IoT),
smart transportation, railways, and finance.
59
Challenge
Traditional GAP products for industrial commu-
nication are typically designed with two sets of
hardware systems, each with its own CPU and
operating system to execute a single task. One
system generally handles data acquisition from
an internal industrial network, while the second
system handles data transfer to an external net-
work. This model has two drawbacks:
Sponsored by
SPONSORED CONTENT
Security risk. Data communication
between the systems is via the TCP/
IP networking protocol, which has
the potential to expose an unguarded
network to cyber-attacks that could
corrupt the systems.
Cost, complexity, and footprint. Sys-
tem proliferation in an industrial envi-
ronment imposes a financial burden in
terms of hardware purchasing, system
support, and maintenance expense,
while also introducing layers of com-
plexity in terms of cabling, shielding,
and networking.
Solution
Develop a secure gateway product
that can:
Consolidate the workloads of the
two tasks into one CPU, and
Replace the conventional, less-
secure TCP/IP communication be-
60
tween the two systems with secure ed gateway solution and form the rationale
inter-core communication to elimi- behind Intels recent introduction of the
nate this potential point of risk and Intel Industrial Solutions System Consoli-
help thwart cyber-attacks. dation Series.
Implement the Intel Industrial Solutions The three trends are as follows:
System Consolidation Series, which Growing pressure for efficiency and
simplifies the task of using virtualization productivity requires manufacturers and
technology to combine multiple dis- IT departments to streamline, simplify,
crete workloads on a single computing and consolidate their complex array of
device. Intels solution integrates key systems to increase output while using
software and hardware components to fewer resources.
streamline the development of consoli- As factories streamline, traditionally iso-
dated, virtualized solutions. lated industrial networks and systems
are evolving into connected, intelligent,
Introduction Internet-enabled devices capable of
Three trends are converging in the industri- analyzing and harnessing data for busi-
al segment, transforming the factory floor ness valuethe Internet of Things
from a system of isolated, single-purpose (IoT). (See related sidebar.) Manage-
devices into an interconnected network ment is pushing for greater connectivity
of intelligent, efficient, multi-purpose ma- as this phenomenon transforms facto-
chines. These trends suggest opportunity ries and plants worldwide.
for Baosights new virtualized, consolidat- As these sensitive industrial environ-
Sponsored by
SPONSORED CONTENT
ments become connected to the cloud
and the Internet, the threat of cyber-
attacks is rising. The discovery of
Stuxnet, a computer worm designed
specifically to disrupt industrial pro-
cesses, has raised security concerns
and created demand for solutions that
can achieve both the connectivity that
management requires and the protec-
tion needed for these formerly isolated
networks.
Baosight Software identified the conver-
gence of these three trends as a business
opportunity. Company managers under-
stand that the need to keep industrial
infrastructure protected has never been
greater, and that the cost and complexity
of doing so is escalating. To minimize the
risk of cyber-attacks on industrial systems
in a cost-effective and simplified way,
Baosight harnessed Intels new virtual-
ized platform to develop a consolidated
61
An Integrated Device-to-Internet Solution for Securing the IoT
Baosight developed its iCentroGate-GAP* (ICG-
GAP) product as one of a series of comprehen-
sive solutions oriented to the Internet of Things
(IoT) and the smart city concept.
The IoT is erupting as the new reality in the
business world, ushering in the potential for
hundreds of billions of smart devices ultimately
coming to life in an Internet-connected cos-
mosfrom cars to refrigerators and from stop-
lights to factory robots. As companies imagine
and create intelligent devices that revolve and
communicate in this cosmos, the need to pro-
tect valuable business systems and sensitive
data will only grow more pressing.
Baosights gateway solution addresses the secu-
rity problem for the emerging IoT by minimizing
vulnerability to sensitive internal networks. By lay-
ering its application on top of the integrated Intel
Industrial Solutions System Consolidation Series
base platform, Baosight is enabling companies
to securely draw data out of a manufacturing
environment and onto other private and public
networks. This level of security is not possible in
traditional settings where factory operations have
long run in isolation on unprotected networks.
The ability to plug into the Internet securely and
extract data, and along with it, the ability to
make that data useful and actionable with mini-
mal risk, offers a compelling solution for indus-
trial companies that recognize the transforma-
tive business advantages of the IoT.
Sponsored by
SPONSORED CONTENT
industrial network isolation solution with
advanced levels of security.
Advancing Secure GAP With
Intel Technology
Baosights iCentroGate-GAP* (ICG-GAP)
product is designed to provide advanced
security and a stable, integrated, and
simplified system solution for industrial
applications in Chinas steel, metallurgy,
petro-chemical, manufacturing, and
transportation industries. This secure
gateway solution is ideally suited to in-
dustrial scenarios focused on data acqui-
sition, data processing, and intranet-to-
Internet data transfer where high security
is a crucial requirement.
Baosight implemented the Intel Industrial
Solutions System Consolidation Series in
its ICG-GAP solution, enabling the task of
data acquisition from a companys inter-
nal industrial network and the task of data
62
transfer to an external network to be instru-
mented in two separate operating systems
running on two cores of a single Intel
Core i7 processor. The solution relies on
secure partitions in the virtualized environ-
ment and is supported by the Wind River
Hypervisor and Intel vPro technology.
Consolidating two workloads on a single
processor differentiates Baosights solu-
tion from GAP products currently in the
marketplace and offers users the benefits
of saved hardware resources and their
related operating and maintenance costs.
Moreover, by providing a one-way data
communication path from an internal net-
work to an external network and replac-
ing conventional TCP/IP communication
with secure inter-core communication,
Baosights solution ensures task isola-
tion to help users avoid cyber-attacks and
achieve highly secure and accurate data
transmission. Replacing TCP/IP inter-sys-
tem communication with inter-core com-
munication also delivers advantages in
terms of system stability and performance
by means of strong computing capabil-
ity and greater communication bandwidth
between an intranet and the Internet.
Workload Consolidation
Through Virtualization
Using virtualization technology to combine
previously separate workloads on a single
multi-core hardware platform is a growing
trend in industrial automation today as an
effective way to cut costs and sharpen ef-
ficiency. The Intel Industrial Solutions Sys-
tem Consolidation Series, with its proven
virtualization technologies, facilitates this
effort by pre-integrating, validating, and
testing key software and hardware com-
ponents. For solution developers, this can
mean significantly less upfront engineering
cost, shorter product development cycles,
Sponsored by
SPONSORED CONTENT
and faster time to market.
According to Baosight, the savings
achieved by starting with Intels base
solution are significant. Dong Wensheng,
general manager of Baosights R&D Divi-
sion, says: Using the Intel Industrial
Solutions System Consolidation Series
product saved us an estimated 60 per-
cent of development time and 50 percent
of development cost compared to the
typical time and cost of developing a two-
system GAP solution. In other words,
according to Baosight, a consolidated
system that typically costs $1 million USD
to develop over a one-year design cycle
could be completed within six months for
a cost of approximately $500,000 USD,
considering all cost savings from devel-
opment, facilities, software, and engineer-
ing resources.
For end customers, systems based on
63
Intels integrated solution should trans-
late to less cost and complexity because
multiple functions are consolidated, re-
sulting in decreased operating expense,
system footprint, energy consumption and
integration and support effort.1 Baosight
notes that by consolidating two tasks on
one CPU, total hardware system costs are
reduced by as much as half. In addition, a
consolidated system helps improve over-
all solution efficiency because customers
need less energy and cooling power to run
it. Plus, a consolidated system often has a
better mean time between failures (MTBF)
than the combination of subsystems it
replaces because it has fewer components
that could potentially fail.
Powerful Software Meets
Proven Hardware
The Intel Industrial Solutions System
Consolidation Series at the heart of Bao-
sights new ICG-GAP product contains an
industrial computer with a power-efficient,
quad-core Intel Core i7 processor with In-
tel Virtualization Technology (Intel VT)2
and supports common I/O interfaces for
industrial applications.
It integrates a production-ready virtu-
alization software stack with three pre-
configured virtual machines (VMs), or
partitions, running a combination of real-
time and embedded operating systems
on dedicated processor cores. Solution
designers can use the two instances of
the Wind River VxWorks* RTOS to run
discrete applications with real-time per-
formance requirements, while simultane-
ously running embedded and/or gener-
al-purpose applications on Wind River
Linux* 5.0. Using a single multi-core plat-
form to run multiple applications results in
Sponsored by
SPONSORED CONTENT
a compact footprint and decreases hard-
ware size, weight, power, and cost.
Secure partitioning of the system is en-
abled by the Wind River Hypervisor, an
embedded virtual machine manager that
controls multiple applications and operat-
ing systems on a single multi-core plat-
form to ensure low latency, determinism
and real-time performance.
The virtualization software stack in the
Intel Industrial Solutions System Consoli-
dation Series allows users to have multiple
tasks implemented in parallel in different
operating systems in one CPU, Wensh-
eng says. Intel integrated this solution in
such a way that those partitions are secure
and the workloads can execute in a very
stable, protected, virtualized environment.
Our ability to deliver uni-directional data
communication for sensitive data flows
makes Intels solution the right choice for
our secure GAP product.
64
Collaboration Is Key
Engineers from Intel, Baosight, and a third
collaboration partnerShanghai SymTech
IT Co., Ltd., a,,vendor that contributed
various hardware and software compo-
nentsworked together to define,the
Baosight iCentroGate-GAP product. Intel
also provided technical support across all
hardware,and software levels to integrate
Baosights GAP application with the Intel
Industrial Solutions,System Consolidation
Series.
The result, Baosight says, is a hardware
device thats simpler in design, much more
integrated, and,more compact than exist-
ing products, and one that provides the
highest possible level of security. These
advances empower the company and its
products to be more competitive.
The Intel Industrial Solutions System
Consolidation Series is helping us provide
a unique and innovative solution to our
mainstream customers, which gives Bao-
sight a huge technology advantage, Wen-
sheng says. We have integrated our ap-
plications on top of an Intel-based solution
that provides integrated and fully validated
hardware and a full stack of software. By
starting with the Intel solution, our devel-
opment cost and time have been reduced
significantly, which puts Baosight and our
customers in a unique and favorable mar-
ket and business position.
Learn more at www.intel.com/industrial
consolidation
1
Results have been estimated based on internal Intel
analysis and are provided for informational purposes
only. Any difference in system hardware or software
design or configuration may affect actual results.
2
Intel Virtualization Technology requires a computer
system with an enabled Intel processor, BIOS and
virtual machine monitor (VMM). Functionality, per-
formance or other benefits will vary depending on
hardware and software configurations. Software ap-
plications may not be compatible with all operating
systems. Consult your PC manufacturer. For more
information, visit http://www.intel.com/go/virtualization
Sponsored by