1

State-of-the-art in Power Line Communications:

from the Applications to the Medium
Cristina Cano1 , Alberto Pittolo2 , David Malone3 , Lutz Lampe4 Andrea M. Tonello5 , Anand Dabak6 ,
1
Trinity College Dublin, Ireland, 2 University of Udine, Italy, 3 Maynooth University, Kildare, Ireland,
4
University of British Columbia, Vancouver, Canada, 5 University of Klagenfurt, Austria, 6 Texas Instruments Inc., Dallas, USA.

Abstract—In recent decades, power line communication has at- studied areas of applicability of PLC networks [11], [12], [13],
tracted considerable attention from the research community and [14].
arXiv:1602.09019v1 [cs.NI] 29 Feb 2016

industry, as well as from regulatory and standardization bodies. Applications that provide a means for communication in
In this article we provide an overview of both narrowband and
broadband systems, covering potential applications, regulatory transport systems where an electrical deployment is already in
and standardization efforts and recent research advancements in place can also take advantage of PLC networking [15], [16,
channel characterization, physical layer performance, medium Ch. 10]. In this context, PLC networks have been explored
access and higher layer specifications and evaluations. We also for use in in-vehicle communications [17], [18], naval [19]
identify areas of current and further study that will enable the and aircraft systems [20], as well as in trains [21].
continued success of power line communication technology.
However, the applicability of PLC networks is not restrained
Index Terms—Power Line Communication, narrowband, to these scenarios. A range of novel applications have been
broadband, smart grid, in-home, channel characterization, proposed for PLC networks including robotics [22], authenti-
medium access control.
cation [23], security systems in mining [24], as well as uses
within inductive coupling [25], contactless communication
I. I NTRODUCTION [26] and wireless power transfer [27].
Given the wide range of applications for which PLC
The use of electrical wires to provide data transmission networks can prove useful and the number of associated
capabilities, known as Power Line Communication (PLC), has challenges, PLC has gathered substantial attention from the
recently experienced increased deployment. Chip manufactur- research community as well as industry and has fostered a
ers of PLC devices for in-home and for smart grid applications range of regulatory and standardization efforts. In this article,
report that they are shipping millions of such devices each year we provide a comprehensive overview of the regulation and
and expect the number to continue to grow in the future. standardization processes, we summarize the different research
PLC networks provide a number of advantages that make questions that have been studied (from the physical layer to
them both a useful complement and a strong competitor higher layers in the stack) and we outline important future
to wireless networking solutions. The main appeal of PLC research directions, for both narrowband (NB) and broadband
networks is their low deployment cost when an electrical (BB) systems.
wired infrastructure is already in place. In addition, PLC The remainder of this article is organized as follows. In
networks allow communication through obstacles that com- Section II we give an overview of the regulatory and standard-
monly degrade wireless signals, while delivering high data- ization activities, showing the history of how current standards
rates. Moreover, PLC also provides a low-cost alternative to arose and their implementation. In Section III we describe
complement existing technologies when aiming for ubiquitous the pertinent characteristics of the PLC channel, including
coverage. For instance, as a backhaul for wireless sensor single-input-single-output (SISO) and multiple-input-multiple-
networks or small cells. output (MIMO) channels, modeling of the channel response,
Since one of the main advantages of using PLC networks line impedance and noise properties. In Section IV we sum-
is the possibility of re-using the existing wired electrical marize the research effort assessing the performance of the
network to provide communication capabilities, the smart grid physical layer, outlining main results and describing potential
remains one of the most appealing applications of PLC and improvements. In Section V we describe the MAC protocols
consequently the research carried out in this area is vast. Some defined in the standards and suggest future research directions
feasibility and experimental studies include the works in [1], and areas that are not deeply explored at present. We conclude
[2], [3], [4], [5], [6], [7]. In the same line, smart city [8], the article with some final remarks in Section VI.
in-home automation [9] and telemetry [10] applications can
naturally benefit from the fact that new cabling is not required II. R EGULATION , S TANDARDIZATION ACTIVITIES AND
and that wireless propagation issues are avoided. I NDUSTRIAL S OLUTIONS
The high data-rates that can currently be achieved with PLC The diversity of grid and application domains to which
— comparable with WiFi and domestic Ethernet — make it PLC can be applied has naturally led to a large ecosystem of
suitable for in-home multimedia applications. These scenarios, specifications, many of which have been adopted by standards-
along with the smart grid cases, correspond to one of the most developing organizations (SDOs). Regulatory activities are
 2

TABLE I
essentially concerned with coexistence with other systems that E MISSION STANDARDS AND REGULATIONS FOR NB PLC (< 500 K H Z )
also use the power grid (i.e. machines and appliances that AND BB PLC IN DIFFERENT REGIONS OF THE WORLD .
draw electricity) and wireless systems operating in the same Region Standard/ Remarks
frequency bands as PLC. The frequency range used for today’s Regulation
PLC solutions starts as low as 125 Hz and reaches as high as Narrowband PLC
100 MHz. A useful classification of PLC systems according Europe EN 50065 CENELEC A band for utility use
3–148.5 kHz CENELEC B-D bands for consumer use
to frequency bands has been introduced in [28]: it distin- CSMA/CA in CENELEC C band
guishes between ultra-narrowband (UNB), narrowband (NB) IEEE 1901.2 Not a European Harmonized Standard
and broadband (BB) PLC systems, operating between about 148.5–500 kHz
USA 47 CFR §15 Rules for power line carrier
125–3000 Hz, 3–500 kHz and 1.8–100 MHz, respectively. 9–490/500 kHz or carrier current systems apply
Most recent developments in standardization and regulation Japan ARIB STD T-84 CSMA/CA required
activities over the past 20 or so years apply to NB and BB 10–450 kHz
PLC systems, and we will focus on these in the following. Broadband PLC
Europe EN 50561-1 Dynamic power control
1.6065–30 MHz Static and dynamic notching
USA 47 CFR §15 Subpart G for access BB PLC
A. Regulation Activities 1.705–80 MHz Interference mitigation and avoidance
As for any electric load that is connected to the power grid, Excluded bands and zones
PLC systems are subject to regulations that limit the strength
of the signals coupled into power lines. In most cases, it is
desirable that the PLC signal is fully contained within the §15.113 are relevant. This paragraph permits power utilities
proximity of the power line. However, since the power grid has to use PLC in the 9–490 kHz band “on an unprotected, non-
not been designed to conduct relatively high-frequency signals, interference basis”. There is one caveat though, in that these
electromagnetic radiation occurs (e.g. [29]). This is mostly specifications do not apply to “electric lines which connect
relevant for BB PLC systems whose signals often have short the distribution substation to the customer or house wiring”
wavelengths compared to the length of the power lines. Hence, (§15.113(f)). Hence, PLC for many smart grid applications
the relevant regulatory constraints are generally different for involving, for example, smart meters would not fall under this
NB and BB PLC systems. provision. For such cases, one has to consider limits for what
1) NB PLC: Europe is a very active market for PLC is defined as “carrier current system” in §15.3(f). Accordingly,
equipment. The regulation and market access principles for paragraph §15.107(c) declares that there are only out-of-band
PLC devices as telecommunication equipment are discussed in conducted emission limits, to protect the 535–1705 kHz band.
[30], [16, Ch. 3]. Under this framework, European harmonized Furthermore, via paragraphs §15.109(e) and §15.209(a), in-
standards are an accepted approach for product compliance band radiated emission limits are specified for the frequency
test. An important such standard is the European Norm range from 9 kHz to 490 kHz.
(EN) 50065, a complete version of which was first issued by Another regulatory document that has been considered in
CENELEC in 1992 [31]. The EN distinguishes four frequency system specifications is the Standard T84 [35] by the Japanese
bands, which are commonly referred to as CENELEC-A (3– Association of Radio Industries and Businesses (ARIB). This
95 kHz), CENELEC-B (95–125 kHz), CENELEC-C (125– permits the use of PLC in the 10–450 kHz band. Table I
140 kHz) and CENELEC-D (140–148.5 kHz) respectively. It summarizes the emission standards and relevant regulations
specifies in-band and out-of-band emission limits in terms of from the discussed three regions for NB PLC.
maximum voltage levels together with the measurement pro- 2) BB PLC: For BB PLC, radiated emissions become a
cedures. Besides band-specific limits, the standard mandates bigger concern due to the higher signal frequencies and the
that the CENELEC-A band is reserved for power utilities and asymmetries in power line networks. In Europe, the specifica-
that the CENELEC B-D bands can only be used by con- tion of harmonized emission limits has been complicated by
sumer installations. Moreover, it specifies a mandatory carrier- the fact that relevant standards differentiate between a mains
sense multiple-access with collision avoidance (CSMA/CA) port and a telecommunication port of the tested equipment,
mechanism for the CENELEC-C band. EN 50065 has been and thus do not account for an intentional symmetric signal
decisive for the proliferation of NB PLC systems for home transmitted via the mains port. Therefore, it has been argued
and industry automation and for utility use such as smart that measurement methods and emission limits need to be
metering. There is no harmonized standard for frequencies adjusted for PLC devices, see [36], [30], [16, Ch. 3]. This
between 150 kHz and 500 kHz yet. However, the IEEE has been resolved relatively recently with the approval of
1901.2 standard specifies conducted disturbance limits in terms EN 50561-1 [37] in November 2012. This standard applies
of maximal power spectral densities [32, Sec. 7.5] and the to in-home PLC systems operating in the 1.6–30 MHz band
methods of measurement in an informative appendix [32, App. and it differentiates between a power port (only for power
E], respectively, see also [33]. supply), a telecommunications/network port (only for com-
In the U.S., PLC emissions are regulated through the Code munication signals) and a PLC port (for communication and
of Federal Regulations, Title 47, Part 15 (47 CFR §15) by the power supply). It specifies maximal voltage levels for the
U.S. Federal Communications Commission (FCC) [34]. Here, PLC signals and the corresponding measurement procedures.
the regulations for so-called “power line carrier” systems in The standard also requires dynamic power control and static
 3

and adaptive notching of frequency bands, which renders BB HomePlug ITU-T IEEE
PLC systems cognitive “radios” (which was anticipated by
the research community, e.g. [38], [39]). Further standards, June
namely EN 50561-2 for access networks and EN 50561-3 for 2001
frequencies above 30 MHz, are under development. HomePlug 1.0
released
The 47 CFR §15 by the U.S. FCC [34] defines in-house
and access “broadband over power line” (BPL) systems. The IEEE 1901
Authorization
former fall under the regulations for carrier current systems of project

June
mentioned above. The latter are specifically addressed in Aug ITU G.hn
2005
2005 Start of project
Subpart G for the band 1.705–80 MHz. Subpart G sets Apr
HomePlug AV
out radiated emission limits, differentiating between medium- released
2006

voltage (MV) and low-voltage (LV) installations, interference

Technical
mitigation and avoidance methods including adaptive power proposal
selected
control and frequency notching, administrative requirements
ITU-T G.9960 April
which include registration of deployments in a database, and consented 2008
excluded frequency bands for overhead deployments and geo- Dec Dec
2008 2008
graphic exclusion zones. As in EN 50561-1, power adaptation HomePlug
and notching is intended to avoid harmful interference to radio Green PHY
released ITU-T publishes
Baseline of the
standard confirmed
services. June June
The limits specified in EN 50561-1 and 47 CFR §15 for BB 2010 2010
Dec
G.9960 (PHY), G.9961 (DLL)
PLC can be translated into power spectral density masks for G.9972 (Coexistence)
2010

a given termination impedance, e.g. [40, Fig. 6.3]. This leads Dec
IEEE 1901
Jan publication
to a power spectral density (PSD) of about −55 dBm/Hz for 2012 2011

transmission up to 30 MHz at an impedance of 100 Ω, which HomePlug AV2

ITU-T G.9963
published
is consistent with the PSD specifications in the standards released

IEEE 1901 [41] and ITU-T G.9964 [42].

Fig. 1. Overview and timeline for the development of BB PLC specifications
and standards.
B. PLC Standardization
There are numerous industry specifications for NB PLC
systems that support link rates of up to a few kbps and operate ITU-T standardization project G.hn in April 2006 [49], [50].
in the application space of home and industry automation and In 2010, this resulted in the publication of the IEEE 1901
for utility applications, e.g. see [16, Ch. 7], [43, Ch. 2.2], [41] and the ITU-T G.9960/61 [51], [52] standards, which
[44]. Several of these were adopted as international standards specify the physical and data link layers as well as coexistence
in the late 1990s and early 2000s, including ISO/IEC 14908-3 mechanisms and PSD masks, see columns 2 and 3 of Fig. 1.
(LonWorks), ISO/IEC 14543-3-5 (KNX), and IEC 61334-5- The IEEE standard uses the 2–30 MHz frequency band
1/2/4. These systems have established a track record for PLC with an optional extended band of up to 50 MHz. It in-
as a proven technology for low data-rate applications. In the cludes two multicarrier physical (PHY) layers, which are
following, we first review the standards developement for BB commonly referred to as OFDM via Fast Fourier Transform
PLC and then NB PLC standards for relatively higher data (FFT-OFDM) and Wavelet OFDM, respectively. The former
rates. is classic (windowed) OFDM, while the latter is a discrete
1) BB PLC: The late 1990s saw a spur of activities in wavelet multitone (DWMT) modulation [53]. They are non-
the PLC community developing BB PLC solutions for the interoperable but their coexistence is enabled by an inter-
access and in-home domains, eventually targeting data rates of PHY protocol (IPP) [54], which later was extended to allow
hundreds of Mbps [45], [46]. This resulted in several industry coexistence also with G.9960, and this extension was called
specifications, mainly those backed by the HomePlug Power- inter-system protocol (ISP). The FFT-PHY applies a Turbo
line Alliance (HomePlug), the Universal Powerline Alliance code for forward error correction (FEC), while the Wavelet-
(UPA) and the High-Definition Power Line Communication PHY uses concatenated Reed-Solomon (RS) and convolutional
(HD-PLC) alliance. The left column of Fig. 1 shows some codes, which can optionally be replaced by low-density parity-
important steps in the evolution of HomePlug specifications, check (LDPC) convolutional codes. The PHY layers support
starting with HomePlug 1.0 released in June 2001 [47], multiple signal constellations and spectral masking as required
[48], which was adopted by the Telecommunications Industry by regulations discussed in Section II-A2. On top of these two
Association (TIA) as the international standard TIA-1113 in physical layers resides, via PHY layer convergence protocols,
2008. However, the existence of different non-interoperable a common medium-access control layer that enables both
specifications has not been ideal for broad market success. CSMA and time-division multiple access (TDMA).
Against this background, the consolidation of BB PLC systems While IEEE 1901 has provisions for in-home and access
in international standards was started by the IEEE P1901 networks, ITU-T G.hn applies specifically to home network-
Corporate Standards Working Group in June 2005 and the ing. It does not apply only to PLC but also to communication
 4

38 117 138
PRIME/ 125 144
ITU-T IEEE
G3-PLC

PRIME ARIB1
ARIB1, FCC-Low FCC-above-CENELEC
specification IEEE 1901.2
made public CEN A B
d ARIB 2
Feb
2008 FCC
ITU-T G.9902
FCC 1 FCC 2
G.hnem
ITU G.hnem
May Start IEEE 1901.2 CEN A B
d CD
d
Aug of project Authorization
2009
of project
2009 FCC
Jan ITU-T G.9903
PRIME Mar CEN A B
d
G3-PLC 2010 G3-PLC
Alliance 2010 ARIB
established specification
made public
ITU-T G.9955 (PHY) Jan ITU-T G.9904
ITU-T G.9956 (DLL) 2011 CEN A
(G3-PLC and PRIME as annexes) PRIME
Sep
IEEE 1901.2
2011 Dec 0 100 200 300 400 500 kHz
first draft
2011
36 91 98 122 155 403 488
G3-PLC 34 42 89 150 478
Alliance
established Oct
2012 IEEE 1901.2
publication Fig. 3. Frequency bandplans for standards and specifications of HDR NB
ITU-T publishes
Dec
PLC systems2 , following bands available in different regions of the world (see
G.9902 (G.hnem)
G.9903 (G3-PLC) 2013 Section II-A1). The numbers are the center frequencies of the start and end
G.9904 (PRIME) tone for each of the bands rounded to the next kHz-integer value. Adapted
supersede G.9955/56
from [61].

Fig. 2. Overview and timeline for the development of HDR NB PLC

specifications and standards. PLC systems. “HDR” means that data rates of tens to hundreds
of kbps are achieved using the 3–500 kHz frequency band. In
particular, in accordance with the frequency bands available in
over phone lines and coaxial cables. For PLC, it includes different regions of the world as described in Section II-A1, the
three bandplans, from 2 MHz to 25, 50, and 100 MHz, specifications listed in Fig. 2 have defined different bandplans
respectively. The spectral mask to comply with the emission as shown in Fig. 3.
limits outlined in Section II-A2 is consistent with that used in The industry specifications PRIME (Powerline Related In-
IEEE 1901. Also, as in IEEE 1901, windowed OFDM with telligent Metering Evolution) and G3-PLC have been de-
flexible bit loading is applied, and CSMA and TDMA are veloped by the PRIME Alliance and the G3-PLC Alliance
used for medium access. In contrast to the IEEE 1901 PHY respectively. Following the BB PLC example, multicarrier
modes, the physical layer of ITU-T G.hn uses LDPC block modulation and in particular OFDM has been adopted for the
codes. The IEEE 1901 and ITU-T G.hn standards are non- PHY layer. In contrast to BB PLC, differential modulation
interoperable. However, coexistence is enabled through the ISP is mandatory in both standards, but support of coherent
specified in IEEE 1901 and ITU-T G.9972 [55], whose support modulation was added to G3-PLC later on. Differential mod-
is mandatory for IEEE 1901 devices. ulation avoids the need for channel estimation and is thus
En-route towards Gbps transmission, a multiple-input better suited for transmission of shorter messages and more
multiple-output (MIMO) transmission extension to G.hn has robust to channel variations. Further emphasizing simplicity,
been specified as ITU-T G.9963 [56]. Similarly, HomePlug convolutional coding is used. In case of PRIME, even this is
published the HomePlug AV2 standard [57], which is back- optional, while G3-PLC adds an outer RS code. G3-PLC also
ward compatible with HomePlug AV and IEEE 1901 and specifies a robust mode that uses additional repetition. Such a
includes MIMO transmission as well as efficient notching and mode has been added to the latest version of PRIME.
power back-off to reduce emissions. At the other end of the In 2011, the ITU-T published recommendations ITU-T
data-rate spectrum, ITU-T G.9960 includes a low-complexity G.9955 for the PHY layer and ITU-T G.9956 for the link
profile for reduced component cost and power consumption layer, which included PRIME and G3-PLC as well as the
targeting the smart grid market. Similarly, the HomePlug new G.hnem technology. The latter uses coherent transmission.
Green PHY specification [58] has been developed as a subset This has been reorganized into standards ITU G.9902-04,
of the HomePlug AV standard for low power consumption and as shown in Fig. 2. ITU-T G.9904 (2012) adopted PRIME
low cost, targeting the home-area network domain of smart v1.3.6 as is, whereas G3-PLC adopted in ITU-T G.9903
grids. Fig. 1 summarizes the mentioned standards along the evolved since its first submission to ITU-T and went through
timeline of their publication dates. three major revisions (2012/2013/2014). In 2013, the IEEE
2) NB PLC: The development and standardization of BB published the IEEE 1901.2 standard, which is based on G3-
PLC systems have been followed by a wave of activity to PLC. However, as outlined in [61], IEEE 1901.2 and ITU-T
specify NB PLC solutions for relatively high data-rate (HDR) G.9903 have differences that render them non-interoperable.
transmission. These efforts have been driven by the demands But IEEE 1901.2 includes a NB-PLC coexistence protocol that
for an effective smart grid communication infrastructure [59], has also been adopted in ITU-T G.9903 (2014), which enables
[60]. Fig. 2 provides an overview of the development of major 2 The PRIME v1.4 PHY specification extends the usable frequency band to
industry specifications and SDO standards for such HDR NB 42–472 kHz.
 5

devices using these standards to coexist. 1.5

RMS Delay Spread [µs]

Indoor samples
1.8-100 Fit (ITA)
1 1.8-30 Fit (USA)
C. Industrial Solutions 2-30 Fit (ESP)
1.8-30 Fit (ITA)
Examples from each class of PLC, namely UNB, NB and 0.5
BB, have been implemented in products and find different
applications. Example of UNB technologies include the Turtle
0
system from Landis+Gyr and two way automatic communi- (a)
0
cations system (TWACS) from Acalara supporting data rates

Phase slope [rad/MHz]

from sub 1 bits/sec to 10 of bits/sec while reaching distances
of 150 km. For NB, several chip vendors including Renesas, -1

STM, Maxim, Texas Instruments, SemiTech, Semtech support

PRIME, G3, IEEE 1901.2 standards. Challenges in developing -2
Indoor samples
solutions for these standards include building high perfor- 1.8-100 Fit (ITA)
mance modems that can handle PLC interference, interface to -3
(b)
low impedance lines, challenge at the MAC level and building 1.5

RMS Delay Spread [µs]

reliable mesh networks. Large scale deployments of these Outdoor MV samples
1.8-30 Fit Outdoor MV
technologies have been announced in France and Spain. BB- 1 1.8-30 Fit LV OPERA
PLC vendors include Broadcom and Qualcomm implementing
Homeplug AV2 standard for in-home applications with MIMO
0.5
support. In a different application area, for wireless power
transfer/charging and communications, several products have
0
been announced for the Qi, A4WP wireless charging stan- -70 -60 -50 -40 -30 -20 -10 0
dards. (c) Average Channel Gain [dB]

Fig. 4. RMS-DS versus ACG for the in-home scenario in the 1.8–100 MHz
III. PLC M EDIUM band (a) and the outdoor MV scenario in the 1.8–30 MHz band (c), with the
corresponding robust regression fit. The robust fits for the 2–30 MHz Spanish
The power distribution network was not conceived as a (ESP) [64], 1.8–30 MHz USA [65] and our (ITA) 1.8–30 MHz measurements
medium for data transmission. As a medium it has peculiar (a), and the 1.8–30 MHz LV OPERA [66] model (c) are also shown. The
characteristics in the frequency band of interest, i.e. above unwrapped phase slope versus ACG for the in-home scenario, together with
the robust fit, are reported in (b).
10 kHz and up to 300 MHz. The primary characteristics are
the high frequency selectivity and attenuation: these are due
to multipath signal propagation caused by the presence of
multiple branches (discontinuities), unmatched loads and high Looking at Fig. 4a, note that the RMS-DS and the ACG
frequency selective low impedance loads. Time variations are are negatively related [63]. This indicates that the channel
also exhibited when the network topology changes and/or the attenuation is due to multipath propagation and it increases for
loads change. Furthermore, the PLC medium experiences high highly dispersive channels. The figure reports also the robust
levels of noise injected by devices connected to the power fit of the measurements carried out in Spain and in the USA,
grid or coupled through electromagnetic phenomena. In the described in [64] and [65] respectively. Despite the different
following, the PLC channel in home and outdoor networks wiring practices, the robust fit curves are very similar.
are described. The single-input single-output (SISO) channel, Fig. 4b shows the relation between the phase slope of the
the multiple user (MU) and the multiple-input multiple-output robust fit of the unwrapped phase of the channel frequency
(MIMO) channels are considered individually in the following response (CFR) and the ACG. The phase slope offers some
to highlight the main distinctive properties. information about the average delay introduced by the channel
and thus, in turn, on the length of the backbone, i.e. the shortest
electrical path between the transmitter and the receiver. The
A. SISO Channel greater the magnitude of the phase slope, the larger the
In this section, the channel properties are assessed in terms expected wire length and number of branches. Consequently,
of the main and most commonly used statistical metrics, the higher the attenuation, the lower the ACG, since the
namely the average channel gain (ACG), the root-mean- channel attenuation increases with the distance and the number
square delay spread (RMS-DS) and the coherence bandwidth of branches connected to the backbone.
(CB). Furthermore, data from measurements made in different Another important aspect is the definition of the channel.
countries are compared. In contrast to the wireless scenario, there is no expectation
The in-home scenario is considered first. In Fig. 4a and that the PLC channel is Rayleigh distributed. The amplitude
Fig. 4b the RMS-DS and the unwrapped phase versus the ACG of the channel frequency response is well fitted by the log-
(in dB scale) respectively are reported. The circles correspond normal distribution, as first reported in [63] and then in
to the scatter plot of all measured values in a campaign [65]. However, this is scenario dependent and deviations in
conducted in Italy (described in [62]) in the band 1.8–100 the distribution tails can be encountered [62]. Furthermore,
MHz, while the lines correspond to the robust regression fit. a correlation is manifested between the channel response at
 6

1
different frequencies [67]. Finally, the channel response can
Same Tx & site
exhibit a periodically time-variant behavior as a result of the 0.9 Same site only
periodic variations, with the mains AC voltage, of the load 0.8
impedance [68]. This is particularly true at frequencies below

Correlation Coefficient (ρ)

0.7
2 MHz.
Now, let us turn attention to the outdoor PLC channel. In 0.6
Fig. 4c, the outdoor MV channel from measurements con-
0.5
ducted in Italy [69] is considered. The robust fit for the outdoor
LV channel, from the EU OPERA project measurements [66], 0.4

is also depicted. In particular, note that the RMS-DS robust fit 0.3 P => N
slope of the outdoor MV channels is approximately half the P => E
0.2 P => CM
slope of the in-home channels considered in Fig. 4a. This is N => E
because the MV channels are more attenuated due to longer 0.1 N => CM
E => CM
cables and, furthermore, they exhibit lower RMS-DS due to 0
reduced multipath propagation into a network topology that 0 50 100 0 50 100
(a) Frequency [MHz] (b) Frequency [MHz]
has fewer branches. Contrariwise, the slope of the OPERA LV
channels is almost double the slope of the in-home channels. Fig. 5. MU correlation coefficient between SISO PLC channels sharing, or
This is due to the large number of signal reflections introduced not, the same transmitter (a), and spatial correlation coefficient among all the
by the typical network structure that consists of a backbone star-style receiving mode combinations for the MIMO channels (b).
with many short branches connecting premises. The high
attenuation in the OPERA LV channels may be explained by
the resistive characteristics of the deployed cables. communication link up to a certain node (named the pinhole
The channel at lower frequencies, e.g. in the NB spectrum of or keyhole) where branches then depart towards the final
9–500 kHz, is less attenuated than the channel at frequencies receiving destinations. This structure gives rise to a phe-
beyond 2 MHz, i.e. the BB channel both of the indoor and nomenon known as the keyhole effect, which was documented
in the outdoor environments. This is shown in Table II where in the context of cooperative multi-hop PLC in [72] and,
the average ACG, RMS-DS and CB are reported for different later, considering the physical layer security in PLC networks
bands and scenarios. Data were obtained from [62], [69], [66], in [73]. It is important to note that the MU concept holds
[70] (NB-PLC measurements in Indian and Chinese sites), [71] for both the indoor and the outdoor scenarios, and for any
(for the CENELEC-A band of 3–95 kHz). The NB channel implemented transmission scheme, such as SISO or MIMO.
characterization has been less documented than the BB one, However, within this section, the presented results are based
especially for the outdoor scenario. Given the relevance of on the 1300 SISO in-home channel measurements discussed
recently developed NB PLC technology, it would be beneficial in [62].
to further investigate the NB channel both in indoor and A key aspect of the MU channel is that the communication
outdoor scenarios and to report a detailed analysis. links are correlated or, in other words, there is a certain
level of determinism. This is significantly different from the
TABLE II wireless, where MU channel diversity is often introduced by
AVERAGE STATISTICAL METRICS FOR DIFFERENT CHANNEL SCENARIOS IN
DIFFERENT FREQUENCY BANDS . rich scattering (multipath) propagation. Now, to quantitatively
show this, we can compute the spatial correlation coefficient
∗
Scenario Band ACG RMS-DS CB ρ, which is defined as ρ = E[H (i) (f )H (j) (f )] for pairs of
(dB) (µs) (kHz)
In-Home 1.8–100 MHz −35.41 0.337 288.11
distinct i, j channels. In Fig. 5a, ρ is reported for channels
In-Home 1.8–30 MHz −31.91 0.394 216.48 of a given site sharing the same transmitter as well as when
Outdoor MV 1.8–30 MHz −40.53 0.491 458.58 the constraint of having the same transmitter is removed. The
OPERA LV 1.8–30 MHz −54.64 0.581 140.63
data base of measurements in [62] is used. It should be noted
OPERA LV 9–500 kHz −32.02 2.345 30.69
Outdoor LV [71] 3–95 kHz ∼ −35 ∼ 19 ∼4 as ρ takes high values, approximately equal to 0.5, along
Outdoor LV [70] 3–500 kHz −(15–33) 2.2–4 - almost the entire frequency range for channels sharing the
same transmitter. This high spatial correlation reduces the
available channel diversity.
B. Multiuser Channel Spatial correlation is exhibited also in the MIMO channel,
When we consider a network of nodes connected to the as discussed in the next section.
same power grid, it becomes important to characterize the
multiple user (MU) channel. In this respect, the underlying
C. MIMO Channel
network structure deeply affects the channel properties, and in
turn the achievable MU communication performance. MIMO systems are popular in the wireless domain where
The MU PLC network has, in general, a tree structure, they deploy multiple transmitting and receiving antennas. Also
so that pairs of nodes share part of the wireline network. in the PLC context, MIMO transmission can be established
For instance, if we consider a pair of channels from a given by exploiting the presence of multiple conductors. In home
transmitter to two distinct receivers, they share part of the networks, for instance, the power network comprises three
 7

P P A first study of the performance improvement achieved

through the use of a 2 × 4 MIMO communication method
N N w.r.t. the SISO system, in the 2–30 MHz frequency band, is re-
ported in [77]. Additive white Gaussian noise (AWGN), spatial
multiplexing (beamforming at the transmitter) and zero forcing
E E (ZF) detection are assumed. Precoded spatial multiplexing is
analyzed in [78], while [75] describes implementation results.
CM Finally, the performance improvements with the exploitation
of a 2 × 2 MIMO scheme with differential signal transmission
and reception are discussed in [79].

D. Channel Response Modeling

Fig. 6. Possible MIMO transmission modes in a PLC network, according to Modeling the PLC channel impulse and frequency response
the STF-410.
has attracted considerable attention. Models can be categorized
as top-down when the approach is phenomenological, and as
wires: the phase (P), the neutral (N) and the protective earth bottom-up when a physical description of signal propagation
(E) wires3 . using transmission line theory is used. Furthermore, the model
At the transmitter side, the differential voltage signal can be can be deterministic or statistical (random).
injected between pairs of wires giving rise to three different A deterministic top-down multipath propagation model was
signals, referred to as delta ∆ modes. Due to the Kirchhoff’s firstly proposed in [80] targeting the BB frequency spectrum.
laws, only two ∆ signals can be injected at the same time. Then, it was improved in [81], extended in statistical terms
Contrariwise, if a reference plane is used at the receiver firstly in [82] and refined in [83]. Other top-down BB random
side, the signal can be observed between one conductor and channel models were proposed in [84] (in frequency domain)
the reference plane. This configuration is referred to as star- and [65] (in time domain). Inspired by [82], a MIMO statistical
mode (S-mode). The number of possible S-modes is three, top-down model was presented in [85] for the BB spectrum.
First attempts for bottom-up NB modeling were presented in
corresponding to the number of wires. However, an additional
[86] for the in-home scenario, while in [87] for the NB outdoor
mode, named common mode (CM), can be exploited [74],
low-voltage scenario. More recently, bottom-up models based
[75]. Large TV screens, for instance, include a large metal
on the s-parameters and ABCD-matrix representations were
plane that can act as a reference plane. If the reference plane
proposed in [88], [89], [90]. While [88] and [90] consider
is not available, then, the delta reception mode can be used.
the BB frequency range, [89] considers the NB spectrum
The CM flows with the same intensity and direction through
providing a validation of the model in time domain. In what
the P, N and E wires. For EMI reasons, the CM is used only
follows, all the referred works consider the BB spectrum. In
at the receiver side. Hence, as shown by Fig. 6, a 2×4 MIMO
particular, it was shown that a random extension of the bottom-
transmission can be established. However, it should be noted
up model is possible by using a random (although simplistic)
that, due to Kirchhoff’s laws, the fourth signal collected by
topology representation in [91]. This bottom-up modeling
the start-style coupler is linearly dependent on the other three
approach can also be exploited to include the periodic time
signals, but it can provide a signal-to-noise ratio gain as well
variant channel changes by adding the time variant behavior
as a capacity gain [40, Ch. 1 and 5].
of load impedances [92]. A statistically representative random
In the MIMO scenario, the PLC channels are correlated due
topology model for home networks was presented in [93]
to the symmetry and determinism of the wiring structure. The
together with an efficient way to compute the channel transfer
correlation of the channel responses among different S-style
function in complex networks referred to as voltage-ratio
receiving modes (where e.g. P⇒N indicates the correlation
approach. This model was used to infer the BB channel
among phase and neutral) is reported in Fig. 5b. The figure
statistics as a function of the network geometry in [94].
shows that the correlation coefficient is even higher than that
While bottom-up modeling offers a tight connection with
found in the MU channel sharing the same transmitter in
physical propagation of PLC signals in a certain network, top-
Fig. 5a. In particular, high levels of correlation are exhibited
down modeling is particularly attractive for its low complexity.
among the channels P⇒E and N⇒E, with the highest values
It is therefore foreseen that refined top-down models will be
experimented among the P⇒N channels responses, especially
developed in the future. Recently, it has been shown that
for lower frequencies. These high correlation values are due
the simplest way to model the SISO CFR is to directly
to the fact that the power is delivered through phase and the
generate the amplitude and phase as a vector of correlated
neutral wires, which are positioned one next to the other and
complex random variables, whose marginals have log-normal
follow the same path from the transmitter to the receiver.
amplitude and uniform phase distribution [67]. An analytic
Not only the MIMO channel responses are correlated but
expression for the correlation matrix in the frequency domain
also the noise. Colored and spatially correlated noise are con-
(i.e. between the channel samples in frequency) can be derived
sidered in [76] for the 2 × 4 scheme (see also Section III-F).
from experimental measures. Finally, by exploiting the relation
3 The protective earth acts as a return path for the power supply in the case existing between the Pearson (linear) and the Spearman (rank)
of an insulation fault. correlation, the multivariate CFR distribution can be generated.
 8

noise can be done by observing it at the receiver port [98] or at

the source [99]. The aperiodic noise is the most unpredictable
102
Re{Z in} [Ω]

component and it is due to the connection and disconnection

of appliances from the power delivery network. The amplitude
of the aperiodic noise can be significantly larger than that of
100
the other impulsive noise components. Beside the amplitude,
the aperiodic impulsive noise is typically described by the
(a)
2 duration and the inter-arrival time [100]. The statistics of
these quantities depend on how the impulsive noise events
1
are identified and measured.
Im{Zin} [kΩ]

0 The PLC background noise is usually modeled with a

stationary Gaussian colored process having a frequency de-
-1 Measurements creasing power spectral density (PSD) profile, according to
Average value
-2
a polynomial [91] or exponential [79] function of frequency.
0 10 20 30 40 50 60 70 80 90 100 Typical noise PSD trends, having different floor levels and
(b) Frequency [MHz]
profiles, have been reported in [67] for the in-home, the
Fig. 7. Real part (a) and imaginary part (b) of the input line impedance for outdoor LV and MV scenarios. The main differences are
the in-home scenario in the broadband frequency range 1.8–100 MHz. related to the network structure and topology, as well as to
the type of connected loads. For example, indoor networks
are characterized by many loads interconnected through a grid
E. Line Impedance deploying short cables. This prevents the noise attenuation,
Not only the channel response is important, but also the line leading to high levels of noise at the receiver side. Outdoor
impedance has to be considered since it affects the design of networks, instead, are affected by the noise contribution gen-
the analog front-end of the PLC modem. A low line impedance erated by the overall industrial and residential consumers, by
at the transmitter port makes the injection of the voltage the inverters used in renewable generation plants, as well as
signal challenging. The measurement results have shown that by the RF interference coupling into the grid.
the line impedance can be significantly low (in the order A model for the periodic noise terms that is based on a
of few ohms) especially at low frequencies. For instance, in deseasonalized autoregressive moving average is presented in
the access network, this is due to the fact that the home [101]. Several authors suggest to fit the amplitude of the noise
network acts as many parallel loads attached to the access in the time-domain with the Middleton’s class A distribution,
port [95]. In the home network, the line impedance (at the e.g., [102]. In [103] it was speculated that the Nakagami-m
outlets) exhibits a highly frequency-dependent behavior, as distribution is more appropriate. However, the results in [103]
shown in Fig. 7. Interestingly, it is mostly inductive and the are obtained from too few measurements to be considered con-
real part increases at high frequency. This behavior can be clusive. Some further studies reveal that the normal assumption
especially noted looking at the single realizations in Fig. 7a on the noise statistics holds true if the periodic time-variant
(each one represented with a different color). Therefore, it nature of the noise is accounted for [104] and the impulsive
is expected that broadband PLC can be less affected by this noise contributions are removed from the measures [105].
issue than NB PLC and the broad spectrum provides channel Based on this, filtering a stationary process with an LPTV
and impedance diversity which can simplify the design of system is proposed in [106] to model cylostationary Gaussian
impedance adaptation techniques [96]. distributed noise. In [97] a Markov-chain model is proposed to
model the ensemble of components. In [107] the noise at the
source is modeled with a non-Gaussian distribution as in [91]
F. Noise Properties and Models and then the noise at the receiver is obtained by filtering it
The PLC medium is affected by severe noise with station- with the channel generated with a top-down channel response
ary (typically referred to as background) and nonstationary generator. It is found that with a sufficiently large number of
(referred to as impulsive) components, which depend on the noise sources, the overall noise at the receiver approaches the
considered application scenario (e.g. indoor or outdoor and Middleton’s class A distribution. In [108] another top-down
LV or MV). The former is a combination of conducted noise channel generator is used, instead, to filter the source noise.
and coupled radio signal contributions. The latter can be Noise in the MIMO context has not been thoroughly studied
cyclostationary, with a repetition rate that is equal to, or yet. Experimental results in the home scenario are reported in
double, that of the mains period, or bursty and cyclostationary, [109], [110] and a model to account for the spatial correlation
with a repetition rate that is high between 50 and 200 kHz, or of noise is proposed in [76]. In this respect, Fig. 8a shows
aperiodic. The first two components are referred to as periodic the noise PSD measured in the home scenario at the P, N, E
noise synchronous and asynchronous with the mains frequency and CM ports using an S-style coupler. While, Fig. 8b reports
[97]. The periodic synchronous noise originates from silicon the cross-PSD (C-PSD) among each pair of noise modes. The
controlled rectifiers (SCR) in devices, while the asynchronous C-PSD allows us to highlight the spatial cross-correlation that
noise is due to the switching activity of power supplies. The exists between the noise signals in frequency domain. It is
characterization in the time and frequency domain of PLC defined as the covariance between the noise signals observed at
 9

-120 1
P (STF-410) P => N 1.8-30 MHz SISO capacity
N (STF-410) P => E
0.95 1.8-100 MHz MIMO (uncorr. noise)
-125 E (STF-410) P => CM 1.8-300 MHz MIMO (corr. noise)

CCDF (Prob[C>x] and Prob[C >x])

CM (STF-410) N => E 0.9 SISO secrecy cap.
MIMO secrecy cap.

S
N => CM
E => CM 0.85
-130
Noise PSD [dBm/Hz]

CCDF (Prob [C>x])

0.8
-135
0.75

-140 0.7

0.65
-145
0.6

-150 0.55

0.5
0 50 100 0 50 100 0 0.5 1 1.5 0 0.5 1 1.5
(a) Frequency [MHz] (b) Frequency [MHz] (a) Capacity (C) [Gb/s] (b) C and CS [Gb/s]

Fig. 8. Noise PSD profiles for the star-style receiving modes measured by Fig. 9. Performance improvement due to bandwidth extension (a) and to the
the STF-410 (a) and the computed C-PSD among the different modes (b). MIMO transmission in the 1.8–100 MHz band together with a colored and
correlated Gaussian background noise assumption (b).

modes Si and Sj that is given by Ri,j (f ) = E[NSi (f )NS∗j (f )]

with i 6= j, where NSi (f ) is the Fourier transform of the noise of noise properties and associated statistics has yet to be
experienced by the mode Si [76]. In particular, from Fig. 8a achieved. Typically, capacity is computed under the stationary
note that the CM experiences the highest noise PSD, while Gaussian noise assumption, as we do in the following. We also
the other three modes, i.e. P, N and E, have approximately the assume that the transmitted signal has a PSD of −50 dBm/Hz
same PSD. The PSD significantly increases above 87 MHz due up to 30 MHz and −80 dBm/Hz beyond 30 MHz (according
to coupled FM broadcasting radio signals. Fig. 8b, instead, to the HomePlug AV2 standard [57]).
shows that the C-PSD profiles resemble the PSD trends Firstly, we focus on the SISO in-home channel and on
depicted in Fig. 8a, although they are lower. This is because the the gain attainable by a signal bandwidth extension. Fig. 9a
C-PSD elements are a linear combination of the PSD profiles reports the capacity complementary cumulative distribution
in Fig. 8a and depend on a time-domain correlation coefficient function (CCDF) with measured channel responses and typical
that ranges between 0 and 1, as discussed in [76]. The lowest background noise PSD (see [62] for details). A bandwidth
noise C-PSD profile is exhibited between the modes P and the extension is beneficial, indeed. For 50% of cases a rate
N since the corresponding PSD profiles are the lowest ones. exceeding 1.7 Gb/s can be achieved with a band 1.8–300 MHz,
Contrariwise, the combinations P⇒CM, N⇒CM and E⇒CM almost doubling the 1 Gb/s achieved in the 1.8–100 MHz band.
are affected by higher and similar noise levels. However,
Fig. 8b shows that the noise C-PSD exhibits non-negligible Now, we turn our attention to the MIMO scenario by
levels and significant differences among the spatial modes. exploiting the channels measured by the ETSI special task
force 410 (STF-410) discussed in [74]. Herein, we also wish
to show the effect of the noise when its time and space
IV. P HYSICAL L AYER P ERFORMANCE
correlation is taken into account. The spatial correlation of
From the characterization of the medium, it is possible the noise is considered according to the model described in
to assess the performance of the PHY layer. The Shannon [76]. Fig. 9b shows that MIMO provides significant gains
capacity is the common metric used to determine the theo- over SISO with the same total PSD constraint. The difference
retical achievable rate limit. The secrecy capacity is another between the SISO results in Fig. 9a and Fig. 9b are not
metric and it refers to the rate achievable by a communication pronounced and are attributable to the distinct CFR databases
that grants perfect confidentiality and secrecy. The results and slightly different background noise PSDs used for each
discussed in the first two sections, namely Section IV-A and experiment. The worst MIMO performance is obtained with
Section IV-B, deal with the BB indoor scenario considering the colored and spatially uncorrelated noise assumption, while
both SISO and MIMO transmission. Afterwards, PHY layer colored and spatially correlated noise lead to a further capacity
aspects and possible improvement directions are discussed improvement if precoding is implemented. This is due to
considering both NB and BB systems in Section IV-C and the fact that the knowledge of the noise correlation matrix
Section IV-D, respectively. facilitates its mitigation at the receiver side.

A. Capacity B. Secrecy Capacity

The capacity depends on the channel, e.g., SISO or MIMO, The secrecy capacity is another less common performance
the bandwidth and the noise assumptions. The true capacity metric. It indicates the amount of information that can be
of the PLC channel is unknown since a full characterization reliably and securely exchanged between two nodes, without
 10

any disclosure of information towards a third party, or adver- high speed communication for multimedia services, thus, they
sary, commonly referred to as eavesdropper or wiretapper. The require more complex techniques, e.g., turbo codes in the
secrecy capacity is studied in the context of so called physical HPAV and IEEE 1901 standards [41], while LDPCs in ITU-
layer security (PLS). While PLS has been studied extensively T G.9960 (G.hm). Besides the above mentioned techniques,
in wireless, the first study in PLC was reported in [73] and some other coding schemes are currently under investigation
then extended in [111]. In the following, the main concepts by the research community. For instance, the permutation
concerning the PLS in PLC networks are briefly summarized. trellis codes which combine permutation and convolutional
From an information-theoretic point of view, a wiretap codes are particularly suited to combat burst of errors [114].
channel consists of a transmitter (Alice) that wants to send The modulation scheme is also important, especially be-
a confidential signal to a legitimate receiver (Bob) without cause, as we saw in Section II-A2, spectral masks have to be
any leakage of information towards a malicious eavesdropper fulfilled as specified by the standards. Thus, it is important
(Eve), which tries to disclose the message. In the presence of to realize flexible spectrum usage with the ability to create
Gaussian noise, the secrecy capacity under a power constraint spectral notches and allow coexistence with other systems.
is defined as CS = maxfx ∈F [I(x, y) − I(x, z)]+ , where x is The most commonly used modulation scheme is pulse-shaped
the signal transmitted by Alice, while y and z are the signals OFDM (PS-OFDM), a multicarrier scheme similar to OFDM,
received by Bob and Eve, respectively. The quantity fx denotes but with the usage of a window which is better than the
the probability density function (pdf) of x, whereas F is the rectangular time-domain window adopted in OFDM. PS-
set of all possible pdfs of the input signal x. The terms I(x, y) OFDM is deployed in latest NB and BB PLC standards (see
and I(x, z), instead, represent the mutual information among Section II-B). It is also interesting to note that while BB
x and y or z, respectively. Moreover, [q]+ = max(q, 0) so that system use coherent modulation, i.e., M-PSK and M-QAM,
CS is set to zero when Eve has a better channel realization NB systems use also differential PSK. In particular, ITU-T
than Bob. Since the mutual information terms are convex in fx , G.9903 (G3-PLC) deploys the conventional time-differential
a lower bound RS for the secrecy capacity can be formulated phase modulation, while ITU-T G.9904 (PRIME) deploys
+
as CS ≥ [maxfx ∈F [I(x, y)] − maxfx ∈F [I(x, z)]] = RS frequency differential phase modulation where the information
[112]. In [73] it was proved that, differently from capacity, is encoded in the phase difference between adjacent OFDM
the secrecy capacity is upper bounded by a constant value sub-channels.
even if the power indefinitely increases. Furthermore, it was Recently, in view of an evolution for further improvements,
shown in [73] and [111] that the channel statistics (log-normal more attention has been directed to the study of other types
in PLC and not Rayleigh as in wireless), as well as the spatial of filter bank modulation (FBM) that privilege the frequency
correlation of the channel (introduced by the keyhole effect), confinement of the sub-channel pulses, e.g. filtered multitone
may further limit the secrecy capacity. (FMT) modulation [115]. FBM, such as FMT, offers sev-
In Fig. 9b, the secrecy capacity CS for the SISO and eral advantages over PS-OFDM, as the higher sub-channel
the MIMO transmission schemes is compared to the capac- frequency confinement and the higher notching selectivity,
ity C. The figure shows that the SISO secrecy capacity is allowing a reduction in the number of sub-channels required
considerably lower than the unconstrained capacity but, with to be deactivated to meet EMI constraints. In order to reduce
the use of MIMO transmission, it can increase [111], [113]. the FBM implementation complexity, the use of a different
The secrecy capacity is in general low because it is upper architecture, where the linear convolutions are replaced with
bounded (as a function of power [73]) and it is obtained as circular convolutions, was proposed in [116]. In this case,
the difference among the rates of the intended receiver and of the transmission takes place in blocks, similarly to OFDM,
the eavesdropper. resulting in a scheme referred to as cyclic block FMT (CB-
FMT). The circular convolution is also applied in the filter
bank analysis at the receiver, offering an efficient frequency
C. Practical PHY Layer Issues: Modulation domain implementation [117].
In order to achieve the channel capacity, advanced modu- As an example, the achievable rate for a specific channel re-
lation and coding schemes have to be deployed. Current PLC alization (which corresponds to the median channel, ranked in
technology deploys powerful channel coding schemes, such terms of capacity, selected from the database of measurements
as concatenated Reed-Solomon codes, convolutional codes, in [62]) is reported in Fig. 10 as a function of the number of
turbo codes, low density parity check (LDPC) codes, or a used sub-channels. Different lengths Lg of the prototype filter
combination of these techniques, together with high order at the transmitter side are considered. The cyclic prefix (CP)
modulation. Furthermore, to overcome burst of errors in- length is chosen in order to offer the highest rate, under the
troduced by noise and channel frequency response notches, further constraint of fulfilling the notching mask. The figure
interleaving can be deployed in order to spread in time and in shows that the rate in CP-OFDM increases with the number of
frequency (over the sub-channels in multicarrier modulation sub-channels. This happens since the overhead introduced by
schemes) coded blocks of bits or symbols. At the moment, the CP is reduced and a better notching capability is obtained.
NB systems use simpler coding techniques e.g., convolutional However, the performance gap from the theoretical channel
codes with bit interleaving in ITU-T G.9904 (PRIME) and capacity is high. This gap can be significantly reduced by
convolutional and Reed-Solomon codes in ITU-T G.9903 (G3- deploying FMT. In this example, FMT uses a long root-raised-
PLC) and in IEEE 1901.2. Contrariwise, BB systems provide cosine pulse with length 20 or 10 sub-channel symbols. To
 11

300 it is a shared medium with unpredictable channel behavior.

Theoretical
FMT (Lg =20 N)
Consequently, the link layer and other layers immediately
250 FMT (Lg =10 N) above face similar challenges to a protocol designed for
CB-FMT (L g =4 N)
wireless. As we will see, this means that the link layer
OFDM
Achievable rate [Mbps]

200 and routing layer often have common elements with their
wireless counterparts. There are also differences: for example,
as mentioned above the PLC medium can be time-varying
150
with the AC mains cycle, and so transmissions or beaconing
are aligned with the mains cycle.
100

50
A. MAC Protocols for BB Applications
The MAC protocols defined in the HomePlug [47], IEEE
0 1901 [41] and G.hn (G.9961) [52] standards define a
128 256 512 1024 2048
contention-based (random access) as well as a contention-free
Sub-channels
(TDMA-like) procedure to access the channel. However, it
Fig. 10. Comparison between OFDM, FMT and CB-FMT in terms of is the random access procedure that usually is observed for
maximum achievable rate as a function of the number of sub-channels used Internet traffic [129], and thus we will focus on this in the
K.
following. Although the three standards differ on how the
channel access is realized, the random access procedure is
reduce the FMT complexity, the figure shows that CB-FMT equivalent. The general procedure adopted is a CSMA/CA
with a rectangular frequency domain pulse (which renders the technique, similar to the Distributed Coordination Function
system to be the dual of OFDM) provides better performance (DCF) defined in the IEEE 802.11 standard [130]. Each time
than PS-OFDM and, with 2048 sub-channels, it is not far away a node has a new packet to transmit, the backoff stage (i ∈
from FMT, despite the fact that its complexity is only 36 % [0, m−1]) is initialized to 0 and a random backoff counter (BC)
of the FMT one. is selected from [0, W0 ]. The backoff countdown is frozen
when activity is detected on the channel and restarted when the
medium becomes idle again. The packet is actually transmitted
D. PHY Layer Improvements
when the backoff countdown expires. If an acknowledgment
There are several areas that are currently being investigated is received, the packet is considered successfully transmitted.
for the improvement of the PHY layer. The main ones are: Otherwise, the node starts the retransmission procedure: the
channel coding and signal processing for mitigating impulsive backoff stage changes to i = min(i + 1, m − 1) and a new
noise and interference [114], [118]; synchronization, channel BC is selected from [0, Wi ], Wi being the contention window
estimation and equalization [119], [16]; transmission schemes of stage i. The similarity of the MAC to 802.11’s DCF has
that can allow coexistence at the PHY layer between different lead to MAC modeling (e.g. [131], [132], [133], [134]) often
PLC systems [54], between high speed PLC systems together in the style of Bianchi’s 802.11 model [135].
with sensor PLC networks [120] and between PLC systems 1) The Deferral Counter - Its Impact on Fairness: In
and DSL systems [121], [122]; adaptation and resource al- contrast to the DCF specification, in the HomePlug, IEEE 1901
location for maximum spectral efficiency, e.g., bit loading MAC and G.hn (G.9961), a new counter, called the Deferral
[16, Ch. 6] or adaptation and allocation of the time/frequency Counter (DC), is introduced. This counter is initialized to Mi
resources in multicarrier systems [123], [124]; cooperation and at each backoff stage and decreased by one after overhearing
relaying to extend coverage [125], [126]; diversity combining a data packet or a collision. If a new packet or a collision
techniques that mix PLC with wireless transmission [127]. is overheard and the value of the DC is equal to zero, the
In NB PLC for smart grid applications, research effort is node acts as if a collision has happened: the backoff stage is
spent to increase robustness and coverage, for instance, look- increased if it has not yet reached its maximum value and a
ing at increased spectrum usage, possibly beyond 148 kHz, new backoff is selected from [0, Wi ]. The goal of the DC is to
or exploiting better coupling techniques that can resolve the avoid collisions when high contention is inferred by decreasing
limitations due to the low line impedance and high noise [128]. the aggressiveness of transmission attempts.
In BB PLC, one question is how to go beyond the current The use of the deferral counter does reduce the collision
high throughput offered to support very high speed multimedia probability when there is high contention. However, as shown
services and home networking applications. One direction is to in [136], this modification to the DCF does not always provide
better exploit MIMO and another is to increase the bandwidth better performance, especially considering heterogeneous and
using an EMC friendly mechanism, for instance, as recently exposed terminal scenarios. More importantly, it has been
introduced by the HomePlug AV2 standard [57]. shown in [137] that it has an impact on short-term fairness
as some stations may substantially reduce their transmission
V. L INK AND H IGHER L AYERS probability by overhearing consecutive neighboring transmis-
As we have seen, while the PLC medium has differences to sions at a given time interval. The trade-off between collision
the wireless medium, it also has many similarities. In particular probability and fairness has been studied in [138].
 12

FRAME RIFS CIFS FRAME 250

ACK PRS0 PRS1 BACKOFF ...
TRANSMISSION TRANSMISSION
t 200 CA3
av. 181.3285

Frequency
150 min. 65.5855
Fig. 11. Allocation of priority resolution slots (refer to [47] and [41]). max. 204.1066
100

50

2) Strict Prioritization - Benefits and Drawbacks: To pro- 0

0 20 40 60 80 100 120 140 160 180 200
SCA3 [Mbits/s]
vide channel access differentiation, HomePlug [47], IEEE
1901 [41] and G.hn (G.9961) [52] define four Access Cat- 2500

egories (CAs), ranging from CA0–3. CA3 and CA2 share Wi 2000 CA0
av. 0.32049

Frequency
and Mi values, as do CA1 and CA0. Two Priority Resolution 1500 min. 0
max. 115.7184
Slots (called PRS0 and PRS1) are allocated at the end of suc- 1000

cessful frame exchanges, see Fig. 11. These slots allow nodes 500

to announce the priority of packets pending transmission. The 0

0 20 40 60 80 100 120 140 160 180 200
highest priority (CA3) is signaled by transmitting a symbol in SCA0 [Mbits/s]

both PRS0 and PRS1; the CA2 category is signaled in PRS0

Fig. 12. Histrograms of throughput for 2 stations configured at CA3 and
only; CA1 signals in PRS1, if PRS0 was empty; and the lowest CA0. Experimental results extracted from [129].
access category (CA0) does not signal at all. Following this
approach, stations know if there is a station with a frame that
belongs to a higher CA. In such a case, they do not contend considered stale.
for the channel, allowing high-priority frames to be released. 3) Aggregation and Buffer Management - Efficiency vs.
This resolution scheme aims to provide strict access differ- Variability: The standards HomePlug [47], IEEE 1901 [41]
entiation, i.e. using the priority resolution mechanism, packets and G.hn (G.9961) [52] provide a high degree of aggregation
with higher priority are always transmitted before lower- capabilities. Experimental studies on commercial IEEE 1901
priority ones. However, the priority resolution scheme in devices show that although the efficiency can be improved by
HomePlug and IEEE 1901 is only invoked after successful aggregating more data in a single transmission when channel
frame exchanges. These standards suggest that PRS are not conditions are favorable, it may result in a high degree of
present after: i) a collision, ii) frame transmissions resulting in variability on performance [129]. A challenging aspect of
erroneous receptions and iii) the detection of an empty channel studying this is that both aggregation techniques and buffer
for longer than an Extended InterFrame Space (EIFS) period. management decisions are vendor-specific.
Thus, in lightly loaded conditions and after collisions or
channel errors, the priority resolution scheme is not employed
and channel access differentiation only occurs through the B. MAC Protocols for NB Applications
different parameters of the access categories. Thus, we expect The MAC protocols for NB applications share many simi-
strict prioritization if we have a single station in a high CA, but larities but they also have several differences that are important
less strict prioritization if multiple stations are in the highest to consider [61]. We give an overview of their common
CA because of collisions. characteristics and differences next.
Channel differentiation in PLC networks has been evaluated The standard PRIME defines a Contention Free access
in [139], [140], [141], [142] and [129]. Although the priority Period (CFP) in which devices transmit using TDMA, as well
resolution mechanism is able to provide strict protection to as a Contention Access Period (CAP) where channel accesses
high-priority traffic, a series of issues has been identified. are randomized. During CAP, stations wait a random backoff
First, one immediate effect of this strong protection to high- before attempting transmission. The random backoff depends
priority frames is the starvation faced by lower-priority traffic on the priority of the frame and on the number of experienced
[129]. To illustrate the importance of this effect we show in channel attempts. After the backoff expires, stations carrier
Fig. 12 histograms of throughput from experimental results sense the medium a number of times if it is found busy. The
in a real testbed extracted from [129]. Note that in presence number of times the carrier sensing procedure is carried out
of a higher access category, the station configured at CA0 is also depends on the priority of the frame to be transmitted.
not effectively able to transmit. This effect is caused by the Contrary to PRIME, the MAC protocol defined in IEEE
inability to transmit any lower-priority frame when stations 1901.2 is based on the IEEE 802.15.4 standard for wireless
with higher-priority frames are saturated, i.e. they always have sensor networks [143]. The same applies to G3-PLC, except
a packet pending to transmit in their queues. Moreover, as for the fact that G3-PLC does not define a CFP period. In G3-
shown in [142], the behavior of the network is extremely hard PLC and in the CAP period of IEEE 1901.2, stations perform
to predict when we vary the number of stations contending a random backoff before attempting transmission, similarly
for the channel or when the traffic load changes. Additionally, to IEEE 802.15.4. G3-PLC and IEEE 1901.2 extend the
given that control messages for tone map update are sent at contention procedure of IEEE 802.15.4 to account for fairness
CA2, a possible oscillatory behavior in throughput has been (stations with a high number of busy channel detections
identified [129]. This is because it is impossible to release increase the aggressiveness of their transmission attempts) and
these control messages in presence of a saturated CA3 source, different priorities (by defining different contention periods for
which prevents a given station transmitting as its tone map is different access categories). A common feature of the random
 13

access procedure in G3-PLC, IEEE 1901.2 and PRIME is that devices for extending the reach of networks are available
the assessment of the channel status only occurs when the off-the-shelf [151]. The IEEE 1905.1 standard provides a
backoff expires. convergence layer to facilitate the use of WiFi, PLC, Ethernet
Finally, the NB MAC procedure defined in G.hnem is more and MoCA within the home [152], [153]. A generic exten-
like the CSMA/CA approach defined in the IEEE 802.11 sion mechanism has also been recently standardized in IEEE
standard [130] for wireless local area networks. In this case, no 1905.1a [154]. Other uses of PLC have been proposed, for
CFP is defined, and thus stations access the channel following example a mix of WiMAX and PLC has been proposed for
a contention-based approach. If a transmission is detected collecting data in a hospital [155]. Visible light communication
on the channel, stations defer their attempt until the next is another promising technology that is complementary to WiFi
contention period, which takes place once the current ongoing and PLC [156].
transmission is completed. G.hnem also accounts for different Another question is what protocol should be run over PLC.
prioritization levels. However, in contrast to G3-PLC and IEEE Broadband PLC is often used like Ethernet, and so can be
1901.2, the differentiation is not so strict as different access used in much the same way as any LAN, running IPv4,
categories have contention windows that partially overlap. IPv6 or other protocols. However, in NB PLC, sometimes
Although the PLC research community can rely on the resources are at more of a premium. For example, in G.9903
extensive work on IEEE 802.15.4 and IEEE 802.11 to predict 6LoWPAN is used to carry IPv6 frames on the PLC network.
the performance of the network, these standards include some 6LoWPAN provides a number of functions that might be
modifications which are, as far as we know, unexplored optimized specifically for PLC (e.g. routing [157]).
at present. These are: i) the modifications in G3-PLC and Of course, when integrated with the network ecosystem,
IEEE 1901.2 of the backoff procedure to provide fairness, ii) PLC needs to be managed. A number of PLC vendors provide
the strict prioritization mechanisms defined in G3-PLC and tools, including open source tools such as faifa4 and open-
IEEE 1901.2, and iii) the prioritization mechanism defined in plc-utils5 . Efforts have also been made to provide an SNMP
G.hnem. The impact of these extensions on performance is interface to PLC devices [158], [159], [160]. Within hybrid
not straightforward and further analysis is needed in order to networks, IEEE 1905.1’s convergence layer also provides
understand the behavior of the network. abstractions to help with management, including features for
estabilishing the topology and link metrics [152], [153], [154].
C. Routing Issues
G.hn and IEEE 1901 support link layer multihop operation, E. Challenges and Future Directions
where nodes that are not in direct communication can have
frames received and retransmitted via intermediate nodes, and Compared to research on IEEE 802.11 and IEEE 802.15.4
the protocol can take advantage of link quality information networks, and also compared to the advances on the physical
provided by the lower layers. There is also a possibility to take layer of PLC networks, the MAC protocols for PLC are
advantage of other aspects of PLC networks for routing. For unexplored. Work has begun to fill this gap, however there are
example, previously we discussed that the topology often has still many aspects that remain unclear and several issues that
a tree-like structure, which might be exploited by a routing need to be studied in order to ensure the successful penetration
system [144]. Likewise, for many PLC devices, it is likely of the technology.
that they are attached at a physically fixed location, and so In particular, extensions to the analytical models of Home-
geographic routing may be practical [145], [146]. Plug and IEEE 1901 [134], [138], [133] in order to consider
Indeed the NB PLC standards are divided on whether aggregation and buffer management techniques are needed
routing should be carried out at the link layer or above [147], in order to fully understand the protocol behavior and the
[148]. G.9903 and G.9904 include link-layer routing, where impact on performance. Similarly, amendments to the stan-
all nodes appear to be connected, even if relaying is taking dards related to the deferral counter and the strict priority
place. In contrast, IEEE 1901.2 and G.9902 allow routing at resolution scheme may also be desirable. Also, as previously
the link layer or above, where in the latter case a higher layer stated, the impact on performance of the extensions to the
protocol must handle forwarding between nodes not in direct IEEE 802.15.4 and IEEE 802.11 baselines considered in NB
communication. PLC standards regarding fairness and prioritization remain
Of course, rather than receive and retransmit, it is possible relatively unstudied.
to have nodes relay in real-time, as in cooperative transmission A combined understanding of PHY effects and MAC layer
in wireless. Cooperative relaying has been considered for issues can raise interesting issues. An example of this is the
PLC (e.g. [149], [72], [126], [125], [150]). While diversity challenge of building a stream protocol for smart grid on
gains are often lower than in wireless, power gains through top of stop-and-wait MAC protocols common in PLC [161],
multihop transmission are still practical, which can be useful where MAC delays can indirectly result from using long
for improving range. OFDM symbols to mitigate burst interference. Though IEEE
1905.1 provides basic mechanisms, such as link metrics, for
addressing the use of hybrid PHY layers, the optimal use of
D. Integration with the Networking Ecosystem
Some integrations of PLC into the broader network have 4 Available at https://github.com/ffainelli/faifa
been successful. Consumer modems, or integrated PLC-WiFi 5 Available at https://github.com/qca/open-plc-utils
 14

its routing and multipath forwarding features are still open [10] L. R. Castor, R. Natale, J. Silva, M. E. Segatto, et al., “Experimental
questions. investigation of broadband power line communication modems for
onshore oil & gas industry: A preliminary analysis,” in IEEE Intl. Symp.
on Power Line Commun. and Its Appl. (ISPLC), 2014.
VI. F INAL R EMARKS [11] W. Wu, B. Chen, and C. Li, “Experimental research on triple play via
PLC networks have become a fruitful technology which power line,” in IEEE Intl. Symp. on Power Line Commun. and Its Appl.
(ISPLC), 2012.
can provide a means of communication for a wide range [12] A. Gnazzo, A. Bergaglio, M. Palma, F. Pittoni, M. Giunta, and
of applications. There have been numerous regulatory and F. Ballesio, “Powerline technology over coaxial cables for in-home
standardization efforts over recent decades and a programme multimedia applications: Performances and EMC issues,” in IEEE Intl.
Symp. on Power Line Commun. and Its Appl. (ISPLC), 2011.
of work by the research community has addressed different [13] S.-C. Yeh, C.-K. Lin, and H. H. Chen, “Available bandwidth charac-
challenges, making great advances on the use of a channel teristics and estimation of in-home power line networks,” in IEEE Intl.
not initially designed for data communication. Symp. on Power Line Commun. and Its Appl. (ISPLC), 2008.
[14] C.-K. Lin, H.-W. Chu, S.-C. Yen, M.-T. Lu, J. Yao, and H. Chen,
In this article we have reviewed germane contributions and “Robust video streaming over power lines,” in IEEE Intl. Symp. on
stated the main results in the literature, for both NB and Power Line Commun. and Its Appl. (ISPLC), 2006.
BB systems. We have considered standardization, channel [15] P. Degauque, I. Stievano, S. Pignari, V. Degardin, F. Canavero,
characterization and modeling, as well as physical and higher F. Grassi, and F. Canete, “Power-line communication: Channel charac-
terization and modeling for transportation systems,” IEEE Veh. Technol.
layer techniques defined in the different PLC standards. Mag., vol. 10, no. 2, 2015.
We have also highlighted areas of further study. Regard- [16] L. Lampe, A. Tonello, and T. Swart, Power Line Communications:
ing the physical layer, we have pointed out future research Principles, Standards and Applications from Multimedia to Smart Grid,
2nd ed. Chichester, UK: John Wiley & Sons, in print.
directions that include channel coding and signal process- [17] P. Tanguy, F. Nouvel, and P. Maziearo, “Power line communication
ing, mechanisms to ensure coexistence among different PLC standards for in-vehicule networks,” in Intl. Conf. on Intelligent Trans-
systems and among PLC and other communication technolo- port Syst. Telecom. IEEE, 2009, pp. 533–537.
gies, resource allocation in multicarrier systems, techniques [18] S. Barmada, M. Raugi, M. Tucci, and T. Zheng, “Power line communi-
cation in a full electric vehicle: Measurements, modelling and analysis,”
to extend coverage based on cooperation and relaying, and in IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC),
combining the use of PLC with wireless transmission using 2010.
diversity-combining techniques. On the higher layers, we have [19] M. Antoniali, A. M. Tonello, M. Lenardon, and A. Qualizza, “Mea-
surements and analysis of PLC channels in a cruise ship,” in IEEE Intl.
emphasized the need for further studies on the differences Symp. on Power Line Commun. and Its Appl. (ISPLC), 2011.
of the protocols defined for PLC networks compared to [20] V. Degardin, E. Simon, M. Morelle, M. Liénard, P. Degauque, I. Jun-
their analogs for wireless or sensor networks, mechanisms qua, and S. Bertuol, “On the possibility of using PLC in aircraft,” in
IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), 2010.
to resolve effects due to strict priority resolution, combined
[21] S. Barmada, A. Gaggelli, A. Musolino, R. Rizzo, M. Raugi, and
behavior of the medium access control and the physical layer M. Tucci, “Design of a PLC system onboard trains: selection and
dynamics and the integration with the networking ecosystem. analysis of the PLC channel,” in IEEE Intl. Symp. on Power Line
We believe research outcomes in these areas will increase the Commun. and Its Appl. (ISPLC), 2008.
[22] F. Minamiyama, H. Koga, K. Kobayashi, and M. Katayama, “Power
penetration of PLC in the years to come. supply overlaid communication and common clock delivery for coop-
erative motion control,” in IEEE Intl. Symp. on Power Line Commun.
R EFERENCES and Its Appl. (ISPLC), 2011.
[23] A. T. Sherman, D. Phatak, B. Sonawane, and V. G. Relan, “Location
[1] A. Pinomaa, J. Ahola, A. Kosonen, and P. Nuutinen, “HomePlug Green authentication through power line communication: Design, protocol,
PHY for the LVDC PLC concept: Applicability study,” in IEEE Intl. and analysis of a new out-of-band strategy,” in IEEE Intl. Symp. on
Symp. on Power Line Commun. and Its Appl. (ISPLC), 2015. Power Line Commun. and Its Appl. (ISPLC), 2010.
[2] V. N. Papilaya, K. Ouahadaz, A. Mengi, M. Weinand, M. Koch, et al., [24] L. Zhang, X. Liu, and D. Xu, “A novel security monitoring system
“Analysis of the devolo’s 500 kHz G3-PLC access technology based of coal mine based on power line communication dynamic routing
on smart grid field trials,” in IEEE Intl. Symp. on Power Line Commun. technology,” in Industry Applications Society Annual Meeting, 2014
and Its Appl. (ISPLC), 2014. IEEE. IEEE, 2014, pp. 1–6.
[3] A. Pinomaa, J. Ahola, A. Kosonen, and P. Nuutinen, “Applicability
[25] S. Tsuzuki and Y. Yamada, “Feasibility study of ubiquitous sensor
of narrowband power line communication in an LVDC distribution
networks by inductively coupled PLC over PV power systems,” in IEEE
network,” in IEEE Intl. Symp. on Power Line Commun. and Its Appl.
Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), 2015.
(ISPLC), 2014.
[4] A. Goedhart, R. Heymann, and H. Ferreira, “Adapting HomePlug C&C [26] A. de Beer, H. Ferreira, and A. Vinck, “Contactless power-line com-
PLC for use in a low voltage smart grid,” in IEEE Intl. Symp. on Power munications,” in IEEE Intl. Symp. on Power Line Commun. and Its
Line Commun. and Its Appl. (ISPLC), 2012. Appl. (ISPLC), 2014.
[5] A. Pinomaa, J. Ahola, and A. Kosonen, “Power-line communication- [27] S. Barmada, M. Raugi, and M. Tucci, “Power line communication
based network architecture for LVDC distribution system,” in IEEE integrated in a wireless power transfer system: A feasibility study,”
Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), 2011. in IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC),
[6] Y. He, N. Jenkins, J. Wu, and M. Eltayeb, “ICT infrastructure for smart 2014.
distribution networks,” in IEEE Intl. Symp. on Power Line Commun. [28] S. Galli, A. Scaglione, and Z. Wang, “For the grid and through the
and Its Appl. (ISPLC), 2010. grid: The role of power line communications in the smart grid,” Proc.
[7] J. Liu, B. Zhao, J. Wang, Y. Zhu, and J. Hu, “Application of power IEEE, vol. 99, no. 6, pp. 998–1027, Jun 2011.
line communication in smart power consumption,” in IEEE Intl. Symp. [29] R. Vick, “Radiated emission caused by in-house PLC-systems,” in Intl.
on Power Line Commun. and Its Appl. (ISPLC), 2010. Symp. on Power Line Commun. and Its Appl. (ISPLC), 2001, pp. 197–
[8] P. Mlynek, M. Koutny, J. Misurec, and Z. Kolka, “Measurements and 202.
evaluation of PLC modem with G3 and PRIME standards for street [30] S. Galli, M. Koch, H. Latchman, S. Lee, and V. Oksman, “Chap.
lighting control,” in IEEE Intl. Symp. on Power Line Commun. and Its 7: Industrial and International Standards on PLC-based Networking
Appl. (ISPLC), 2013. Technologies,” in Power Line Communications: Theory and Applica-
[9] G. Dickmann, “Digitalstrom : R A centralized PLC topology for home tions for Narrowband and Broadband Communications over Power
automation and energy management,” in IEEE Intl. Symp. on Power Lines, 1st ed., H. Ferreira, L. Lampe, J. Newbury, and T. Swart, Eds.
Line Commun. and Its Appl. (ISPLC), 2011. Chichester, UK: John Wiley & Sons, 2010.
 15

[31] European Committee for Electrotechnical Standardization (CEN- [52] ——, “ITU-T Recommendation G.9961, Data link layer (DLL) for
ELEC), “Signalling on low-voltage electrical installations in the fre- unified high-speed wire-line based home networking transceivers,” Jun
quency range 3 kHz to 148,5 kHz — Part 1: General requirements, fre- 2010. [Online]. Available: http://www.itu.int/rec/T-REC-G.9961
quency bands and electromagnetic disturbances,” Standard EN 50065- [53] M. Sun and D. Lun, “Power-line communications using DWMT
1, 1992, 2001, 2011,. modulation,” in IEEE Intl. Symp. on Circuits and Syst. (ISCAS), vol. 4,
[32] “IEEE standard for low-frequency (less than 500 kHz) narrowband 2002, pp. IV–493 – IV–496 vol.4.
power line communications for smart grid applications,” IEEE Std [54] S. Galli, A. Kurobe, and M. Ohura, “The inter-PHY protocol (IPP): A
1901.2-2013, pp. 1–269, 2013. simple coexistence protocol for shared media,” in IEEE Intl. Symp. on
[33] S. Ponzelar, A. Mengi, and M. Koch, “Electromagnetic compatibility in Power Line Commun. and Its Appl. (ISPLC), Mar 2009, pp. 194–200.
europe for OFDM based PLC access data transmission in the frequency [55] International Telecommunications Union (ITU), “ITU-T Rec-
range 150 kHz to 500 kHz,” in IEEE Intl. Symp. on Power Line ommendation G.9972, Coexistence mechanism for wireline
Commun. and Its Appl. (ISPLC), March 2015, pp. 195–198. home networking transceivers,” Jun 2010. [Online]. Available:
[34] FCC, “Title 47 of the code of federal regulations (CFR),” http://www.itu.int/rec/T-REC-G.9972
U.S. Federal Communications Commission (FCC), Tech. Rep. [56] ——, “ITU-T Recommendation G.9963, Unified high-speed wire-line
47 CFR §15. [Online]. Available: http://www.fcc.gov/encyclopedia/ based home networking transceivers — Multiple Input/Multiple
rules-regulations-title-47 Output (MIMO),” Sep 2011. [Online]. Available: http://www.itu.int/
[35] Association of Radio Industries and Businesses (ARIB), “Power line rec/T-REC-G.9963
communication equipment (10kHz-450kHz),” Nov 2002, sTD-T84, [57] L. Yonge, J. Abad, K. Afkhamie, L. Guerrieri, S. Katar, H. Lioe,
Ver. 1.0, (in Japanese). [Online]. Available: http://www.arib.or.jp/ P. Pagani, R. Riva, D. M. Schneider, and A. Schwager, “An overview of
english/html/overview/doc/1-STD-T84v1 0.pdf the HomePlug AV2 technology,” J. Electr. Comput. Eng., 2013, Article
[36] H. Hirsch and M. Koch, “Chap. 3: Electromagnetic Compatibility,” in ID 892628, doi:10.1155/2013/892628.
Power Line Communications: Theory and Applications for Narrowband [58] HomePlug Powerline Alliance, “HomePlug Green PHY specification,
and Broadband Communications over Power Lines, 1st ed., H. Ferreira, release version 1.0.0,” Jun 2010. [Online]. Available: http://www.
L. Lampe, J. Newbury, and T. Swart, Eds. Chichester, UK: John Wiley homeplug.org/tech-resources/green-phy-iot/
& Sons, 2010. [59] L. Lampe, A. Tonello, and D. Shaver (Guest Editors), “Power line com-
[37] European Committee for Electrotechnical Standardization (CEN- munications for automation networks and smart grid,” IEEE Commun.
ELEC), “Power line communication apparatus used in low-voltage in- Mag., vol. 49, no. 12, Dec 2011.
stallations — Radio disturbance characteristics — Limits and methods [60] J. Khan, A. Tonello, L. Vangelista, L. Lampe, A. Maeder, T. Godfrey,
of measurement — Part 1: Apparatus for in-home use,” Nov 2012. and S. Mohanty, “Chap. 4: Smart Grid Communication Technologies,”
[38] B. Praho, M. Tlich, P. Pagani, A. Zeddam, and F. Nouvel, “Cognitive in IEEE Vision for Smart Grid Communications: 2030 and Beyond,
detection method of radio frequencies on power line networks,” in IEEE S. Goel, S. Bush, and D. Bakken, Eds. IEE, 2013.
Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC). IEEE, [61] S. Galli and T. Lys, “Next generation narrowband (under 500 kHz)
2010, pp. 225–230. power line communications (PLC) standards,” China Communications,
[39] S. W. Oh, Y. L. Chiu, K. N. Ng, R. Mo, Y. Ma, Y. Zeng, and no. 2, pp. 1–8, Mar 2015.
A. Syed Naveen, “Cognitive power line communication system for
[62] A. M. Tonello, F. Versolatto, and A. Pittolo, “In-home power line com-
multiple channel access,” in IEEE Intl. Symp. on Power Line Commun.
munication channel: Statistical characterization,” IEEE Trans. Com-
and Its Appl. (ISPLC). IEEE, 2009, pp. 47–52.
mun., vol. 62, no. 6, pp. 2096–2106, Jun 2014.
[40] A. Schwager and L. Berger, “Power Line Communication Electro-
[63] S. Galli, “A simplified model for the indoor power line channel,” in
magnetic Compatibility Regulations,” in MIMO Power Line Commu-
Proc. of IEEE International Symposium on Power Line Communica-
nications: Narrow and Broadband Standards, EMC, and Advanced
tions and Its Applications (ISPLC), March 2009, pp. 13–19.
Processing. New York, NY: CRC Press - Taylor & Francis Group,
2014. [64] J. A. Cortés, F. J. Cañete, L. Dı́ez, and J. L. G. Moreno, “On the
[41] IEEE Std 1901, “Standard for broadband over power line networks: statistical properties of indoor power line channels: Measurements and
Medium access control and physical layer specifications,” ANSI/IEEE models,” in IEEE Intl. Symp. on Power Line Commun. and Its Appl.
Std 1901, 2010. (ISPLC), Apr 2011, pp. 271–276.
[42] International Telecommunications Union (ITU), “ITU-T Recommen- [65] S. Galli, “A novel approach to the statistical modeling of wireline
dation G.9964, Unified high-speed wireline-based home networking channels,” IEEE Trans. Commun., vol. 59, no. 5, pp. 1332–1345, May
transceivers — Power spectral density specification,” Dec 2011. 2011.
[Online]. Available: https://www.itu.int/rec/T-REC-G.9964 [66] M. Babic, M. Hagenau, K. Dostert, and J. Bausch, “D4: Theoretical
[43] O. Hersent, D. Boswarthick, and O. Elloumi, Eds., The Internet of postulation of the PLC channel model OPERA,” IST Integrated Project
Things: Ket Applications and Protocols, 2nd ed. Chichester, UK: No. 507667 funded by EC, Tech. Rep., Mar 2005, oPERA Deliverable
John Wiley & Sons, 2012. D4.
[44] D. Dzung, I. Berganza, and A. Sendin, “Evolution of powerline [67] A. M. Tonello, A. Pittolo, and M. Girotto, “Power line communications:
communications for smart distribution: From ripple control to OFDM,” Understanding the channel for physical layer evolution based on filter
in IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), bank modulation,” IEICE Trans. on Commun., vol. E97-B, no. 8, pp.
Apr 2011, pp. 474–478. 1494–1503, Aug 2014.
[45] H. Latchman and L. Yonge (Guest Editors), “Power line local area [68] F. J. Cañete, J. A. Cortés, L. Dı́ez, and J. T. Entrambasaguas, “Analysis
networking,” IEEE Commun. Mag., vol. 31, no. 4, Apr 2003. of the cyclic short-term variation of indoor power line channels,” IEEE
[46] S. Galli, A. Scaglione, and K. Dostert (Guest Editors), “Broadband J. Select. Areas Commun., vol. 24, no. 7, pp. 1327–1338, Jul 2006.
is power: Internet access through the power line network,” IEEE [69] F. Versolatto, A. M. Tonello, C. Tornelli, and D. D. Giustina, “Sta-
Commun. Mag., vol. 31, no. 5, May 2003. tistical analysis of broadband underground medium voltage channels
[47] HomePlug Powerline Alliance, “HomePlug 1.0 specification,” 2001. for PLC applications,” in IEEE Intl. Conf. on Smart Grid Commun.
[Online]. Available: http://www.homeplug.org/ (SmartGridComm), Nov 2014, pp. 493–498.
[48] M. K. Lee, R. E. Newman, H. A. Latchman, S. Katar, and L. Yonge, [70] M. Nassar, J. Lin, Y. Mortazavi, A. Dabak, I. H. Kim, and B. L. Evans,
“HomePlug 1.0 powerline communication LANs — protocol descrip- “Local utility power line communications in the 3–500 kHz band:
tion and performance results,” Int. J. Commun. Syst., vol. 16, no. 5, Channel impairments, noise, and standards,” IEEE Signal Processing
pp. 447–473, May 2003. Mag., vol. 29, no. 5, pp. 116–127, Sep 2012.
[49] S. Galli and O. Logvinov, “Recent developments in the standardization [71] J. A. Cortés, A. Sanz, P. Estopiñán, and J. I. Garcı́a, “Analysis of
of power line communications within the IEEE,” IEEE Commun. Mag., narrowband power line communication channels for advanced metering
vol. 46, no. 7, pp. 64–71, Jul 2008. infrastructure,” EURASIP J. on Adv. Signal Processing, vol. 2015, no. 1,
[50] V. Oksman and S. Galli, “G.hn: The new ITU-T home networking pp. 1–13, 2015.
standard,” IEEE Commun. Mag., vol. 47, no. 10, pp. 138 – 145, Oct [72] L. Lampe and A. J. H. Vinck, “Cooperative multihop power line
2009. communications,” in IEEE Intl. Symp. on Power Line Commun. and
[51] International Telecommunications Union (ITU), “ITU-T Recom- Its Appl. (ISPLC), Mar 2012, pp. 1–6.
mendation G.9960, Unified high-speed wire-line based home [73] A. Pittolo and A. M. Tonello, “Physical layer security in PLC networks:
networking transceivers — Foundation,” Jun 2010. [Online]. Available: Achievable secrecy rate and channel effects,” in IEEE Intl. Symp. on
https://www.itu.int/rec/T-REC-G.9960/en Power Line Commun. and Its Appl. (ISPLC), Mar 2013, pp. 273–278.
 16

[74] ETSI TR 101 562-1 V 1.3.1, “Powerline telecommunications (PLT); [97] M. Zimmermann and K. Dostert, “Analysis and modeling of impulsive
MIMO PLT; part 1: Measurement methods of MIMO PLT,” European noise in broad-band powerline communications,” IEEE Trans. Electro-
Telecommunication Standardization Institute, Tech. Rep., 2012. magn. Compat., vol. 44, no. 1, pp. 249–258, Feb 2002.
[75] D. Schneider, A. Schwager, J. Speidel, and A. Dilly, “Implementation [98] M. Gotz, M. Rapp, and K. Dostert, “Power line channel characteristics
and results of a MIMO PLC feasibility study,” in IEEE Intl. Symp. on and their effect on communication system design,” IEEE Commun.
Power Line Commun. and Its Appl. (ISPLC), 2011, pp. 54–59. Mag., vol. 42, no. 4, pp. 78–86, Apr 2004.
[76] A. Pittolo, A. M. Tonello, and F. Versolatto, “Performance of MIMO [99] M. Antoniali, F. Versolato, and A. M. Tonello, “An experimental
PLC in measured channels affected by correlated noise,” in IEEE Intl. characterization of the PLC noise at the source,” to appear IEEE Trans.
Symp. on Power Line Commun. and Its Appl. (ISPLC), Mar 2014, pp. Power Delivery, 2015.
261–265. [100] M. H. L. Chan and R. W. Donaldson, “Amplitude, width, and in-
[77] L. Stadelmeier, D. Schill, A. Schwager, D. Schneider, and J. Speidel, terarrival distributions for noise impulses on intrabuilding power line
“MIMO for inhome power line communications,” in Intl. ITG Conf. communication networks,” IEEE Trans. Electromagn. Compat., vol. 31,
on Source and Channel Coding (SCC), 2008, pp. 1–6. no. 3, pp. 320–323, Aug 1989.
[78] D. Schneider, J. Speidel, L. Stadelmeier, and D. Schill, “Precoded [101] F. Gianaroli, F. Pancaldi, E. Sironi, M. Vigilante, G. M. Vitetta, and
spatial multiplexing MIMO for inhome power line communications,” A. Barbieri, “Statistical modeling of periodic impulsive noise in indoor
in IEEE Global Telecom. Conf. (GLOBECOM), 2008, pp. 1–5. power-line channels,” IEEE Trans. Power Delivery, vol. 27, no. 3, pp.
[79] R. Hashmat, P. Pagani, and T. Chonavel, “MIMO communications for 1276–1283, Jul 2012.
inhome PLC networks: Measurements and results up to 100 MHz,” in [102] D. Middleton, “Statistical-physical models of electromagnetic interfer-
IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), Apr ence,” IEEE Trans. Electromagn. Compat., vol. 19, no. 3, pp. 106–127,
2010, pp. 120–124. Aug 1977.
[80] H. Phillips, “Modelling of powerline communication channels,” in [103] H. Meng, Y. L. Guan, and S. Chen, “Modeling and analysis of noise
IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), Mar effects on broadband power-line communications,” IEEE Trans. Power
1999, pp. 14–21. Delivery, vol. 20, no. 2, pp. 630–637, Apr 2005.
[81] M. Zimmermann and K. Dostert, “A multipath model for the powerline [104] M. Katayama, T. Yamazato, and H. Okada, “A mathematical model
channel,” IEEE Trans. Commun., vol. 50, no. 4, pp. 553–559, Apr 2002. of noise in narrowband power line communication systems,” IEEE J.
[82] A. M. Tonello, “Wideband impulse modulation and receiver algorithms Select. Areas Commun., vol. 24, no. 7, pp. 1267–1276, Jul 2006.
for multiuser power line communications,” EURASIP J. on Adv. Signal [105] J. A. Cortés, L. Dı́ez, F. J. Cañete, and J. J. Sánchez-Martı́nez,
Processing, vol. 2007, pp. 1–15, 2007. “Analysis of the indoor broadband power-line noise scenario,” IEEE
Trans. Electromagn. Compat., vol. 52, no. 4, pp. 849–858, Nov 2010.
[83] A. M. Tonello, F. Versolatto, B. Béjar, and S. Zazo, “A fitting algorithm
for random modeling the PLC channel,” IEEE Trans. Power Delivery, [106] M. Nassar, A. Dabak, I. H. Kim, T. Pande, and B. Evans, “Cyclosta-
vol. 27, no. 3, pp. 1477–1484, Jun 2012. tionary noise modeling in narrowband powerline communication for
Smart Grid applications,” in Proc. of IEEE International Conference
[84] M. Tlich, A. Zeddam, A. Moulin, and F. Gauthier, “Indoor power-
on Acoustics, Speech and Signal Processing (ICASSP), March 2012,
line communications channel characterization up to 100 MHz — part
pp. 3089–3092.
I: One-parameter deterministic model,” IEEE Trans. Power Delivery,
[107] L. di Bert, P. Caldera, D. Schwingshackl, and A. M. Tonello, “On
vol. 23, no. 3, pp. 1392–1401, Jul 2008.
noise modeling for power line communications,” in IEEE Intl. Symp.
[85] R. Hashmat, P. Pagani, A. Zeddam, and T. Chonave, “A channel model
on Power Line Commun. and Its Appl. (ISPLC), Apr 2011, pp. 283–
for multiple input multiple output in-home power line networks,” in
288.
IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC), Apr
[108] M. Tlich, H. Chaouche, A. Zeddam, and P. Pagani, “Novel approach
2011, pp. 35–41.
for PLC impulsive noise modelling,” in IEEE Intl. Symp. on Power
[86] J. S. Barnes, “A physical multi-path model for power distribution Line Commun. and Its Appl. (ISPLC), Mar 2009, pp. 20–25.
network propagation,” in IEEE Intl. Symp. on Power Line Commun.
[109] ETSI TR 101 562-3 V 1.1.1, “Powerline telecommunications (PLT);
and Its Appl. (ISPLC), Mar 1998, pp. 76–89.
MIMO PLT; part 3: Setup and statistical results of MIMO PLT channel
[87] O. G. Hooijen, “On the relation between network-topology and power and noise measurements,” European Telecommunication Standardiza-
line signal attenuation,” in IEEE Intl. Symp. on Power Line Commun. tion Institute, Tech. Rep., 2012.
and Its Appl. (ISPLC), Mar 1998, pp. 45–56. [110] D. Rende, A. Nayagam, K. Afkhamie, L. Yonge, R. Riva, D. Veronesi,
[88] H. Meng, Y. L. Guan, C. L. Law, P. L. So, E. Gunawan, and T. Lie, F. Osnato, and P. Bisaglia, “Noise correlation and its effect on capacity
“Modeling of transfer characteristics for the broadband power line of in-home MIMO power line channels,” in IEEE Intl. Symp. on Power
communication channel,” IEEE Trans. Power Delivery, vol. 19, no. 3, Line Commun. and Its Appl. (ISPLC), Apr. 2011, pp. 60–65.
pp. 1057–1064, Jul 2004. [111] A. Pittolo and A. M. Tonello, “Physical layer security in power line
[89] J. Anatory, N. Theethayi, and R. Thottappillil, “Power-line commu- communication networks: an emerging scenario, other than wireless,”
nication channel model for interconnected networks — part I: Two- IET Communications, vol. 8, no. 8, pp. 1239–1247, May 2014.
conductor system,” IEEE Trans. Power Delivery, vol. 24, no. 1, pp. [112] E. A. Jorswieck, A. Wolf, and S. Gerbracht, Trends in Telecommuni-
118–123, Jan 2009. cations Technologies. InTech, March 2010, ch. 20: Secrecy on the
[90] S. Galli and T. C. Banwell, “A deterministic frequency-domain model Physical Layer in Wireless Networks, pp. 413–435.
for the indoor power line transfer function,” IEEE J. Select. Areas [113] Y. Zhuang and L. Lampe, “Physical layer security in MIMO power
Commun., vol. 24, no. 7, pp. 1304–1316, Jul 2006. line communication networks,” in IEEE Intl. Symp. on Power Line
[91] T. Esmailian, F. R. Kschischang, and P. Glenn Gulak, “In-building Commun. and Its Appl. (ISPLC), Mar 2014, pp. 272–277.
power lines as high-speed communication channels: Channel charac- [114] H. C. Ferreira, A. J. H. Vinck, T. G. Swart, and I. de Beer, “Permutation
terization and a test channel ensemble,” Intern. J. of Commun. Syst., trellis codes,” IEEE Trans. Commun., vol. 53, no. 11, pp. 1782–1789,
vol. 16, no. 5, pp. 381–400, Jun 2003. Nov 2005.
[92] F. J. Cañete, J. A. Cortés, L. Dı́ez, and J. T. Entrambasaguas, “A [115] A. Tonello and F. Pecile, “Efficient architectures for multiuser FMT
channel model proposal for indoor power line communications,” IEEE systems and application to power line communications,” IEEE Trans.
Commun. Mag., vol. 49, pp. 166–174, Dec 2011. Commun., vol. 57, no. 5, pp. 1275–1279, 2009.
[93] A. M. Tonello and F. Versolatto, “Bottom-up statistical PLC channel [116] A. M. Tonello and M. Girotto, “Cyclic block FMT modulation for
modeling — part I: Random topology model and efficient transfer broadband power line communications,” in IEEE Intl. Symp. on Power
function computation,” IEEE Trans. Power Delivery, vol. 26, no. 2, Line Commun. and Its Appl. (ISPLC), Johannesburg, South Africa, Mar
pp. 891–898, Apr 2011. 2013, pp. 247–251.
[94] ——, “Bottom-up statistical PLC channel modeling — part II: Inferring [117] A. Tonello and G. Mauro, “Cyclic block filtered multitone modulation,”
the statistics,” IEEE Trans. Power Delivery, vol. 25, no. 4, pp. 2356– EURASIP J. on Adv. Signal Processing, vol. 2014, no. 1, 2014.
2363, Oct 2010. [118] N. Andreadou and A. M. Tonello, “On the mitigation of impulsive noise
[95] H. C. Ferreira, L. Lampe, J. Newbury, and T. G. Swart, Power Line in power-line communications with LT codes,” IEEE Trans. Power
Communications. John Wiley & Sons Ltd, 2010. Delivery, vol. 28, no. 3, pp. 1483–1490, Jul 2013.
[96] M. Antoniali, A. M. Tonello, and F. Versolatto, “A study on the optimal [119] J. Cortés, A. Tonello, and L. Dı́ez, “Comparative Analysis of Pilot-
receiver impedance for SNR maximization in broadband PLC,” J. based Channel Estimators for DMT Systems Over Indoor Power-line
Electr. Comput. Eng., vol. 2013, p. 11, 2013. Channels,” in Proc. of IEEE International Symposium on Power Line
 17

Communications and Its Applications (ISPLC), March 2007, pp. 372– [143] I. S. 802.15.4, “Wireless LAN Medium Access Control (MAC) and
377. Physical Layer (PHY) Specifications for low-rate wireless personal area
[120] A. M. Tonello, F. Versolatto, and M. Girotto, “Multitechnology (I- networks,” IEEE Std 802.15.4, 2003 Edition (Revised 2006).
UWB and OFDM) coexistent communications on the power delivery [144] L. Zhang, X. Liu, Y. Zhou, and D. Xu, “A novel routing algorithm for
network,” IEEE Trans. Power Delivery, vol. 28, no. 4, pp. 2039–2047, power line communication over a low-voltage distribution network in
Oct 2013. a smart grid,” Energies, vol. 6, no. 3, pp. 1421–1438, 2013.
[121] J. Maes, M. Timmers, and M. Guenach, “Spectral compatibility of in- [145] M. Biagi and L. Lampe, “Location assisted routing techniques for
home and access technologies,” in IEEE Intl. Symp. on Power Line power line communication in smart grids,” in IEEE Intl. Conf. on Smart
Commun. and Its Appl. (ISPLC), Apr 2011, pp. 7–11. Grid Commun. (SmartGridComm). IEEE, 2010, pp. 274–278.
[122] K. Ali, G. Messier, and S. Lai, “DSL and PLC Co-Existence: An [146] M. Biagi, S. Greco, and L. Lampe, “Neighborhood-knowledge based
Interference Cancellation Approach,” IEEE Transactions on Commu- geo-routing in PLC,” in IEEE Intl. Symp. on Power Line Commun. and
nications, vol. 62, no. 9, pp. 3336–3350, Sept 2014. Its Appl. (ISPLC). IEEE, 2012, pp. 7–12.
[123] A. M. Tonello, S. D’Alessandro, and L. Lampe, “Cyclic prefix design [147] S. Galli and J. Le Clare, “Chap. 11: Narrowband Power Line Stan-
and allocation in bit-loaded OFDM over power line communication dards,” in MIMO Power Line Communications: Narrow and Broadband
channels,” IEEE Trans. Commun., vol. 58, no. 11, pp. 3265–3276, Nov Standards, EMC, and Advanced Processing. New York, NY: CRC
2010. Press - Taylor & Francis Group, 2014.
[124] A. M. Tonello, J. A. Cortés, and S. D’Alessandro, “Optimal time [148] S. Galli and T. Lys, “Next generation narrowband (under 500 kHz)
slot design in an OFDM-TDMA system over power-line time-variant power line communications (PLC) standards,” China Commun., vol. 12,
channels,” in IEEE Intl. Symp. on Power Line Commun. and Its Appl. no. 3, pp. 1–8, 2015.
(ISPLC), Mar 2009, pp. 41–46. [149] M. Kuhn, S. Berger, I. Hammerström, and A. Wittneben, “Power line
[125] L. Lampe, R. Schober, and S. Yiu, “Distributed space-time coding for enhanced cooperative wireless communications,” IEEE J. Select. Areas
multihop transmission in power line communication networks,” IEEE Commun., vol. 24, no. 7, pp. 1401–1410, 2006.
J. Select. Areas Commun., vol. 24, no. 7, pp. 1389–1400, Jul 2006. [150] X. Cheng, R. Cao, and L. Yang, “Relay-aided amplify-and-forward
[126] S. D’Alessandro and A. M. Tonello, “On rate improvements and power powerline communications,” IEEE Trans. Smart Grid, vol. 4, no. 1,
saving with opportunistic relaying in home power line networks,” pp. 265–272, 2013.
EURASIP J. on Adv. Signal Processing, vol. 2012, no. 1, p. 16, 2012. [151] P. Tinnakornsrisuphap, P. Purkayastha, and B. Mohanty, “Coverage and
[127] S. W. Lai and G. G. Messier, “Using the wireless and PLC channels capacity analysis of hybrid home networks,” in Intl. Conf. on Comput.,
for diversity,” IEEE Trans. Commun., vol. 60, no. 12, pp. 3865–3875, Networking and Commun. (ICNC). IEEE, 2014, pp. 117–123.
Dec 2012. [152] IEEE Std 1905.1-2013, “Standard for a Convergent Digital Home
[128] A. M. Tonello and A. Pittolo, “Considerations on Narrowband and Network for Heterogeneous Technologies,” ANSI/IEEE Std 1905.1,
Broadband Power Line Communication for Smart Grids,” in Proc. of 2013.
6th IEEE International Conference on Smart Grid Communications [153] E. Cohen, D. Ho, B. Mohanty, and P. Rajkotia, “Chap. 15: IEEE 1905.1:
(SmartGridComm), 2015. Convergent Digital Home Networking,” in Power Line Communica-
[129] C. Cano and D. Malone, “When priority resolution goes way too far: tions: Principles, Standards and Applications from Multimedia to Smart
An experimental evaluation in PLC networks,” in IEEE Intl. Conf. on Grid, 2nd ed., L. Lampe, A. Tonello, and T. Swart, Eds. Chichester,
Commun. (ICC), 2015. UK: John Wiley & Sons, in print.
[130] IEEE Std 802.11, “Wireless LAN Medium Access Control (MAC) and [154] IEEE Std 1905.1a-2014, “Standard for a Convergent Digital Home
Physical Layer (PHY) Specifications,” ANSI/IEEE Std 802.11, 1999 Network for Heterogeneous Technologies Amendment 1: Support of
Edition (Revised 2007). New MAC/PHYs and Enhancements,” ANSI/IEEE Std 1905.1a, 2014.
[131] M. Y. Chung, M.-H. Jung, T.-J. Lee, and Y. Lee, “Performance analysis [155] P. Wang, A. Marshell, K. Noordin, X. Huo, G. Markarian, et al.,
of HomePlug 1.0 MAC with CSMA/CA,” IEEE J. Select. Areas “Hybrid network combining PLC and IEEE802.16 for hospital envi-
Commun., vol. 24, no. 7, pp. 1411–1420, 2006. ronment,” in IEEE Intl. Symp. on Power Line Commun. and Its Appl.
[132] E. Kriminger and H. Latchman, “Markov chain model of HomePlug (ISPLC). IEEE, 2010, pp. 267–272.
CSMA MAC for determining optimal fixed contention window size,” [156] H. Ma, L. Lampe, and S. Hranilovic, “Integration of indoor visible light
in IEEE Intl. Symp. on Power Line Commun. and Its Appl. (ISPLC). and power line communication systems,” in IEEE Intl. Symp. on Power
IEEE, 2011, pp. 399–404. Line Commun. and Its Appl. (ISPLC). IEEE, 2013, pp. 291–296.
[133] C. Cano and D. Malone, “On efficiency and validity of previous [157] K. Razazian, A. Niktash, V. Loginov, and J. Yazdani, “Enhanced
Homeplug MAC performance analysis,” to appear Comput. Networks, 6LoWPAN ad hoc routing for G3-PLC,” in IEEE Intl. Symp. on Power
2015. Line Commun. and Its Appl. (ISPLC). IEEE, 2013, pp. 137–142.
[134] C. Vlachou, A. Banchs, J. Herzen, and P. Thiran, “Performance analysis [158] C.-K. Park, J.-M. Kang, M.-J. Choi, J. W.-K. Hong, Y.-h. Lim, S. Ju,
of MAC for power-line communications,” in ACM SIGMETRICS Intl. and M.-s. Choi, “Definition of common PLC MIB and design of MIB
Conf. on Meas. and Modeling of Comput. Syst., 2014. mapper for multi-vendor PLC network management,” in IEEE Intl.
[135] G. Bianchi, “Performance analysis of the IEEE 802.11 distributed Symp. on Power Line Commun. and Its Appl. (ISPLC). IEEE, 2008,
coordination function,” IEEE J. Select. Areas Commun., vol. 18, no. 3, pp. 152–157.
pp. 535–547, 2000. [159] C.-K. Park, J.-M. Kang, M.-J. Choi, J. W.-K. Hong, Y.-H. Lim,
[136] C. Cano and D. Malone, “Evaluation of the backoff procedure of S. Ju, and M.-S. Choi, “Development and testing of an SNMP-
Homeplug MAC vs. DCF,” in IEEE Intl. Symp. on Personal, Indoor based integrated management system for heterogeneous power line
and Mobile Radio Commun. (PIMRC), 2013. communication networks,” Intern. J. of Network Management, vol. 20,
[137] C. Vlachou, J. Herzen, and P. Thiran, “Fairness of MAC protocols: no. 1, pp. 35–55, 2010.
IEEE 1901 vs. 802.11,” in IEEE Intl. Symp. on Power Line Commun. [160] D. N. De Oliveira, T. L. Vasques, F. H. Vieira, G. A. de Deus Júnior,
and Its Appl. (ISPLC), 2013. M. S. De Castro, S. G. De Araújo, E. M. De Souza, J. G. Vieira, O. C.
[138] C. Vlachou, A. Banchs, J. Herzen, and P. Thiran, “On the MAC for de Oliveira Júnior, A. Borges, et al., “A management system for PLC
power-line communications: Modeling assumptions and performance networks using SNMP protocol,” in IEEE Intl. Symp. on Power Line
tradeoffs,” in 2014 IEEE 22nd International Conference on Network Commun. and Its Appl. (ISPLC). IEEE, 2010, pp. 78–83.
Protocols (ICNP), 2014, pp. 456–467. [161] G. Bumiller, “Powerline-channel adopted layer-design and link-layer
[139] A. Chowdhery, S. Jagannathan, J. M. Cioffi, and M. Ouzzif, “A polite for reliable data transmission,” in IEEE Intl. Symp. on Power Line
cross-layer protocol for contention-based home power-line communi- Commun. and Its Appl. (ISPLC). IEEE, 2015, pp. 256–261.
cations,” in IEEE Intl. Conf. on Commun. (ICC), 2009, pp. 1–6.
[140] B. Zarikoff and D. Malone, “Construction of a PLC test bed for network
and transport layer experiments,” in IEEE Intl. Symp. on Power Line
Commun. and Its Appl. (ISPLC), 2011, pp. 135–140.
[141] M. Meftah, L. Toutain, D. Ros, and A. Kortebi, “NS-2 model of
HomePlug AV PLC technology,” in Intl. ICST Conf. on Simulation
Tools and Tech., 2011, pp. 231–239.
[142] C. Cano and D. Malone, “Performance evaluation of the priority
resolution scheme in PLC networks,” in IEEE Intl. Symp. on Power
Line Commun. and Its Appl. (ISPLC), 2014.
 18

Cristina Cano obtained the Lutz Lampe (M’02SM’08)

Telecommunications Engineering received the Dipl.-Ing. and Dr.-Ing.
Degree at the Universitat Politecnica degrees in electrical engineering
de Catalunya (UPC) in February from the University of Erlangen,
2006. Then, she received the Erlangen, Germany, in 1998 and
M.Sc. (2007) and Ph.D. (2011) 2002, respectively. Since 2003, he
on Information, Communication and has been with the Department
Audiovisual Media Technologies of Electrical and Computer
from the Universitat Pompeu Fabra Engineering, The University of
(UPF). From July 2012 to December 2014, she was working British Columbia, Vancouver,
as a research fellow at the Hamilton Institute in the National BC, Canada, where he is a Full
University of Ireland, Maynooth (NUIM). Part of this research Professor. His research interests are broadly in theory and
group is now based in Trinity College Dublin (School of application of wireless, optical wireless, and power line
Computer Science and Statistics), where she was working communications. Dr. Lampe was the General (Co-)Chair
from January 2015 to January 2016 as a senior research for the 2005 International Conference on Power Line
fellow. Now she is with the FUN team at Inria-Lille. Her Communications (ISPLC), the 2009 IEEE International
research interests are related to coexistence of wireless Conference on Ultra-Wideband (ICUWB) and the 2013 IEEE
heterogeneous networks, distributed scheduling, performance International Conference on Smart Grid Communications
evaluation and stability analysis. (SmartGridComm). He is currently an Associate Editor of
the IEEE WIRELESS COMMUNICATIONS LETTERS
Alberto Pittolo received the Laurea and the IEEE COMMUNICATIONS SURVEYS AND
degree in electrical engineering TUTORIALS and has served as an Associate Editor and
(2009) and the Laurea Specialistica a Guest Editor of several IEEE transactions and journals.
degree (2012) in electrical and He was a (co-)recipient of a number of best paper awards,
telecommunications engineering, including awards at the 2006 IEEE ICUWB, the 2010 IEEE
with honors, from the University International Communications Conference (ICC), and the
of Udine, Udine, Italy. He is 2011 IEEE ISPLC.
currently completing the Ph.D.
degree in industrial and information Andrea M. Tonello (M00-
engineering at the University of SM12) received the Laurea degree
Udine. His research interests and (summa cum laude) and the PhD in
activities are channel modeling, physical layer security and electronics and telecommunications
resource allocation algorithms for both wireless and power from the University of Padova, Italy,
line communications. in 1996 and in 2003, respectively.
From 1997 to 2002, he was with
David Malone received B.A.(mod), Bell Labs- Lucent Technologies,
M.Sc. and Ph.D. degrees in Whippany, NJ, USA, first as a
mathematics from Trinity College Member of the Technical Staff.
Dublin. During his time as a Then, he was promoted to Technical
postgraduate, he became a member of Manager and appointed to Managing Director of the Bell
the FreeBSD development team. He Labs Italy division. In 2003, he joined the University of
is currently a member of Maynooth Udine, Italy, where he became Aggregate Professor in 2005
University’s Hamilton Institute and Associate Professor in 2014. He also founded WiTiKee,
and Department of Mathematics a spin-off company working on telecommunications for
& Statistics. His interests include smart grids. Currently, he is Professor and Chair of the
mathematical modeling and measurement of WiFi, PLC Embedded Communication Systems Group at the University
and password use. He also works on IPv6 and systems of Klagenfurt, Austria. Dr. Tonello received several awards,
administration, and is the is a co-author of O’Reilly’s “IPv6 including five best paper awards, the Bell Labs Recognition
Network Administration”. of Excellence Award (1999), the Distinguished Visiting
Fellowship from the Royal Academy of Engineering, U.K.
(2010), the IEEE VTS Distinguished Lecturer Award (2011-
2015), the Italian Full Professor Habilitation (2013). He is
the Chair of the IEEE Communications Society Technical
Committee on Power Line Communications. He serves as
Associate Editor of IEEE Trans. on Communications and
IEEE Access. He was the General Chair of IEEE ISPLC
2011 and IEEE SmartGridComm 2014.
 19

Anand G. Dabak received

the B.Tech. degree in electrical
engineering from the Indian Institute
of Technology, Bombay, India in 1987
and the M.Sc. and Ph.D. degrees in
electrical engineering, from Rice
University, Houston, TX, in 1989
and 1992, respectively in statistical
signal processing. After receiving the
Ph.D. degree, he worked for Viasat Inc., Carlsbad, California,
before joining Texas Instruments Incorporated, Dallas, TX,
in 1995. He has since then worked on the system and
algorithm issues related to communication systems including
wireless and powerline communications (PLC). He was
involved in wireless standardization and chipset solution
development activity for GSM, 3G, LTE, Bluetooth systems
and PLC standardization in IEEE 1901.2 and solutions on
TI MCU platforms. He is currently IEEE Fellow and TI
fellow and continues to work Ultrasound based water and gas
flowmetering and other signal processing applications that
can be implemented on TI platforms. He has published more
than 30 journal and conference papers and has more than 100
issued patents.