Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

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

Smart meter communication standards in Europe – a comparison

Sabine Erlinghagen a,n, Bill Lichtensteiger b, Jochen Markard a
a
Department of Management, Technology, and Economics, Group for Sustainability and Technology, Swiss Federal Institute of Technology (ETH) Zurich,
Weinbergstrasse 56/58, 8092 Zurich, Switzerland
b
Member of the IEEE, 2001L Street, NW. Suite 700, Washington, DC 20036-4910, USA

art ic l e i nf o a b s t r a c t

Article history: Smart grids and smart meters are central for the integration of distributed, renewable electricity
Received 21 April 2014 generation. For smart grids to operate smoothly, technology standards are essential. In Europe, a
Received in revised form multitude of different standards have been developed, making standard selection a complex task for
2 November 2014
electric utilities. With this paper we make three contributions: firstly, we identify 17 wired and wireless
Accepted 8 November 2014
communication standards for smart metering in Europe. Secondly, we develop and describe a
comprehensive set of technical and non-technical criteria for standard selection. Finally, we compare
Keywords: the existing standards against these criteria. We find that none of the standards is superior in all aspects,
Smart meter resulting in major trade-offs for decision makers. Moreover, we reveal the competing nature and missing
Communication technology
interoperability of standards, which makes a seamless integration of smart grids across different regions
Standards
difficult and poses major challenges for the development of complementary smart grid technologies.
& 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Smart meter technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.1. Use cases of smart meters in smart grids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.2. Introduction to smart meter communication technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.3. Communication standards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.4. Status of smart meter diffusion in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4. Criteria for standard comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4.1. Technical criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4.2. Non-technical criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5. Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5.1. Technical comparison of smart meter communication standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5.1.1. Narrowband PLC standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
5.1.2. Wireless standards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5.1.3. Application standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
5.2. Non-technical comparison of smart meter communication standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.2.1. Current and expected installed base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.2.2. Standard openness and EU endorsement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.2.3. Ownership of communication networks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.2.4. Costs and mode of meter replacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.3. Summary of comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6. Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Appendix A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

n
Corresponding author.
E-mail address: serlinghagen@ethz.ch (S. Erlinghagen).

http://dx.doi.org/10.1016/j.rser.2014.11.065
1364-0321/& 2014 Elsevier Ltd. All rights reserved.
1250 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
1. Introduction
Smart grids are a central element of the ongoing energy
transitions as they facilitate the integration of distributed, renew-
able electricity generation [1–3]. Smart meters are household
meters with two-way communication technology. They are often
considered the first step in building a smarter grid [1].
For large numbers of smart meters to operate smoothly,
technology standards for data exchange are essential [4–7].
Standards can be expected to enable interoperability, reduce costs
through economies of scale and create mass markets [6,8]. More
generally, common standards facilitate the diffusion of new
technologies and the development of entire technological fields
[9,10]. Classical examples are technology standards for computers
[11] or mobile phones [12].
In emerging technological fields, it can be very challenging to
arrive at a common standard and the literature on standardization
reports many cases of firms and governments struggling about
standards. The battle of VCR standards with VHS winning over
Sony's Betamax [13], the success of Blu-ray in the case of high
definition DVDs [14] or the struggle of IBM, Apple and Sun for
dominance in the field of IT platforms [11] are prominent
examples.
Also in the case of smart grids, different standards are asso-
ciated with different strategic interests. Technology providers such
as Landis þGyr, Echelon or Connode, utilities such as Enel, ERDF or
Iberdrola and even governments are involved in standard devel-
opment [10]. Today, 17 different smart meter standards exist in
Europe. What makes this situation even more challenging is that
the entire energy sector is currently changing – as a consequence
of market liberalization and the recent rise of renewable energies
in many countries. In fact, the ongoing transformation of the
energy sector can be viewed as a sustainability transition [15],
which depends on the simultaneous development of complemen-
tary technologies, including new renewable energies, energy
storage, energy efficiency and smart grids.
In this situation of general uncertainty about sector develop-
ment and many competing standards, utilities have a hard time
choosing a smart meter standard. Standard selection depends on
the intended use of smart meters and on different criteria such as
costs, data rates, robustness or standard openness.
A comprehensive review of technical and non-technical criteria
for smart metering and a comparison of standards against these
criteria are still missing. So far, scholars have provided in-depth
analyses of the performance of individual standards such as
PRIME, G3, GSM, DLMS/COSEM or oneM2M [16–21]. Existing
studies also discuss the applicability of standards from other
domains for smart metering [21–23]. Other studies discuss the
opportunities and challenges of smart meters and smart grid
communication in more general terms [24,25]. Existing reviews
take a broad perspective, comparing communication standards
independent of their application. They survey communication
standards used for several smart grid applications [5,7,26] and
they only focus on technical criteria.
In this paper we specifically focus on communication standards
used for smart metering and we compare them against technical
and non-technical criteria. In doing so, we make three contribu-
tions: firstly, we identify standards used for smart meter commu-
nication in Europe. Secondly, we provide a comprehensive review
of technical and non-technical selection criteria relevant for smart
metering. Finally, we compare the identified standards against
these criteria. The results provide utilities and other stakeholders
with greater transparency for standard evaluation and selection.
This paper is structured as follows: we start by introducing the
technological field, including a description of the main use cases.
In Section 3, we then describe the methodological approach for
the identification of standards, the review of criteria and the
comparison of standards against these criteria. Section 4 intro-
duces the criteria for the comparison. The results of the compar-
ison are described in Section 5. Section 6 discusses the findings
and concludes.
2. Smart meter technology
2.1. Use cases of smart meters in smart grids
The main purpose of household electricity meters is billing –
meters are the cash registers for utilities. Smart meters are
equipped with two-way communication technologies. In the first
place, two-way communication enables automatic, remote
(instead of manual) meter readings for utilities [21]. This saves
manual reading efforts and enables utilities to send regular (e.g.
monthly) bills based on actual consumption rather than on
estimates.
Detailed information about electricity consumption from smart
meters can also help utilities to identify and stop fraud (energy
theft). Smart meters provide information on actual consumption
in intervals between one day and 15 min. An analysis of this data
enables utilities to identify irregular consumption patterns [5] that
are indicators for fraud [27]. Moreover, the two-way communica-
tion allows utilities to remotely disconnect meters [21,24] not only
in cases of fraud or unpaid bills but also in emergency situations
(e.g. risk of a black-out).
With the aid of smart meters utilities can also provide con-
sumers with real-time information on their electricity consump-
tion. This consumption data can be displayed on so-called in-home
displays, consumer internet portals or mobile apps [28] that help
consumers to identify energy saving potentials [24]. Studies have
shown that consumers reduce their energy consumption by 5–15%
when they use such displays and portals [28,29].
Beyond these consumer-oriented use cases, smart meters can
also help utilities to integrate increasing amounts of distributed,
renewable energy production in the electricity grid. Especially
fluctuating electricity production can lead to imbalances between
demand and supply. Utilities typically have little information of
what is going on in the distribution part of the grid. Smart meters
can help monitor and provide near-real-time information on
actual distributed production and consumption [24].
Moreover, smart meters can enable utilities to influence
demand as they allow introducing flexible tariff schemes or other
advanced demand response measures. Today, the electricity price
for households is mostly the same every day and throughout the
day. In some cases, a lower tariff applies during the night. With
smart meters utilities can change pricing schemes anytime and
remotely [21]. In some cases, utilities might not want to rely on
consumers' reaction to price incentives. Instead, they might seek
to control industrial, commercial and household loads directly. For
such demand response programs utilities need to know how much
load is available for them to reduce and they need to verify the
actual reduction of loads. Smart meters can provide this transpar-
ency and proof [24].
Finally, smart meter communication can be used in even more
advanced cases connected to other domains such as transport,
comfort, security or healthcare. The communication channel to the
meter could be used to remotely manage heating or ventilation or
to send alarm signals in case of emergencies. Each of the above-
described use cases has particular requirements in terms of
quantity, speed, latency and reliability of data transfer. A central
step when choosing a communication technology standard is
therefore to select and specify the applications for which smart
meters shall be used.
 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
2.2. Introduction to smart meter communication technologies
Smart meter communication networks typically consist of two
types of communication links: the wide area network (WAN) and
the local area network (LAN).1 In some cases, a third type of
communication link is added, the home area network (HAN)
[1,30]. The HAN enables communication from the meter into the
home, for example, to inform consumers about actual consump-
tion. The LAN establishes the communication link between the
meter and a data concentrator, e.g., in a substation. The WAN is
used to transport the information from the data concentrator to a
utility back-end system (such as a billing system). The require-
ments for all three types differ substantially especially with
respect to the distances covered and the data volumes. Therefore
different technological options may be considered for each type of
communication link. The most controversial debate around smart
metering concerns the LAN (and if needed the HAN) connection. A
great variety of possible technologies and standards exist, each
with advantages and disadvantages.
The LAN connection for smart meters can be established using
wired or wireless technologies. Our review will include standards
for both. In terms of wired technologies, the focus lies on power
line communication (PLC) technology, the most widely used
technology for wired LAN connection of meters in Europe.2
PLC technology uses the power line as communication channel,
modulating the communication signal onto the power line in a
defined frequency band. A dedicated PLC network needs to be
established for smart metering. It consists of communication
modules that are typically integrated into the meters and data
concentrators, typically installed in substations [21]. The data
concentrator sends and receives communication signals from all
meters connected to its substation.
PLC technologies have an inherent trade-off between band-
width and range. While broadband technologies achieve higher
data rates, they do not cover large distances. Additional repeaters
need to be installed to transport the signal over longer distances,
which is costly. For the connection of smart meters in Europe,
narrowband PLC technologies are therefore typically chosen (for
pilots with broadband power line communication see [31]). Other
wired technologies (such as optical technologies) are used only in
exceptional cases, as they require additional wiring or can be
disconnected by consumers too easily (e.g. by disconnecting the
internet router at home).
In terms of wireless technologies the review includes point-to-
point technologies (mobile communication), point-to-multipoint
(star topology) and radio mesh networks. These technologies
modulate signals onto radio waves in defined frequency bands.
While point-to-point and point-to-multipoint technologies use
existing network infrastructures such mobile phone networks or
TV and radio antennas, radio frequency (RF) mesh networks
require dedicated, field-based infrastructure such as data collec-
tors or data concentrators.
RF mesh technologies also differ from the other two wireless
technologies in terms of network topology. RF mesh technologies
form a network where meters route information to a neighboring
meter that is not necessarily directly connected to a data collector.
This is also referred to as dynamic multi-hop communication. Like
PLC technologies RF mesh technologies form a dedicated LAN,
which is connected to the WAN through data concentrators.
In contrast, point-to-point and point-to-multipoint technolo-
gies span the LAN and the WAN as each meter is directly linked to
1
LANs are also referred to as neighborhood area networks (NANs).
2
Note that the actual performance of a communication link is driven by
general characteristics of the technology and on the particular specifications of the
standard in a given geographical environment.
1251
Smart Data
meter collector
Application Application
Presentation Presentation
Session Session
Transport Transport
Network Network
Data Link Data Link
Physical Physical
Fig. 1. OSI layer model for smart meter communication.
a central antenna or base station. This has implications for the
network design, as each meter has to be reached by an antennas or
cellular base stations. For houses in remote locations and for
meters in cellars of buildings this can cause problems.
2.3. Communication standards
Both, wired and wireless technologies have been specified
through standards. Standards can be defined as “rule(s) for com-
mon and voluntary use, decided by one or several people or
organizations.” [32]. This broad definition includes both open and
proprietary standards. Communication technology standards, in
particular, define rules or conventions for the information
exchange between devices (here: meters and data concentrators
or base stations) in a network. These rules include specifications of
syntax, semantics and the synchronization of communication in
so-called protocols. Typically, such protocols contain rules for data
and address formats for data exchange, address mapping, routing,
detection of transmission errors, direction of information flow etc.
Communication protocols need to be interchangeable to func-
tion in diverse settings. For this purpose they are typically
structured with layers. The open systems interconnection (OSI)
model is a widely used standard (ISO/IEC 7498-1) for structuring
protocols. The OSI model consists of seven layers (see Fig. 1). The
lower layers (physical and data link layer) specify signal modula-
tion on the physical communication medium (fiber cable, power
line, radio frequency (RF), etc.), the upper layers (session, pre-
sentation and application layer) determine the functionality. The
layers in-between enable the transport.
For a communication link to be established (e.g. between meter
and data concentrator) the entire OSI stack needs to be specified.
This can be achieved in two ways: Either a standard specifies all
layers or several standards, which each specify a subset of layers,
are combined. Such combinations only work if the standards are
specified in a compatible way. For this review, a mapping of
standards to the OSI model will allow us to identify competing
standards as well as complementarities.
2.4. Status of smart meter diffusion in Europe
The EU has set a goal to have 80% of European households
equipped with smart meters by 2020. But only few countries have
achieved this target so far. Italy was the first country to roll-out
smart meters in 2001, even before the EU set its goals. Sweden
followed with its roll-out between 2003 and 2009. Malta and
Finland have finalized their roll-outs in 2013. Governments in
several other countries, including France, Spain, UK and Nether-
lands, have set targets for the implementation of smart meters, but
have not finalized the roll-out yet [33]. Germany has decided to
install smart meters only in new or substantially renovated houses
1252 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Fig. 2. Number of installed smart meters (vs. non-smart meters) in selected European countries (2012) based on [34].
as well as in high-consumption households. Germany is likely not
to realize a comprehensive roll-out. Again others (Belgium, Czech
Republic and Lithuania) have decided against a smart meter roll-
out, based on a negative cost-benefit analysis [33]. Fig. 2 provides
an overview of the smart meter penetration in major European
countries in 2012.
Typically, national regulation makes the distribution system opera-
tors (DSOs) responsible for metering and consequently also for the
roll-out of smart meters. In exceptional cases, such as the UK, the
retailer is responsible for metering. Within the limits of government
mandates DSOs (or retailers) have to determine the functional and
non-functional (e.g. security) requirements for smart metering and
select the communication standard. These requirements depend
among others on the targeted use cases as introduced above.
3. Methods
In order to identify the most relevant smart meter standards,
we analyzed a comprehensive database containing information on
more than 200 smart meter projects in Europe. This database was
compiled in an earlier study (please refer to [34] for more detail on
the data sources) and, among others, contained information on the
standard(s) used in each smart meter project. This information
allowed us to generate a list of all deployed standards. To include
only the most relevant standards, we limited the scope to
standards deployed (or to be deployed) in projects larger than
50,000 m endpoints. As a result we identified 11 standards in
scope of the study. Through a review of standardization activities
in major SDOs, we identified another six standards, which have
not been deployed in larger projects so far due to their novelty.
The 17 standards included in this review are of three types:
wired PLC standards (eight standards), wireless standards (seven
standards) and standards that specify the application layer only
(two standards). Table 1 provides an overview of the standards.
Note that several PLC and wireless standards (i.e. Meters&More,
LON/OSGP, Meshnet3, Flexnet and KamstrupRF) also specify the
application layer (cf. Table 4).
To develop assessment criteria and to compare the standards,
we proceeded in two steps: in a first step, we analyzed secondary
data to compile preliminary criteria and an initial comparison. In a
second step, we gathered primary data through expert interviews,
which helped us to validate the criteria and refine the comparison.
Table 1
Standards included in this study.
Type Standard name
Wired (PLC) PLAN, AMIS, Meters&More, LON/OSGP, PRIME, G3, G.9902, 1901.2
Wireless GSM/GPRS, UMTS, LTE, MeshNet3, Flexnet, KamstrupRF, 802.15.4
Application* DLMS/COSEM, oneM2M
n
Standards that specify only the upper layer of the OS.
The analysis of secondary data was based on a comprehensive
document analysis. It included publications in peer-reviewed
journals, academic conference proceedings, technical specifica-
tions of standards, web sites of standards developers and industry
reports.3 This analysis allowed us to establish a first list of criteria.
Moreover, it provided us with first information how the selected
standards compare against these initial criteria.
In a second step, we conducted 14 interviews with experts,
who were involved in the development of the standards under
analysis. We selected experts to have at least one interviewee per
standard, ideally two. Note that most experts were involved in
more than one standard development committee (cf. Table 2).4
This multiple engagement of experts was also very helpful to
compare standards directly. The informants worked for utilities,
technology suppliers, independent research institutes or consult-
ing firms. The interviews were conducted between October 2012
and April 2013. They lasted 30–60 min and were conducted in
person (9/14) where possible and by phone otherwise. Extensive
notes were taken during each interview.
3
For the comparison of current and expected installed base (see non-technical
criteria in Section 4.2), we analyzed project data in the above mentioned project
database to determine the installed base in 2012 and the expected installed base in
2020. For each project, the database contained information on the standard
deployed, the project end date, the total amount of meters and the meters installed
by 2012. Based on this information, we determined the number of meters per
standard installed in 2012. To determine the expected installed base we assumed
for each project that was not completed by 2012 that an equal share of the
remaining meters would be installed each year until the planned end date of the
project. Since none of the projects had an end date later than 2020, this method
could be applied for all projects.
4
This was true for all open standards that were developed collectively by
several organizations. For the few standards that were developed by a single
organization (e.g. Kamstrup RF) experts only participated in the development of
their own standard.
 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262 1253

Table 2 The reach inside buildings and the achievable range are impor-
Overview of interviewed informants. tant criteria because they determine whether each node (here:
meter) can be reached. In Europe, meters are often located inside
No. Employer Interview focus
of buildings (often even in cellars) and might therefore be difficult
1 Technology provider CENELEC standards to reach, e.g., for weak radio signals. The range determines the
2 Technology provider Kamstrup RF maximum distance between sender and receiver (here meters and
3 Technology provider OSGP, LON data concentrators5) in a network. The network needs to be
4 Technology provider PRIME, G3, 1901.2
5 Technology provider MeshNet, IEEE802.15.4g/e
established in a way that each meter can be reached in a cost
6 Technology provider GSM, UMTS, LTE, oneM2M effective way [24]. If the communication standard covers a too
7 Technology provider AMIS short range, additional data concentrators or additional repeaters
8 Independent research institute CENELEC standards are required. Both substantially increase the costs for communica-
9 Independent research institute OSGP, LON
tion per meter. This distance between meter and substation can be
10 Independent consultant G3
11 Utility Meters&More very different, e.g., in rural and urban areas.
12 Utility PRIME Thirdly, communication standards have to work in frequency
13 Utility PRIME bands available to the utility. The usage of both power line and
14 Utility G3 radio frequency bands is regulated internationally. Some fre-
quency bands are reserved for specific entities (such as utilities
or police forces) whereas others are restricted to certain applica-
Each interview contained two parts, a general part about tions (such as low voltage network monitoring). Ideally, the utility
standard selection criteria and a specific part on the standard has the exclusive usage rights for the band they use for smart
(s) for which the interviewee possessed expert knowledge. After metering. Such exclusivity prevents congestions or interference
each interview we updated our comparison with the new infor- through other users and reduces sources for errors. But such
mation and took this as a basis for the next interview. If new exclusivity can come at high costs as substantial license fees can
information from an interview yielded contradictions, we went be involved (e.g. licenses for frequency bands for mobile phone
back by phone or email to previous interviewees for clarification. communication). Details on European frequency band regulations
These interviews enabled us to complement and triangulate for powerline communication are explained in Box 1, for radio
the preliminary findings from the document analysis. Not only did frequencies in Box 2 in Appendix A.
they allow us to fill remaining gaps, they also helped us to refine The fourth criterion concerns data rates and latencies. Data
our criteria and the assessment of standards against these criteria. rates are a measure of how much data can be transmitted
Experts mainly added information on technical characteristics of per second. Latencies describe the time between sending and
standards, especially in the case of proprietary standards. They receiving of a communication signal. With higher data rates,
also provided us with additional documents with technical stan- latencies decrease proportionally. A communication standard
dard specifications such as mappings to the OSI model. Moreover, should achieve sufficient data rates and latencies [5] to fulfill
they helped us to define the boundaries for the categories in the functional and non-functional (e.g. security) requirements. Differ-
overview table (cf. Table 4). Finally, the interviews were helpful to ent smart grid functions have different latency requirements,
clarify contradicting findings from the document analysis, e.g., on which will affect the choice of communication technology
achievable data rates or openness. deployed in the LAN (and WAN). Table 3 provides an overview of
To summarize our results, we aggregated the results of our some typical latency requirements for smart meter and grid
comparison in a table (cf. Table 9). Each standard was assessed applications. While requirements for billing only are not very high
relative to the highest and lowest empirically observed values for (1–2 kbps and 30 s), more advanced functional and non-functional
each criterion. We distinguished three categories: ‘ þ þ ’ for the (security) requirements need higher data rates and lower latency.
highest values ‘ þ’ for the values in the middle and ‘o’ for the Another important criterion for standard selection is robust-
lowest values. For the criterion ‘data rates’, for example, we ness. The communication to and from the node needs to be reliable
categorized data rates above 1Mbps as ‘þ þ’ and data rates below in order to transmit the data without errors [5]. This is already
10kbps as ‘o’. important when ‘only’ billing is concerned and it is even more
crucial for more advanced functionality. Especially, if meters are
used for real time functions like advanced demand response or
4. Criteria for standard comparison renewable integration, communication errors can result in lost
benefits, penalty payments or destabilized networks. Different
The choice of smart meter communication standards depends network architecture and modulation techniques have proven to
on technical and non-technical criteria. We identified seven be more or less robust. Mesh networks are typically more robust
technical and five non-technical criteria. While this classification than star networks as the signal can use multiple ways to reach its
might be debatable for some criteria, it highlights the relevance of destination. Modulation techniques contribute to robustness in
non-technical aspects in decision-making, which were neglected two ways: by reducing interference from side bands (esp. OFDM
in many previous studies. modulation) and by providing increased data rates allowing for
the specification of additional robustness techniques such as
channel encoding, interleaving or data repetition. Moreover, in
4.1. Technical criteria shared frequency bands, quality of data transmission is increased
when standards specify features for the coordination of multiple
The seven technical criteria include reach inside buildings, users (in the MAC layer of the OSI model).
range, frequency bands, data rates/latencies, robustness, intero- Utilities will also look for interoperability [30] in lower and upper
perability and semantics/data models. While the first five criteria layer standards. Three types of interoperability can be distinguished:
are relevant for lower layer standards (cf. OSI model), interoper-
ability is also important for application standards. The last criter-
ion, ‘semantics and data models’ is only relevant for application 5
In the case of point-to-point or point-to-multipoint networks the distance
standards. between meters and base stations/antennas.
1254 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Firstly, interoperability between different versions of a standard to
ensure seamless replacement of old meters with newer versions;
secondly, interoperability between implementations of the same
standard by different vendors. Such interoperability assures seamless
interaction between meters and communication modules from
different vendors in one network. It makes utilities less dependent
on one vendor. Thirdly, interoperability can be achieved among
different standards.
Finally, the fulfillment of functional depends on the application
layers (layers 5–7). The semantics and data models specified for the
application layer should be designed to allow for the implementa-
tion of all required functionality [20,30]. In the first place these use
cases will be electricity or energy specific, but in a second step
they might go beyond energy and also cover transportation (e.g.
electrical vehicles) or even applications like health.
4.2. Non-technical criteria
We include five non-technical criteria in our review: The
current and the expected installed base (2 criteria), the openness
of a standard, the ownership of the network, and costs.
A large installed base allows later adopters to benefit from
experiences made in earlier installations. A standard with a large
installed base can be considered field-proven and scalable. Beyond
this, the current and especially the expected installed base have
crucial influence on which standard eventually becomes domi-
nant. A large installed base typically creates bandwagon effects
leading new users to choose the standard with the highest
prospects of becoming dominant [35,36]. A large installed base
also creates economies of scale. Users can benefit from lower
prices (see cost criterion below) and long-term security of supply.
The openness and EU endorsement of a standard is another
relevant selection criterion [35,37]. Proprietary standards entail
the risk of high dependency on one vendor or even lock-in with a
single vendor. With open standards, many suppliers can offer the
standard. Standards can be considered open, if (1) anybody can
contribute to the development and if (2) the standard can be
implemented by anybody under so-called FRAND conditions (fair,
reasonable and non-discriminatory) [38,39] or even entirely
(license) free. Through the contribution of many parties in stan-
dard development, users (utilities) benefit from a broader knowl-
edge base. Moreover, utilities might also want to consider EU
endorsement of a standard. Under the EU mandate M/441, Eur-
opean standardization organizations have specified a smart meter
reference architecture and framework. The EU requires standards
to be open and compliant with this framework. Such EU endorse-
ments can increase legitimacy of a standard. This might be
important, for example, for end-user acceptance.
Utilities might also have preferences with regard to the own-
ership of the communication network. Some might prefer to own
and operate the network, others might be happy to outsource
both. For utilities, owning and operating a communication
Table 3
Overview of typical latency requirements for different smart grid functionality.
Functionality Typical latency
requirement
Service disconnect 1–10 min
Classical meter reading 10–30 s
Load control o5 s
Distriubtion automation 1–2 s
Substation protection and transformer o 5 ms
monitoring
Voltage phasor measurement in generation o 1 ms
network requires internal resources and competences. Outsour-
cing means relying on and managing the interface to a third party
and to pay for the provided services.
Finally, costs cannot be neglected as a criterion. Costs for
communication modules and for operation of the network can
differ. Module costs typically decrease significantly with large
market volumes (economies of scale, see above) and increase with
additional data rates.6 In addition, costs are influenced by the
mode of smart meter replacement (‘comprehensive roll-out’: all
conventional meters in one geographical area are replaced by
smart meters at once or ‘selective exchange’: smart meters are
installed only in households that fulfill certain criteria such as high
electricity consumption). Technology options that can rely on
existing network infrastructure are often more cost efficient in
case of selective exchange, whereas technology options that
require the establishment of dedicated communication networks,
typically yield advantages in comprehensive roll-outs.
Table 4 provides an overview of the above-described technical
and non-technical criteria. For each criterion it shows the oper-
ationalization for distinguishing the three categories (‘þ þ’, ‘þ’
and ‘o’). This operationalization is later used to aggregate the
findings of the standard comparison in one overview
table (Table 9). Only for the two criteria ‘ownership’ and ‘cost
effectiveness in different replacement modes’ we distinguish only
two categories: “yes” or “no”, as the ownership of such a network,
for example, was either possible or not.
In summary, we identified a broad range of different technical
and non-technical criteria. These criteria not only need to be
carefully assessed, they might also be partly contradicting, result-
ing in trade-offs to be made in standard selection. Moreover, we
see that the relevance of these technical and non-technical criteria
depends on the targeted use cases (see also Section 6). Some
‘basic’ use cases such as daily meter readings can be realized with
very low data rates for example, whereas more advanced use cases
such as demand response, require higher data rates.7 For utilities
with clear strategic priorities and target use cases these criteria
provide a basis for standard evaluation and selection even beyond
the 17 standards we analyzed for this paper.
5. Comparison
The following comparison reveals the strengths and weak-
nesses of each standard with regard to the identified technical
and non-technical criteria. The mapping of standards to the OSI
model provides an initial overview and helps identify alternative
and complementary standards.
5.1. Technical comparison of smart meter communication standards
The mapping of standards to the OSI layer model (cf. Section
2.3) not only allows us to distinguish application standards from
physical layer standards, it also helps to identify complementa-
rities and alternative standards. Standards can be considered
alternatives, if they specify the same layers.
Table 5 shows the mapping of all 17 standards to the OSI layer
model. We see that ten of the standards specify only the lower
layers (layer 1 and 2) that determine the physical communication
channel, two specify only the upper, application layers and five
specify both.
6
We reflected these two aspects, which are relevant to costs, in the criteria
data rates and installed base (both current and expected). We did not consider
them in the criterion cost to avoid overemphasis.
7
An assessment of standards relative to each use case and strategic priority of
utilities is beyond the scope of this paper.
 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Table 4
Categories for ranking of standards against technical and non-technical criteria.
Selection criteria Ranking
Categories
þþ
Technical
1 Reach nodes inside buildings Always
2 Range 45 km
3 Dedicated/licensed frequency band Dedicated license-free band
4 High data rates/low latencies 41 Mbps
5 Robustness Robust architecture and protocol
6 Interoperability Between different vendors
7 Horizontal data model/semantics Designed for many sectors
Non-technical
1 Current installed base 45mio
2 Expected installed base 45mio
3 Openness and M/441 EU endorsement Open and EU endorsed
4 Ownership of the network
5 Cost efficient for selective replacement
Cost efficient for comprehensive roll-out
Among the standards that specify the lower layers, eight
specify standards for power line communication (depicted in the
columns to the left of the table) PLAN, AMIS, PRIME or G3 are
examples here. These standards compete not only among each
other, but they also compete with seven wireless standards
including, for example, MeshNet3, Kamstrup RF or GSM/GPRS
(see columns to the right of the table).
In terms of application standards (layer 7), AMIS, Meters&More
DLMS/COSEM, LON/OSGP, oneM2M, MeshNet3 and Kamstrup RF
represent alternative solutions. While AMIS, Meters&More, LON/
OSGP, Meshnet and Kamstrup RF are specifically designed to work
with their respective physical and data link layers (layers 1 and 2).
DLMS/COSEM and oneM2M are designed to work with all lower
layer standards that use IPv4 or IPv6 as a protocol in the network
layer. These include all ten standards that specify only the lower
layers.8
In the following technical comparison we first compare PLC
standards against each other (see Section 5.1.1), we then compare
the seven wireless standards, starting with point-to-point stan-
dards, followed by RF mesh and point-to-multipoint standards
(see Section 5.1.2). Finally, we compare the application standards
(see Section 5.1.3). In Section 5.2 we compare all standards against
non-technical criteria.
5.1.1. Narrowband PLC standards
PLC standards have a long history in smart meter communica-
tion in Europe. The first generation dates back to the 1990s (P-LAN,
LON/OSGP and Telegestore now Meters&More9). The next genera-
tion has been developed in the mid-2000s (PRIME, G3, AMIS,
G.9902 and 1901.2). G.9902 and 1901.2 are the most recently
finalized standards, released in 2012.
Using PLC networks for smart metering, implies using the same
grid for the transportation of electrons and data. No additional
wiring is needed since the meter is connected to the grid in any
case. Therefore PLC systems are often considered a “cost-effective
and straight forward solution” [5]. When compared to wireless
8
Meshnet3 and PLAN do not use IP but they can also be used with DLMS/
COSEM via a specialized adapter.
9
Meters&More originated in the Telegestore project and was initially based on
LON technology.
1255
þ o
Generally yes Not always
1–5 km o 1 km
Dedicated licensed band Shared, unlicensed band
10–1000 kbps o 10 kbps
Robust architecture ( þ a) or None
Robust modulation/protocol ( þ m)
Backwards interoperability Not interoperable
Specific energy sector features Designed for energy sector
Deployed Piloted/not deployed
Planned deployement known No deployment plans decided
Open Proprietary
Utility can own network Utility cannot own network
Mostly yes Mostly no
technologies, PLC technologies also have advantages in reaching
meters inside homes or even in cellars.
All of these standards work in CENELEC A (3–95 kHz) frequency
band, which is reserved for distribution monitoring use (which
includes metering) in Europe (cf. Section 4.1). In addition,
Meters&More and G3 work in ARIB (10–450 kHz), the Japanese
band and FCC (10–490 kHz), the US band. G.9902 also works
in FCC.
In CENELEC A band, these standards achieve data rates
between 1–100 kilobytes per second (kbps). Higher data rates
can be achieved in the FCC band (up to 500 kbps). But using the
FCC band might imply more interference through shared
band usage.
The differences in data rates as well as differences in robustness
are influenced by the different modulation techniques defined in
the standards. PLAN uses Spatial Frequency Shift Keying (S-FSK)
modulation and AMIS uses Differential Code Shift Keying (DCSK).
Both achieve lower data rates (up to 3 kbps), but are quite robust
and in case of S-FSK relatively easy and cost effective to imple-
ment. LON/OSGP and Meters&More use Binary Phase Shift Key (B-
PSK) modulation which achieve higher data rates (up to 57 kbps).
The remaining four standards (G3, PRIME, 1901.2, G.9902) use
OFDM modulation. In general, OFDM modulation is more complex
to implement, in return it is more robust and achieves higher data
rates [16]. Moreover, the increased data rates allow for the
specification of additional features that further increase robust-
ness. These include techniques like channel encoding (e.g. forward
error correction), interleaving and data repetition. In addition,
G.9902 and 1901.2 enable coherent modulation. “The choice
between coherent modulation and non-coherent modulation
reflects complexity vs. performance tradeoffs.” [16].
For operation in shared (FCC) frequency bands, G3 and 1901.2
additionally specify co-existence or contention features (ISP in
1901.2 or CSMA in G3) that enable coordination between different
users of a communication channel.
In general, newer standards achieve higher data rates. Analog to
data rates, latencies (from backend system via data concentrator to
the meter) decrease with each generation of standards. They range
between approximately 20 s using PLAN and ten seconds for
newer standards. G3 is an exception, it ‘sacrifices’ higher data
rates and lower latencies in favor of higher robustness achieved
through multiple repetition of data transmission. Table 6 provides
1256 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Table 5
OSI layer mapping of smart meter LAN standards.
an overview of the major technical differences between the PLC
standards.
Narrowband PLC standards do not greatly differ in terms of
range. Distances of approximately one up to two kilometers
between data concentrator and meter can be covered. This is far
more than broadband PLC standards that typically cover only few
hundred meters in the low voltage network. The extended range
of narrowband PLC standards (in comparison to broadband PLC
standards) comes at the expense of achievable data rates.
Finally, we find that the standards are compatible and interoper-
able among different versions of the same standard. Moreover, PLAN,
LON/OSGP, Meters&More, PRIME and G3 have certification programs
within their respective alliances with the aim of achieving interoper-
ability between standard implementations of different vendors. How-
ever, interoperability between different standards is not given. Only G3
is designed to co-exist with PLAN.
5.1.2. Wireless standards
Wireless technologies generally perform better in outdoor
environments and are susceptible to signal loss in indoor environ-
ments, because radio signals often have difficulties penetrating
building walls and reaching meters when mounted in cellars. This
problem can be mitigated through the use of external antennas,
however at additional cost.
While point-to-point (mobile communication) networks and
RF mesh networks have been deployed for smart metering for
many years, point-to-multipoint networks were only recently used
for this purpose. In this section, we first separately compare point-
to-point standards, followed by RF mesh and point-to-multipoint
standards.
In terms of point-to-point networks we distinguish several
generations of mobile communication standards. GSM and GPRS
are typically referred to as 2nd generation standards, UMTS as 3rd
and LTE as 4th generation. These standards have been agreed
globally [36] and are deployed world-wide. The respective fre-
quency bands are exclusively reserved for mobile operators who
obtain licenses (cf. Table 6). The mobile operators can use them for
any kind of communication, the usage is not limited to a certain
domain (such as smart metering).
As in the case of PLC standards, different modulation techniques
are used to achieve higher data rates with each new generation (cf.
Table 7). GSM/GPRS achieves data rates up to 270 kbps and LTE
reaches several hundred Megabytes per second. Overall these data
rates are significantly higher compared to narrowband PLC standards
(see Table 6). Accordingly latencies are significantly shorter (few
milliseconds to 1 s) once the connection has been established.
The robustness of point to point mobile communication tech-
nologies is determined mainly by the coverage of the network and
the number of obstructions between base station and meter.10
Lower speed GPRS technology is generally more robust than higher
speed UMTS and LTE technologies. Lower frequency (e.g.
900 MHz) mobile networks enjoy better signal propagation and
penetration through walls and are generally more robust than the
higher frequency (1.8 GHz–2.5 GHz) networks.
Also in terms of ranges (here between base stations and
meters), point-to-point standards reach further than PLC stan-
dards. A signal can cover distances of 5–30 km depending on the
topography of the landscape.
Interoperability is an extremely important requirement in mobile
communication standards. They have to be backwards compatible not
only within the same standard but also between different standards.
Moreover, interoperability between different vendors is also achieved.
Otherwise newer end devices (such as mobile phones or meters)
could not use older network infrastructure and equipment from
different vendors could not be used in one network.
In terms of RF mesh standards three standards are compared:
Meshnet3, Kamstrup RF and 802.15.4g-e. Point-to-multipoint
standards have been even less present so far, recently however
the UK has decided to deploy a point-to-multipoint standard
called Flexnet.
RF mesh standards mainly differ in the frequency bands and
power levels they use. Kamstrup RF operates in the unlicensed
433 MHz band and the licensed 444 MHz band at 10 and 500 mW
respectively. Meshnet works in 869.400–869.650 MHz unlicensed
band at 50 mW. 802.15.4g-e are designed for use in multiple
10
Generally speaking mobile networks are targeted at capturing frequent high
data users in dense urban areas and may therefore have poorer coverage in
rural areas.
 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262 1257

Table 6
Overview technical characteristics of PLC standards.

Standard Modulation Data rates Frequency band IP? Other features

PLAN S-FSK 200 bit/s–2.4 kbps CENELEC A

AMIS DCSK 600 bits/s–3 kbps CENELEC A
LON B-PSK 3.6–57.6 kbps CENELEC A
Meters&More B-PSK 4.8–57.6 kbpsn CENELEC A, ARIB, FCC
PRIME OFDM 21.4–128.6 kbps CENELEC A iPv4 Tree routing
G3 OFDM 2.4–33.4 kbpsn CENELEC A, ARIB, FCC iPv6 Robust mode, mesh routing
1901.2 OFDM Approx. 80 kbpsn CENELEC A, ARIB, FCC iPv6 Coherent modulation
G.9902 OFDM Approx. 80 kbpsn CENELEC A, FCC iPv6 Coherent modulation

n
In CENELEC A.

Table 7
Overview technical characteristics of wireless point-to-point standards.

Standard Modulation Data rates Frequency band IP?

GSM/GPRS GMSK/TDMA Up to 270 kbps 900–1800 MHz iPv4

UMTS CDMA 14.4–84 Mbps down 2110–2170 MHz downlink iPv6
5.75–22 Mbps up 1920–1980 MHz uplink
LTE OFDM 326–1000 Mbps down 800–2600 MHz iPv6
86–500 Mbps up

frequency bands world-wide (see [40] for an overview) including The analyzed RF mesh and point-to-multipoint standards are
863–870 MHz for Europe. In this band it operates with power generally not interoperable between different vendors. A certifica-
levels of 25 mW. tion program for 802.15.4 is currently being established though.
While Kamstrup RF and Meshnet3 only achieve data rates up to Table 8 summarizes the major technical characteristics of RF mesh
9.6 kbps, they achieve relatively long ranges of up to 10 km in rural and point-to-multipoint standards.
areas. 802.15.4e-g achieves higher data rates (up to 500 kbps)
through more advanced modulation techniques, but it only covers
5.1.3. Application standards
ranges up to 1 km. More specifically, 802.15.4 g proposes three
In terms of application standards we include seven standards in
alternative modulations for the PHY layer multi-rate frequency
our comparison, two that only specify the application layer i.e.
shift keying (MR-FSK) PHY, multi-rate orthogonal frequency-
DLMS/COSEM and oneM2M [20,21]; and five that have been
division multi-plexing (MR-OFDM) PHY, and multi-rate offset
developed in conjunction with the respective lower OSI layer
quadrature phase shift keying (MR-O-QPSK) [41]. MR-OFDM the-
standards: LON/OSGP, AMIS Meters&More, Meshnet, KamstrupRF.
oretically achieves the highest data rate of up to 800 kbps [40,41].
They are therefore mostly deployed as one stack with their
The mesh architecture11 makes these standards very robust
respective lower layer specifications. LON/OSGP could also be used
(if the network contains enough nodes). Communication can be
with other physical channels, but we have not identified a (larger)
routed through alternative paths in case of congestion or broken
deployment. DLMS/COSEM is used as application layer for several
communication links [7,42]. The network quality increases with
PLC standards including PLAN, PRIME and G3. Moreover it is used
each additional meter in the same area [7]. This then also
with mobile communication standards. DLMS/COSEM does not
reduces the risk of using non-exclusive frequency bands. Using
(yet) work with RF mesh standards, but there is a New Work Item
RF mesh can therefore become a challenge in very sparsely
Proposal (NWIP) within the responsible working group of IEC to
populated areas, where only few meters can form the mesh
close this gap.
network.
These application standards are generally not interoperable. The
Finally, Flexnet, the only point-to-multipoint standard included
DLMS/COSEM data model, for example, is very rich covering far
in this review, achieves data rates up to 172 kbps. It uses the UK
more classes, objects and attributes than are actually required in
licensed frequency band of 412 and 422 MHz. Higher power levels
many countries. Consequently different implementations of this
of typically 2 W in this band provide extra range (up to 30 km) and
standard may not be interoperable with one another because one
robustness.
implementation may have included items missing in the other
implementation. In order to achieve interoperability, a base set of
classes and objects would have to be specified in the interoper-
ability specification and implemented by everyone on a manda-
11
For this mesh architecture to be created it requires not only a standardized tory basis. LON/OSGP, Meters&More and IDIS12 (for DLMS/COSEM)
PHY/MAC such as 802.15.4 g and e (layer 1 and 2 of the OSI stack). It also requires therefore provide a certification that ensures interoperability
additional standards for neighbor discovery, networking, routing, transport, net-
between different vendors. Similarly, the PRIME and G3 standard
work management and security (layers 3 and 4). Meshnet3 and Kamstrup RF
therefore have specifically designed layers 3 and 4. 802.15.4g-e can be used with specify their own (country-specific) configuration of DLMS/COSEM
layer 3 and 4 standards defined by Internet Engineering Task Force (IETF) that are to ensure interoperability of their PLC networks.
known as RFCs. These include standards for the IPv6 network layer and associated
networking schemes, appropriate routing protocol (e.g. RPL) transport protocols (e.
12
g. UDP, TCP) and relevant security mechanisms. IDIS ¼Interoperable Device Interface Specifications Association.
1258 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Most of these application standards use rather concrete, energy
sector specific data models and semantics. As LON/OSGP has its
routes in industry and building automation, data models and
semantics are more abstract and better suited for cross-domain
usage. oneM2M uses the most abstract data models semantics
making it independent from any specific sector, explicitly targeting
cross-domain usages [43]. The specification of oneM2M is driven
by cellular operators, but it is not finalized yet.
In summary, the comparison of wired and wireless standards
reveals major differences between standards and also inherent
trade-offs. None of the standards is per se superior from a
technical perspective. While, point-to-point (mobile) communica-
tion standards achieve highest data rates, they work in non-
exclusive, licensed frequency bands. In addition, they have chal-
lenges to reach meters inside buildings (esp. in cellars). In contrast,
powerline standards work in (exclusive) unlicensed bands. They
reach meters in cellars without problems, but they only achieve
lower data rates.
Overall, we can observe a trend towards increasing data rates
with each new generation of wired and wireless standards. But the
interoperability and compatibility between different standards
remains generally rather low, with the exception of mobile
communication standards.
5.2. Non-technical comparison of smart meter communication
standards
5.2.1. Current and expected installed base
The most widely adopted standard in Europe until today is
Meters&More or more precisely its predecessor Telegestore. The
wide adoption was achieved thanks to the early smart meter roll-
out of Enel in Italy in 2001. Beyond Italy, Meters&More will be
deployed by Endesa, a subsidiary of Enel, in Spain. LON/OSGP,
MeshNet, Kamstrup RF, GSM/GPRS and PLAN have mainly been
deployed in the roll-outs in the Nordic countries between 2003
and 2011. The largest deployment of GSM/GPRS is planned for the
central and southern part of the UK. The northern part will deploy
Flexnet. AMIS has been deployed by some utilities mainly in
Austria. PRIME is currently being rolled-out by Iberdrola in Spain,
by EDP in Portugal and by Energa-Operator in Poland. G3 is the
standard of choice for ERDF in France. To our knowledge, UMTS,
LTE, 802.15.4, 1901.2 and G.9902 have not been deployed in Europe
yet. Fig. 3 shows the current installed base of standards in 2012
and a prediction for 2020 in Europe.
In summary, none of these lower layer standards has reached
the installed base (430mio) of Meters&More until 2012. Looking
forward however, PRIME, G3 and GPRS/UMTS are likely to achieve
higher growth rates, reaching more than ten million meters
by 2020.
Also in terms of application standards, Meters&More is also still
the most widely deployed standard. But DLMS is likely to overtake
Meters&More very soon. The actual and planned deployments of
PLAN, G3, PRIME and GPRS/UMTS all use a version of DLMS for the
application layer. oneM2M has (to our knowledge) not yet been
deployed so far.
5.2.2. Standard openness and EU endorsement
Today, most standards claim to be open, but some differences
can still be observed. These differences, however, are not easy to
uncover [37]. As previously introduced (cf. Section 4.2), one way of
judging openness is to identify how many parties have contributed
to the standard development and who can implement the stan-
dard. This can be done by tracing back the development history of
each standard.
The standards in the scope of our study can be broadly
categorized into three different patterns of standard development
history. In a first group of standards, PLAN, DLMS, 1901.2 and
G.9955, have been developed within standard development orga-
nizations (SDOs) from the start. This implies that several parties
contribute to standard development and that the acceptance of
standards is a result of majority vote (mostly 71% threshold).
Moreover, it implies that standards can be implemented by any-
body under FRAND conditions.
In contrast, G3 and PRIME have been developed as generally
open standards by a small group of firms that expanded to a
formal industry alliances with increasing numbers of members.
Only in a later stage (in 2011) these standards were proposed to
SDOs for acceptance.
Again different, Meters&More (formerly Telegestore), LON/
OSGP, AMIS, Flexnet, Kamstrup RF and MeshNet3, have initially
been developed as proprietary standards by a single firm.
Meters&More, LON/OSGP and AMIS have later been published by
these firms. While Meters&More and LON/OSGP have both formed
alliances, only LON/OSGP has been proposed to SDOs for
acceptance.
So far, only four PLC standards, namely Meters&More, PRIME,
G3 and PLAN with DLMS/COSEM as well as all point-to-point
mobile communication standards have been accepted as being
conformant to the European M/441 criteria for open smart meter
standards.
Jointly, the standard adoption pattern and the analysis of
openness show that standards that have been developed
and deployed in the early years (2001–2007) were mostly pro-
prietary. They only opened up in recent years. More recently
developed standards were open from the start. While older,
proprietary standards have achieved a large installed base, they
have lower growth prospects for the future than newer more open
standards.
5.2.3. Ownership of communication networks
PLC and RF mesh networks give utilities the choice to own and
operate the communication network or to outsource it to a third
party.13 This choice results from the fact that such networks
require dedicated, field-based infrastructure such as data collec-
tors or data concentrators. Such LANs are mainly built and
operated as private networks, often by the DSO itself.
In contrast, the selection of point-to-point and point-to-
multipoint networks for smart meter communication implies the
outsourcing of network ownership and operation, because it uses
existing infrastructure. A dedicated network for smart metering
would not only be cost prohibitive from an infrastructure per-
spective, licenses for the required frequency band are also expen-
sive and typically already awarded to communication service
providers (such as Vodafone or Telefonica). As a consequence
utilities have to outsource ownership and operation of such
networks to cellular or Machine to Machine (M2M) service
providers. This also implies that utilities have to share the network
with other users [7]. This can be a concern, if continuous avail-
ability of communication is a priority (ibid).
5.2.4. Costs and mode of meter replacement
The costs for smart meter communication chips and modules
mostly depends on two factors: The price increases almost
proportionally with data rates (cf. technical comparison) and it
decreases with large overall production volumes of communica-
tion chips and modules thanks to economies of scale. Therefore a
13
In cases where retailers are responsible for the smart meter roll-out, the
usage of PLC technologies is subject to grid access granted by the DSO.
 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Table 8
Overview technical characteristics of wireless point-to-multipoint and RF mesh standards.
Standard Modulation Data rates
Kamstrup RF Undisclosed Approx. 4 kbps
MeshNet3 GFSK 9.6 kbps
802.15.4 MR-FSK; MR-OFDM; MR-O-QPSK MR-FSK: 10–400 kbps
MR-OFDM: 50–800 kbps
MR-O-QPSK: 12.5–500 kbps
Flexnet Undisclosed Up to 172 kbps
Fig. 3. Installed base in Europe per standard in 2012 and projection of expected
installed base in 2020.
large actual or forecasted installed base, contributes to decreasing
prices. Since some very large utilities decide about communication
modules for millions of meters, a single utility can influence prices
significantly14.
At least equally important than the price for communication
chips and modules, is the mode of smart meter replacement.
Technologies that require the establishment of dedicated LAN
communication infrastructures (PLC and RF mesh) are most cost
effective when smart meters are replaced in a comprehensive,
regional roll-out. The installation of data concentrators pays off
best, if all nodes in a geographical area are connected at once and
many meters are connected to one data concentrator.
Since the selection of mobile communication and point-to-
multipoint standards implies the use existing networks, they can
be a cost effective option if the introduction of smart meters is
done selectively, e.g., just for new buildings or for high consump-
tion households. In summary, the comparison of standards against
non-technical criteria shows further differences between the
standards.
5.3. Summary of comparison
Table 9 highlights the strengths and weaknesses of each
standard. It shows that some standards (e.g. G3 and PRIME)
achieve more favorable ratings than others. This does of course
not imply that these standards should be automatically selected.
Instead, it provides a first overview and framework to guide the
standard selection process. On this basis, every utility will then
have to specify its individual use cases and the resulting weighting
of the criteria.
The table shows that PLC standards have advantages, if meters
are mounted in cellars, if targeted use cases require low data rates
14
Several European utilities count more than ten million meters (up to 33
million) in their installed base. This volume is significant not only on a national
scale (where it represents up to 90% market share) but also on European level
where it represents up to 10% market share. In the interviews, large chip and meter
manufacturers confirmed that such large volumes substantially decrease unit costs.
1259
Frequency band IP? Transmission power
433 MHz unlicensed ISM band 10 mW at 433 MHz
444 MHz licensed band 500 mW at 444 MHz
869.400–869.650 MHz IPv4 50 mW at 869 MHz
863–870 MHz (for Europe) IPv6 25 mW at 863–870 MHz
412/422 MHz UK licensed band IPv4 2 W at 412 and 422 MHz
and if meters are replaced in a comprehensive roll-out. While first
generation PLC standards benefit from the experience of a large
installed base, they are technically inferior (data rates and robust-
ness). More recently developed PLC standards in contrast are
technically more advanced but not (yet) field-proven.
RF mesh standards (Meshnet and Kamstrup RF) have similar
strengths as PLC standards. In densely populated areas, they
are more robust thanks to their mesh architecture. However,
they have challenges reaching meters in cellars and the
standards deployed so far are proprietary. In Europe, RF mesh is
likely to remain the exception, unless dedicated or more favorable
frequency bands or power levels are made available in the
future.
Point-to-point and point-to-multipoint communication stan-
dards such as GSM and Flexnet generally show advantages
if meters are distributed over larger areas and if meters
are exchanged selectively rather than in a comprehensive
roll-out. In addition, point-to-point standards allow for
the realization of advanced use cases with very high data rate
requirements (at the expense of higher costs). But these stan-
dards do not leave the choice to own and operate the commu-
nication network.
Overall, the table shows that, the comparison against both
technical and non-technical criteria reveals even more trade-
offs. While, for example, early deployments are based on
proprietary standards, we observe a trend towards open stan-
dards. We also see that higher data rates typically come at the
expense of higher costs and in the case of mobile communica-
tion networks they automatically imply the outsourcing of
ownership and operation of the communication network to a
third party.
6. Discussion and conclusion
In this paper we made three contributions. Firstly, we
identified 17 standards for smart meter LAN communication in
Europe. The large number of alternative standards reflects the
nascent status of this technological field. Secondly, we provided
a comprehensive review of criteria relevant for standard selec-
tion, which also remain relevant in the future, when existing
standards change or new standards emerge. We highlighted that
the relevance of these criteria for each utility depends on the
targeted use cases and on the preferences regarding the own-
ership of communication networks. Finally, we compared the
standards against these criteria. The comparison has revealed
major differences between standards. None of the standards is
superior in all criteria, which means that standard selection
involves trade-offs.
For an actual selection of a smart meter standard, utilities will
have to weigh the relevance of the different criteria for their
specific situation. They need to analyze environmental factors such
as regulation or grid topology and they have to determine what
kind of functionalities ‘their’ smart meters will have to fulfill now
1260 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
Table 9
Overview standard comparison against technical and non-technical criteria.
and in the future. This is all the more difficult as future
requirements are still unclear: for example, does it make sense
to invest in higher data rates now, even though it is unclear
whether they might be needed later? What are the advantages
and disadvantages of owning and operating the communication
network? Or what are the benefits of cooperating with third
parties such as mobile phone operators? A major challenge for
utilities is that the electricity sector is changing fundamentally
and that they have to develop new competences (e.g. in ICT)
and launch new services and business models. Competition will
intensify, new players enter the field [34], innovation cycles
may become much shorter than in former times and sectors
may increasingly overlap (energy and ICT, energy and trans-
portation). Smart grid technology is certainly a central element
in this broader transformation and provides an opportunity to
strategically experiment with new options and to learn about
how to cope with higher degrees of uncertainty and more
dynamic market environments.
Beyond strategic relevance for utilities, this study provides
insights into the development of smart metering and smart
grids more generally. The pre-dominant use of narrowband PLC
standards in Europe currently limits the use cases for smart
metering and the possibilities for leveraging this communica-
tion infrastructure for broader smart grid solutions. The
preference of many utilities for low costs and ownership
of the communication network may therefore create future
bottlenecks. Moreover, the high number of competing smart
metering standards leaves economies of scale unrealized, cre-
ates a patchwork of different solutions and potentially results in
a lack of interoperability. Complementary products or services
(e.g. one bill for charging of electric vehicles in different
electricity networks) might therefore be difficult to realize. This
is a call for policy makers to set incentives for standard
convergence and to launch initiatives that facilitate coordina-
tion among key actors, including negotiation and potential
resolution of conflicting interests. The European Union mandate
(M/441) to three standard development organizations (CEN,
CENELEC and ETSI) to work towards harmonization of smart
meter standards has certainly been a worthwhile endeavor in
this regard.
However, we also have to acknowledge that high levels of
uncertainty, yet emerging applications, conflicting strategic inter-
ests, lack of guidance and struggles over standards are typical for
emerging technological fields [44,45] and broader socio-technical
transitions [46,47]. In the early years of mobile telephony several,
for example, mostly national standards competed for almost two
decades and subsequent technology generations brought about
even new standards as well as changing alliances of promoters
[36,48]. Similar standard battles and frequent standard changes as
technology progressed have also been observed in information
technology [8,9,49]. Shorter technology cycles and more frequent
innovation is something the electricity sector still has to get used
to with its traditionally long lifetimes of technological infrastruc-
ture (30–80 yrs). The ongoing integration of ICT in the energy
sector, of which smart meters are but one example, is very likely to
change many of the long established professional practices in the
sector [34].
It is also important to note that the electricity sector currently
is in a very fundamental phase of transformation that very much
exceeds smart grid technology and smart meter standards. This
‘energy transition’ also includes the development and diffusion of
innovations in renewable energy, energy efficiency and energy
storage and a substitution of fossil and nuclear fuels – at least in
some countries [50,51]. Such broader transitions do not just
encompass a range of different technologies but also far-reaching
changes in the institutional and organizational structures of a
sector [15,52]. The development, assessment and selection of
technology standards are some of the many aspects actors have
to deal with in such times of great changes.
Of course, our analysis also has limitations. Firstly, not all
criteria were equally easy to operationalize and to assess. While
some criteria (e.g. data rates) are well-studied and ‘straight
forward’ to measure, others were less clear. Especially for the
criteria robustness, openness and cost efficiency, we could rely on
very few previous studies. In these cases, expert interviews helped
us further operationalize and assess the criterion (e.g. for robust-
ness we distinguished robust architectures and modulation pro-
tocols). In these cases triangulation of different experts was of
crucial importance to achieve validity. Due to a lack of reliable
studies and possibility to operationalize based on expert state-
ments, we even had to exclude one relevant criterion, namely ‘data
security’.
Secondly, some criteria are somewhat interdependent or show
a certain overlap, e.g., costs vs. installed base. To avoid over-
emphasizing these criteria we carefully operationalized each
criteria. The criterion cost, for example, is limited to the cost
efficiency for certain roll-out types to avoid double counting cost
effects of current and expected installed base.
Thirdly, we used a relative ranking of each standard with an
equal weight for each criterion. In practice, however, utilities will
weigh the criteria according to their use cases and strategic
priorities. While beyond the scope of this study, a multi-criteria
decision making approach could provide an absolute ranking of
standards for a given preference of a utility.
Finally, our study aimed at providing a very comprehensive
comparison of standards against technical and non-technical
criteria. This breadth automatically comes at the expense of depths
in the analysis of each criterion. Future studies should therefore
build on this work and provide in-depth analysis especially for
criteria that are less well studied so far (e.g. cost).
 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262 1261

Box 1–European frequency band regulations for powerline communication (based on [53]).

Box 2–European band regulation for radio frequency (based on [54]).

Appendix A EDP Energias de Portugal

ERDF Électricité Réseau Distribution France
List of abbreviations ETSI European Telecommunications Standards Institute
Abbreviation Full meaning FCC Federal Communications Commission (USA)
AMIS Automated Metering and Information System FRAND Fair, Reasonable And Non-Discriminatory
ARIB Association of Radio Industries and Businesses (Japan) GSM Global System for Mobile Communications
BPSK Binary Phase Shift Key HAN Home Area Network
CEN European Committee for Standardization IEC International Electrotechnical Commission
CENELEC European Committee for Electrotechnical IETF Internet Engineering Task Force
Standardization ISO International Organization for Standardization
COSEM COmpanion Specification for Energy Metering LAN Local Area Network
DCSK Differential Code Shift Keying LTE Long Term Evolution
DLMS/COSEM Device Language Message Specification MAC Media Access Control
DSO Distribution System Operator MHz Mega Herz
1262 S. Erlinghagen et al. / Renewable and Sustainable Energy Reviews 43 (2015) 1249–1262
MR-FSK Multi-Rate Frequency Shift Keying
MR-O-QPSK Multi-rate Offset Quadrature Phase Shift Keying
MR-OFDM Multi-Rate Orthogonal Frequency-Division Multi-
Plexing
OFDM Orthogonal Drequency-Division Multi-Plexing
OSGP Open Smart Grid Protocol
OSI Open Systems Interconnection
P-LAN PowerLAN (Local Area Network)
PLC Power Line Communication
PRIME PoweRline Intelligent Metering Evolution
RF Radio Frequency
S-FSK Spatial Frequency Shift Keying
SDO Standard Development Organization
UMTS Universal Mobile Communication System
WAN Wide Area Network
References
[1] Farhangi H. The path of the Smart Grid. IEEE Power Energy Mag 2010;8:18–28.
[2] Faruqui A, Harris D, Hledik R. Unlocking the [euro] 53 billion savings from
smart meters in the EU: how increasing the adoption of dynamic tariffs could
make or break the EU's smart grid investment. Energy Policy 2010;38:
6222–31.
[3] Fox-Penner P. Smart power: climate change, the smart grid, and the future of
electric utilities. Washington, Covelo, London: Island Press; 2010.
[4] Rohjans S, Uslar M, Bleiker R, Gonzalez J, Specht M, Suding T, et al. Survey of
smart grid standardization studies and recommendations. In: IEEE interna-
tional conference on smart grid communications 2010, IEEE; 2010. p. 583–588.
[5] Ancillotti E, Bruno R, Conti M. The role of communication systems in smart
grids: architectures, technical solutions and research challenges. Comput
Commun 2013;36:1665–97.
[6] OECD/IEA. Smart Grid Technology Roadmap; 2011.
[7] Güngör VC, Sahin D, Kocak T, Ergüt S, Buccella C, Member S, et al. Smart grid
technologies: communication technologies and standards. IEEE Trans Ind
Inform 2011;7:529–39.
[8] Jain S. Pragmatic agency in technology standards setting: the case of ethernet.
Res Policy 2012;41:1643–54.
[9] Garud R, Jain S, Kumaraswamy A. Institutional entrepreneurship in the
sponsorship of common technological standards: the case of Sun Microsys-
tems and Java. Acad Manag J 2002;45:196–214.
[10] Erlinghagen S, Markard J. Technology users as game changers: standardization
strategies in the field of smart meter technology. In: proceedings of the 5th
international conference on sustainability transitions (IST), UTR University;
2014.
[11] West. J. How open is open enough? Res Policy 2003;32:1259–85.
[12] Funk JL, Methe DT. Market- and committee-based mechanisms in the creation
and diffusion of global industry standards: the case of mobile communication.
Res Policy 2001;30:589–610.
[13] Cusumano MA, Mylonadis Y, Rosenbloom RS. Strategic Maneuvering and
Mass-Market Dynamics: The Triumph of VHS over Beta. Bus Hist Rev
1992;66:51–94.
[14] van den Ende J, van de Kaa G, den Uijl S, deVries HJ. The Paradox of Standard
Flexibility: The Effects of Co-evolution between Standard and Interorganiza-
tional Network. Organ Stud 2012;33:705–36.
[15] Markard J, Raven R, Truffer B. Sustainability transitions: an emerging field of
research and its prospects. Res Policy 2012;41:955–67.
[16] Atayero AA, Alatishe AA, Ivanov YA. Power line communication technologies:
modeling and simulation of PRIME physical layer. In: proceedings of the world
congress on engineering and computer science 2012, vol. II, San Francisco,
USA; 2012.
[17] Buayairaksa S, Thepphaeng S, Pirak C. On the performance of G3 power line
communication network with smart energy meter. In: proceedings of the
international conference on electrical engineering/electronics,computer, tele-
communications and information technology, IEEE; 2013. p. 1–5.
[18] Berganza I, Sendin A, Arzuaga A, Sharma M, Varadarajan B. PRIME on-field
deployment – first summary of results and discussion. Getting practical: field
trials, deployments, and lessons learnt (IEEE SmartGridComm), IEEE; 2011, p.
297–302.
[19] Tan RHG, Lee CH, Mok VH. Automatic power meter reading system using GSM
network. In: proceedings of the international power engineering conference,
IPEC 2007. 07. p. 465–9.
[20] Fan Z, Haines RJ, Kulkarni P. M2M Communications for E-Health and Smart
Grid: An Industry and Standard Perspective; 2013.
[21] Khalifa T, Naik K, Nayak A. A survey of communication protocols for automatic
meter reading applications. IEEE Commun Surv Tutor 2011;13:168–82.
[22] Rohjans S, Uslar M, Juergen Appelrath JH. OPC UA and CIM: semantics for the
smart grid. In: transmission and distribution conference and exposition, 2010
IEEE PES, IEEE; 2010, p. 1–8.
[23] Feuerhahn S, Zillgith M, Wittwer C, Wietfeld C. Comparison of the commu-
nication protocols DLMS/COSEM, SML and IEC 61850 for smart metering
applications. In: proceedings of the IEEE international conference on smart
grid communications, IEEE; 2011. p. 410–5.
[24] Depuru SSSR, Wang L, Devabhaktuni V. Smart meters for power grid:
challenges, issues, advantages and status. Renew Sustain Energy Rev
2011;15:2736–42.
[25] Ma R, Chen H, Huang Y, Meng W, Member S. Smart grid communication: its
challenges and opportunities. IEEE Trans Smart Grid 2013;4:36–46.
[26] Usman A, Shami SH. Evolution of communication technologies for smart grid
applications. Renew Sustain Energy Rev 2013;19:191–9.
[27] Depuru SSSR, Wang L, Devabhaktuni V. Electricity theft: overview, issues,
prevention and a smart meter based approach to control theft. Energy Policy
2011;39:1007–15.
[28] Hargreaves T, Nye M, Burgess J. Making energy visible: a qualitative field study
of how householders interact with feedback from smart energy monitors.
Energy Policy 2010;38:6111–9.
[29] Darby S. The effectiveness of feedback on energy consumption. Oxford, UK:
University of Oxford; 2006.
[30] Fan Z, Kulkarni P, Gormus S, Efthymiou C, Kalogridis G, Sooriyabandara M,
et al. Smart grid communications: overview of research activities. IEEE
Commun Surv Tutor 2013;15:21–38.
[31] Berezak-Lazarus N. Smart grid enabled and enhanced by broadband powerline
strategic value and inherent technical characteristics of broadband powerline.
In: proceedings of the third international conference on smart grids, commu-
nications and it energy-aware technologies, ENERGY 2013; 2013. p. 77–82.
[32] Brunsson N, Rasche A, Seidl D. The dynamics of standardization: three perspec-
tives on standards in organization studies. Organ Stud 2012;33:613–32.
[33] Giordano, V, Meletiou, A, Covrig, CF, Mengolini, A, Ardelean, M, Fulli, G, et al.
Smart Grid projects in Europe: lessons learned and current developments
2012 update; 2013.
[34] Erlinghagen S, Markard J. Smart grids and the transformation of the electricity
sector: ICT firms as potential catalysts for sectoral change. Energy Policy
2012;51:895–906.
[35] Van de Kaa G, van den Ende J, de Vries HJ, van Heck E. Factors for winning
interface format battles: a review and synthesis of the literature. Technol
Forecast Soc Change 2011;78:1397–411.
[36] Funk JL, Methe DT. Market- and committee-based mechanisms in the creation
and diffusion of global industry standards: the case of mobile communication.
Res Policy 2001;30:589–610.
[37] West J. The economic realities of open standards: black, white, and many
shades of gray. In: Stango V, editor. Stand. Public Policy, Cambride. MA:
Cambridge University Press; 2007. p. 87–122.
[38] Blind K, Thumm N. Interrelation between patenting and standardisation
strategies: empirical evidence and policy implications. Res Policy
2004;33:1583–98.
[39] Leiponen AE. Competing through cooperation: the organization of standard
setting in wireless telecommunications. Manag Sci 2008;54:1904–19.
[40] Sum C, Kojima F, Harada H. Coexistence of Homogeneous and Heterogeneous
Systems for IEEE 802.15.4g Smart Utility Networks. In: proceedings of the 2011
IEEE international symposium on dynamic spectrum access networks coex-
istence; 2011. p. 510–20.
[41] Sum C-S, Harada H, Kojima F, Lan Z, Funada R. Smart utility networks in TV
white space. IEEE Commun Mag 2011;49:132–9.
[42] B. Lichtensteiger, B. Bjelajac, C. Müller, C.R.F. Wietfeld. Mesh systems for smart
metering: system architecture and performance. In: proceedings of the first IEEE
international conference on smart grid communications; 2010. p. 379–84.
[43] Costa-Perez X, Festag A, Kolbe H, Quittek J, Schmid S, Stiemerling M, et al.
Latest trends in telecommunication standards. ACM SIGCOMM Comput Com-
mun Rev 2013;43:64–71.
[44] Bergek A, Jacobsson S, Sandén B. Legitimation and development of positive
externalities: two key processes in the formation phase of technological
innovation systems. Technol Anal Strateg Manag 2008;20:575–92.
[45] Suarez FF. Battles for technological dominance: an integrative framework. Res
Policy 2004;33:271–86.
[46] Farla J, Markard J, Raven R, Coenen L. Sustainability transitions in the making:
a closer look at actors, strategies and resources. Technol Forecast Soc Change
2012;79:991–8.
[47] Smink MM, Hekkert MP, Negro SO. Keeping sustainable innovation on a leash?
Exploring incumbents' institutional strategies Bus Strateg Environ 2013.
[48] Lyytinen K, Fomin VV. Achieving high momentum in the evolution of wireless
infrastructures: the battle over the 1G solutions. Telecomm Policy
2002;26:149–70.
[49] Van den Ende J, van de Kaa G, den Uijl S, de Vries HJ. The paradox of standard
flexibility: the effects of co-evolution between standard and interorganiza-
tional network. Organ Stud 2012;33:705–36.
[50] Strunz S. The German energy transition as a regime shift. Ecol Econ
2014;100:150–8.
[51] Solomon BD, Krishna K. The coming sustainable energy transition: history,
strategies, and outlook. Energy Policy 2011;39:7422–31.
[52] Geels FW. The multi-level perspective on sustainability transitions: responses
to seven criticisms. Environ Innov Soc Transit 2011;1:24–40.
[53] IEEE Standards Association. IEEE Standard for Low-Frequency (less than
500 kHz) Narrowband Power Line Communications for Smart Grid Applica-
tions, New York; 2013.
[54] ETSI. ETSI TR 102 649-2 v.1.2.1; 2010.