Content uploaded by Dejun Yang
Author content
All content in this area was uploaded by Dejun Yang on Aug 22, 2014
Content may be subject to copyright.
Smart Grid – The New and Improved Power Grid: A Survey
Xi Fang, Student Member, IEEE, Satyajayant Misra, Member, IEEE, Guoliang Xue, Fellow, IEEE,
and Dejun Yang, Student Member, IEEE
Abstract—The Smart Grid, regarded as the next generation
power grid, uses two-way flows of electricity and information to
create a widely distributed automated energy delivery network.
In this article, we survey the literature till 2011 on the enabling
technologies for the Smart Grid. We explore three major systems,
namely the smart infrastructure system, the smart management
system, and the smart protection system. We also propose possible
future directions in each system. Specifically, for the smart
infrastructure system, we explore the smart energy subsystem,
the smart information subsystem, and the smart communication
subsystem. For the smart management system, we explore various
management objectives, such as improving energy efficiency,
profiling demand, maximizing utility, reducing cost, and con-
trolling emission. We also explore various management methods
to achieve these objectives. For the smart protection system, we
explore various failure protection mechanisms which improve
the reliability of the Smart Grid, and explore the security and
privacy issues in the Smart Grid.
Index Terms—Smart grid, power grid, survey, energy, informa-
tion, communications, management, protection, security, privacy.
I. INTRODUCTION
Traditionally, the term grid is used for an electricity system
that may support all or some of the following four operations:
electricity generation, electricity transmission, electricity dis-
tribution, and electricity control.
Asmart grid (SG), also called smart electrical/power grid,
intelligent grid, intelligrid, futuregrid, intergrid, or intragrid, is
an enhancement of the 20th century power grid. The traditional
power grids are generally used to carry power from a few
central generators to a large number of users or customers.
In contrast, the SG uses two-way flows of electricity and
information to create an automated and distributed advanced
energy delivery network. Table Igives a brief comparison
between the existing grid and the SG.
By utilizing modern information technologies, the SG is
capable of delivering power in more efficient ways and
responding to wide ranging conditions and events. Broadly
stated, the SG could respond to events that occur anywhere in
the grid, such as power generation, transmission, distribution,
and consumption, and adopt the corresponding strategies. For
instance, once a medium voltage transformer failure event
occurs in the distribution grid, the SG may automatically
change the power flow and recover the power delivery service.
Manuscript received May 27, 2011; revised September 25, 2011; accepted
September 30, 2011.
Xi Fang, Guoliang Xue, and Dejun Yang are affiliated with Arizona State
University, Tempe, AZ 85281. E-mail: {xi.fang, xue, dejun.yang}@asu.edu.
Satyajayant Misra is affiliated with New Mexico State University, Las Cruces,
NM 88003. Email: misra@cs.nmsu.edu. This research was supported in part
by ARO grant W911NF-09-1-0467 and NSF grant 0905603. The information
reported here does not reflect the position or the policy of the federal
government.
TABLE I: A Brief Comparison between the Existing Grid and
the Smart Grid [70]
Existing Grid Smart Grid
Electromechanical Digital
One-way communication Two-way communication
Centralized generation Distributed generation
Few sensors Sensors throughout
Manual monitoring Self-monitoring
Manual restoration Self-healing
Failures and blackouts Adaptive and islanding
Limited control Pervasive control
Few customer choices Many customer choices
Let us consider another example of demand profile shaping.
Since lowering peak demand and smoothing demand profile
reduces overall plant and capital cost requirements, in the
peak period the electric utility can use real-time pricing to
convince some users to reduce their power demands, so that
the total demand profile full of peaks can be shaped to a nicely
smoothed demand profile.
More specifically, the SG can be regarded as an electric
system that uses information, two-way, cyber-secure commu-
nication technologies, and computational intelligence in an
integrated fashion across electricity generation, transmission,
substations, distribution and consumption to achieve a system
that is clean, safe, secure, reliable, resilient, efficient, and
sustainable. This description covers the entire spectrum of
the energy system from the generation to the end points of
consumption of the electricity [80]. The ultimate SG is a
vision. It is a loose integration of complementary components,
subsystems, functions, and services under the pervasive control
of highly intelligent management-and-control systems. Given
the vast landscape of the SG research, different researchers
may express different visions for the SG due to different
focuses and perspectives. In keeping with this format, in this
survey,we explore three major systems in SG from a technical
perspective:
•Smart infrastructure system: The smart infrastructure
system is the energy, information, and communication
infrastructure underlying of the SG that supports 1) ad-
vanced electricity generation, delivery, and consumption;
2) advanced information metering, monitoring, and man-
agement; and 3) advanced communication technologies.
•Smart management system: The smart management sys-
tem is the subsystem in SG that provides advanced
management and control services.
•Smart protection system: The smart protection system
is the subsystem in SG that provides advanced grid
reliability analysis, failure protection, and security and
privacy protection services.
Other surveys on SG were done in [3,17,29,41,42,90,
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 2
97,211,247,251,254,267]. Chen et al. [41], Yu et al.
[267], and Hassan and Radman [97] briefly reviewed the
basic concepts of SG and some technologies that could be
used in SG. The authors of [211,247] reviewed the existing
SG standardizations and gave concrete recommendations for
future SG standards. Vasconcelos [251] outlined the potential
benefits of smart meters, and provided a short overview of
the legal framework governing metering activities and policies
in Europe. Brown and Suryanarayanan [29] determined an
industry perspective for the smart distribution system and
identified those technologies which could be applied in the
future research in the smart distribution system. Baumeister
[17] presented a review of the work related to SG cyber
security. Chen [42] explored the security and privacy issues in
SG and related these issues to cyber security in the Internet.
Gungor and Lambert [90] explored communication networks
for electric system automation and attempted to provide a
better understanding of the hybrid network architecture that
can provide heterogeneous electric system automation appli-
cation requirements. Akyol et al. [3] analyzed how, where,
and what types of wireless communications are suitable for
deployment in the electric power system. Wang et al. [254]
provided a survey on the communication architectures in the
power systems, including the communication network com-
positions, technologies, functions, requirements, and research
challenges. They also discussed the network implementation
considerations and challenges in the power system settings.
Our survey complements these existing surveys in that
we: 1) comprehensively survey the literature till 2011, and
systematically classify the work for the smart infrastructure
system (energy, information, and communications), the smart
management system, and the smart protection system; and 2)
outline challenges and future research directions for each of
these three major systems. The novelty of this survey is in the
classification, volume of information provided, and outlining
of future research in these three major systems.
This survey is structured as follows. In Section II, we
present an overview of SG. In Section III, we review the
legislations, the standards, the projects, the programs, and
the trials of SG. We then describe three subsystems of the
smart infrastructure system in Sections IV-VI, respectively.
We next describe the smart management system and the smart
protection system in Sections VII and VIII, respectively. In
Section IX, we conclude this survey and present some lessons
learned. In addition, refer to Appendix A for the abbreviations
used in this survey.
II. WHAT IS SMART GRID?
The initial concept of SG started with the idea of advanced
metering infrastructure (AMI) with the aim of improving
demand-side management and energy efficiency, and con-
structing self-healing reliable grid protection against malicious
sabotage and natural disasters [204]. However, new require-
ments and demands drove the electricity industries, research
organizations, and governments to rethink and expand the
initially perceived scope of SG. The U.S. Energy Independence
and Security Act of 2007 directed the National Institute of
Standards and Technology (NIST) to coordinate the research
and development of a framework to achieve interoperability
of SG systems and devices. Although a precise and compre-
hensive definition of SG has not been proposed yet, according
to the report from NIST [177], the anticipated benefits and
requirements of SG are the following:
1) Improving power reliability and quality;
2) Optimizing facility utilization and averting construction
of back-up (peak load) power plants;
3) Enhancing capacity and efficiency of existing electric
power networks;
4) Improving resilience to disruption;
5) Enabling predictive maintenance and self-healing re-
sponses to system disturbances;
6) Facilitating expanded deployment of renewable energy
sources;
7) Accommodating distributed power sources;
8) Automating maintenance and operation;
9) Reducing greenhouse gas emissions by enabling electric
vehicles and new power sources;
10) Reducing oil consumption by reducing the need for
inefficient generation during peak usage periods;
11) Presenting opportunities to improve grid security;
12) Enabling transition to plug-in electric vehicles and new
energy storage options;
13) Increasing consumer choice;
14) Enabling new products, services, and markets.
In order to realize this new grid paradigm, NIST provided a
conceptual model (as shown in Fig. 1), which can be used as a
reference for the various parts of the electric system where SG
standardization work is taking place. This conceptual model
divides the SG into seven domains. Each domain encom-
passes one or more SG actors, including devices, systems,
or programs that make decisions and exchange information
necessary for performing applications. The brief descriptions
of the domains and actors are given in Table II. Refer to
the appendix of the NIST report [177] for more detailed
descriptions. Note that NIST proposed this model from the
perspectives of the different roles involved in the SG.
Fig. 1: The NIST Conceptual Model for SG [177]
In contrast, our survey, which looks at SG from a technical
view point, divides SG into three major systems: smart infras-
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 3
TABLE II: Domains and Actors in the NIST SG Conceptual
Model [177]
Domain Actors in the Domain
Customers The end users of electricity. May also generate, store,
and manage the use of energy.
Markets The operators and participants in electricity markets.
Service Providers The organizations providing services to electrical
customers and utilities.
Operations The managers of the movement of electricity.
Bulk Generation The generators of electricity in bulk quantities. May
also store energy for later distribution.
Transmission The carriers of bulk electricity over long distances.
May also store and generate electricity.
Distribution The distributors of electricity to and from customers.
May also store and generate electricity.
tructure, smart management and smart protection systems.
1) Smart infrastructure system: The smart infrastructure
system is the energy, information, and communication
infrastructure underlying the SG. It supports two-way
flow of electricity and information. Note that it is
straightforward to understand the concept of “two-way
flow of information.” “Two-way flow of electricity”
implies that the electric energy delivery is not unidirec-
tional anymore. For example, in the traditional power
grid, the electricity is generated by the generation plant,
then moved by the transmission grid, the distribution
grid, and finally delivered to users. In an SG, electricity
can also be put back into the grid by users. For example,
users may be able to generate electricity using solar
panels at homes and put it back into the grid, or electric
vehicles may provide power to help balance loads by
“peak shaving” (sending power back to the grid when
demand is high). This backward flow is important. For
example, it can be extremely helpful in a microgrid
(described in Section IV-D) that has been ‘islanded’ due
to power failures. The microgrid can function, albeit at
a reduced level, with the help of the energy fed back
by the customers. In this survey, we further divide this
smart infrastructure into three subsystems: the smart
energy subsystem, the smart information subsystem, and
the smart communication subsystem.
•The smart energy subsystem is responsible for
advanced electricity generation, delivery, and con-
sumption.
•The smart information subsystem is responsible
for advanced information metering, monitoring, and
management in the context of the SG.
•The smart communication subsystem is responsible
for communication connectivity and information
transmission among systems, devices, and applica-
tions in the context of the SG.
Note that the reason why we separate information sub-
system and communication subsystem is to get a handle
on the involvedcomplexity of the SG as a system of sys-
tems. This also makes our survey compliant with IEEE
P2030 [109] to meet the interoperability requirements.
We will briefly describe IEEE P2030 in Section III.
2) Smart management system: The smart management sys-
tem is the subsystem in SG that provides advanced
management and control services and functionalities.
The key reason why SG can revolutionize the grid is
the explosion of functionality based on its smart infras-
tructure. With the development of new management ap-
plications and services that can leverage the technology
and capability upgrades enabled by this advanced in-
frastructure, the grid will keep becoming “smarter.” The
smart management system takes advantage of the smart
infrastructure to pursue various advanced management
objectives. Thus far, most of such objectives are related
to energy efficiency improvement, supply and demand
balance, emission control, operation cost reduction, and
utility maximization.
3) Smart protection system: The smart protection system
is the subsystem in SG that provides advanced grid
reliability analysis, failure protection, and security and
privacy protection services. By taking advantage of the
smart infrastructure, the SG must not only realize a
smarter management system, but also provide a smarter
protection system which can more effectively and effi-
ciently support failure protection mechanisms, address
cyber security issues, and preserve privacy.
Fig. 2shows the detailed classification of these three
major systems. In this paper, we will describe SG using this
classification. We encourage the readers to refer back to this
classification in case of any confusion while reading the text.
III. ANOVERVIEW OF LEGISLATIONS, STANDARDS,
PROJECTS, PROGRAMS AND TRIALS
In 2001, the U.S. Department of Energy (DOE) began a
series of Communications and Controls Workshops focused
on the integration of distributed energy resources [146]. The
broad view of a transformation to SG was reflected in DOE’s
GridWise [51,200]. The U.S. federal government has also
established its policy for SG, which is reflected in two Acts
of Congress. The first one is the Energy Independence and
Security Act of 2007 [239] which specifies studies on the state
and security of SG; establishes a federal advisory committee
and intergovernment agency task force; frames technology
research, development and demonstration; directs the advance-
ment of interoperability; and creates a matching fund program
to encourage investment in SG [146]. The second one is the
American Recovery and Reinvestment Act of 2009 [240],
which includes $3.4 billion in funding for the SG Investment
Grant Program and $615 million for the SG Demonstration
Program. The result of these programs is expected to lead to
a combined investment of over $8 billion in SG capabilities.
Several standardizations have also come up in different
areas, countries, or organizations. We list several major SG
standardization roadmaps and studies in the following:
1) United States: NIST IOP Roadmap [177];
2) European Union: Mandate CEN/CENELEC M/441 [67];
3) Germany: BMWi E-Energy Program [187], BDI initia-
tive -Internet der Energie [111];
4) China: SGCC Framework [233];
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 4
S
Sm
ma
ar
rt
t
G
Gr
ri
id
d
Fig. 2: The Detailed Classification of the Smart Infrastructure System, the Smart Management System, and the Smart Protection
System: In Sections IV-VI, we will describe the smart energy subsystem, the smart information subsystem, and the smart
communication subsystem, respectively. In Section VII, we will describe the smart management system. In Section VIII, we
will describe the smart protection system.
5) Japan: METI Smart Grid roadmap [118];
6) Korea: Smart Grid Roadmap 2030 [247];
7) IEEE: P2030 [109];
8) IEC SMB: SG 3 Roadmap [227];
9) CIGRE: D2.24 [113];
10) Microsoft: SERA [167].
A detailed study comparing them and an overview of other
SG roadmaps (e.g. Austria [209], UK [65], and Spain [75]) can
be found in [211,247]. In order to drive all the dimensions
of the future standards of SG, a cooperative standardization
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 5
Fig. 3: Smart Grid Project Map [226]
roadmap crossing different areas, countries, and organizations
is desired. In the meantime, those existing standards may need
to be developed and revised to adapt to the changes within
technical, political, and regulatory aspects. Considering the
importance of IEEE standards, we briefly describe IEEE P2030
[109]. IEEE P2030 focuses on a system level approach to the
guidance for interoperability components of communications,
power systems, and information technology platforms. SG
interoperability provides organizations the ability to communi-
cate effectively and transfer meaningful data, even though they
may be using a variety of different information systems over
widely different infrastructures, sometimes across different
geographic regions and cultures. P2030 views the SG as a
large, complex “system of systems” and provides guidance
to navigate the numerous SG design pathways throughout the
electric power system and end-use applications.
In order to promote the development of SG, governments,
academia, industry, and research organizations have put a great
deal of effort in pilot projects, programs, and field trials. In
order to help the readers assess the recent progress, especially
in the industrial sector, we summarize 17 major projects,
programs, and trials, shown in Table IV of Appendix B. They
cover smart meter, AMI, transmission grid, distribution grid,
distributed resource, virtual power plant, home application,
microgrid, electric vehicle, and integrated systems. These con-
cepts will be described in the following sections. In addition,
in order to give the readers a direct impression, we use Fig. 3
to show the map of the SG projects collected by Smart Grid
Information Clearinghouse [226]. This map roughly shows
the locations and the objectives of these SG projects. We
can observe that in the U.S., Europe, and East Asia, there
already exist several integrated system projects, although we
are just at the beginning of the SG transition. As pointed out
by Giordano et al. [84], in almost all countries, a significant
amount of investments is devoted to projects which address
the integration of different SG technologies and applications.
Most of the technologies are known, but their integration is
the new challenge.
IV. SMART INFRASTRUCTURE SYSTEM I-SMART ENERGY
SUBSYSTEM
Two-way flows of electricity and information lay the in-
frastructure foundation for the SG. The smart infrastructure
can be subdivided into the smart energy subsystem, the smart
information subsystem, and the smart communication subsys-
tem, respectively. In this section, we explore existing work on
the smart energy subsystem and outline some future research
directions and challenges.
The traditional power grid is unidirectional in nature [70].
Electricity is often generated at a few central power plants by
electromechanical generators, primarily driven by the force
of flowing water or heat engines fueled by chemical com-
bustion or nuclear power. In order to take advantage of the
economies of scale, the generating plants are usually quite
large and located away from heavily populated areas. The
generated electric power is stepped up to a higher voltage
for transmission on the transmission grid. The transmission
grid moves the power over long distances to substations. Upon
arrival at a substation, the power will be stepped down from
the transmission level voltage to a distribution level voltage.
As the power exits the substation, it enters the distribution
grid. Finally, upon arrival at the service location, the power
is stepped down again from the distribution voltage to the
required service voltage(s). Fig. 4shows an example of the
traditional power grid.
In contrast with the traditional power grid, the electric
energy generation and the flow pattern in an SG are more
flexible. For example, the distribution grid may also be capable
of generating electricity by using solar panels or wind turbines.
In this survey, we still divide the energy subsystem into
power generation,transmission grid, and distribution grid.
Fig. 5shows a classification of the work on the smart energy
subsystem.
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 6
Fig. 4: An Example of the Traditional Power Grid
Smart Energy Subsystem
Power Generation
[6,36,114,117]
[155,172,265]
[266,269]
Transmission Grid
[24,141]Distribution Grid
[234,235]
Fig. 5: Classification of the Work on the Smart Energy
Subsystem
A. Power Generation
Electricity generation is the process of generating electricity
from other forms of energy, such as natural gas, coal, nuclear
power, the sun, and wind. During the 1820s and early 1830s,
British scientist Michael Faraday discovered the fundamental
principles of electricity generation: electricity can be generated
by the motion of a loop of wire or a disc of copper between
the poles of a magnet, a principle still being used today. There
are many energy sources used to generate electric power. Fig.
6shows the U.S. electricity generation by source in 2008 and
2009 [58]. As fossil fuels get depleted and generally get more
expensive, it is expected that the renewable energy will play
a more important role in the future power generation. Fig. 7
shows the predicted world energy supply by source [259].
In contrast to the power generation in the traditional power
grid, smarter power generation becomes possible as the two-
way flows of electricity and information are supported. A key
power generation paradigm enabled by SG will be the dis-
tributed generation (DG). DG takes advantage of distributed
energy resource (DER) systems (e.g. solar panels and small
wind turbines), which are often small-scale power generators
(typically in the range of 3 kW to 10,000 kW), in order
to improve the power quality and reliability. For example, a
microgrid (discussed in Section IV-D), which is a localized
grouping of electricity generators and loads, can disconnect
from the macrogrid so that distributed generators continue to
Fig. 6: U.S. Electricity Generation by Source [58]
Fig. 7: World Energy Supply by Source [259]
power the users in this microgrid without obtaining power
from outside. Thus, the disturbance in the macrogrid can be
isolated and the electric power supply quality is improved. A
study [114] from the International Energy Agency pointed out
that a power system based on a large number of reliable small
DGs can operate with the same reliability and a lower capacity
margin than a system of equally reliable large generators.
A review of different distributed energy technologies such
as microturbines, photovoltaic, fuel cells, and wind power
turbines can be found in [269].
However, implementing DG(s) in practice is not an easy
proposition due to several reasons. First, DG involves large-
scale deployments for generation from renewable resources,
such as solar and wind, whose yield is, however, subject to
wide fluctuations. In general, the generation patterns resulting
from these renewables and the electricity demand patterns are
far from being equal [172]. Therefore, effective utilization
of the DG in a way that is cognizant of the variability of
the yield from renewable sources is important. Second, the
authors of [114,269] indicated that the usual operation costs
of distributed generators for generating one unit of electricity
are high compared with that of traditional large-scale central
power plants. Considering the DG’s potential benefits on
power quality, a systematic research on how to balance the
high capital costs and the reliable power supplies brought by
DG is essential.
Although we can only see a limited penetration of DG in
today’s power system, the future SG is expected to adopt a
large number of distributed generators to form a much more
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 7
decentralized power system. As predicted in [114], it may
evolve from the present system in three stages:
1) Accommodating DGs in the current power system;
2) Introducing a decentralized system of DGs cooperating
with the centralized generation system;
3) Supplying most power by DGs and a limited amount by
central generation.
Note that as DG enables the users to deploy their own
generators, the large-scale deployment of DG will also change
the traditional power grid design methodology, in which the
generators are connected to the transmission grid (see Fig. 4).
The development and deployment of DG further leads to a
concept, namely Virtual Power Plant (VPP), which manages
a large group of distributed generators with a total capacity
comparable to that of a conventional power plant [172].
This cluster of distributed generators is collectively run by
a central controller. The concerted operational mode delivers
extra benefits such as the ability to deliver peak load electricity
or load-aware power generation at short notice. Such a VPP
can replace a conventional power plant while providing higher
efficiency and more flexibility. Note that more flexibility al-
lows the system to react better to fluctuations. However, a VPP
is also a complex system requiring a complicated optimization,
control, and secure communication methodology.
Traditional VPPs are studied in [6,36,155,265,266].
Anderson et al. [6] aimed to find and describe a suitable
software framework that can be used to help implement the
concept of a VPP in future power systems, and emphasized the
importance of Service Oriented Architecture in implementing
the VPP. Caldon et al. [36] proposed a cost based optimiza-
tion procedure for harmonizing the concurrent operation of
distribution system operator and VPP which, although acting
in an independent manner, can be coordinated by means of
suitable economic signals. Lombardi et al. [155] focused on
the optimization of the structure of the VPP. By using an
energy management system, a VPP can be controlled in order
to minimize the electricity production costs and to avoid
the loss of renewable energy. You et al. [265] proposed a
market-based VPP, which uses bidding and price signal as two
optional operations, and provides individual distributed energy
resource units with the access to current electricity markets.
You et al. [266] proposed a generic VPP model running under
liberalized electricity market environment, and attempted to
provide an outline of the main functions that are necessary
for the efficient operation of this generic VPP.
In addition, recently the integration of Vehicle-to-Grid
(V2G) technology (explained in Section IV-D) and VPP was
investigated in [117], which outlined an architecture of V2G
integrating VPP, provided a sketch of the trip-prediction algo-
rithm, and the associated optimization problem for the overall
system architecture.
Similar ideas of “virtual” have also been used for other
applications, such as virtual energy buffers [245] and virtual
energy provisioning systems [119].
B. Transmission Grid
On the power transmission side, factors such as infrastruc-
ture challenges (increasing load demands and quickly aging
components) and innovative technologies (new materials, ad-
vanced power electronics, and communication technologies)
drive the development of smart transmission grids. As stated
in [141], the smart transmission grid can be regarded as an
integrated system that functionally consists of three interactive
components: smart control centers,smart power transmission
networks, and smart substations.
Based on the existing control centers, the future smart
control centers enable many new features, such as analytical
capabilities for analysis, monitoring, and visualization.
The smart power transmission networks are conceptually
built on the existing electric transmission infrastructure. How-
ever, the emergence of new technologies (e.g new materials,
electronics, sensing, communication, computing, and signal
processing) can help improve the power utilization, power
quality, and system security and reliability, thus drive the
development of a new framework architecture for transmission
networks.
The vision of the smart substation is built on the existing
comprehensive automation technologies of substations. Al-
though the basic configurations of high-voltage substations
have not changed much over the years, the monitoring, mea-
surement, and control equipment have undergone a sea change
in recent years [24]. Major characteristics of a smart substation
shall include digitalization, autonomization, coordination, and
self-healing. By supporting these features, a smart substation is
able to respond rapidly and provide increased operator safety.
In brief, with a common digitalized platform, in the smart
transmission grid it is possible to enable more flexibility in
control and operation, allow for embedded intelligence, and
foster the resilience and sustainability of the grid.
C. Distribution Grid
For the distribution grid, the most important problem is how
to deliver power to serve the end users better. However, as
many distributed generators will be integrated into the smart
distributed grid, this, on one hand, will increase the system
flexibility for power generation, and on the other hand, also
makes the power flow control much more complicated, in turn,
necessitating the investigation of smarter power distribution
and delivery mechanisms.
An interesting research work was done by Takuno et al.
[235]. Takuno et al. proposed two in-home power distribution
systems, in which the information is added to the electric
power itself and electricity is distributed according to this
information. The first one is a circuit switching system based
on alternating current (AC) power distribution, and the other
is a direct current (DC) power dispatching system via power
packets. Note that the packetization of energy is an interesting
but challenging task since it requires high power switching
devices. Researchers have shown that silicon carbide junction
gate field-effect transistors are able to shape electric energy
packets [234]. Hence, the system proposed in [235] has the
potential as an intelligent power router. More specifically,
supplied electricity from energy sources is divided into several
units of payload. A header and a footer are attached to the unit
to form an electric energy packet. When the router receives
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 8
packets, they are sorted according to the addresses in the
headers and then sent to the corresponding loads. Using energy
packet, providing power is easily regulated by controlling the
number of sent packets. In addition, many in-home electric
devices are driven by DC power and have built-in power
conversion circuits to commutate AC input voltage. Thus, DC-
based power distribution is feasible. These systems will make
in-home power distribution systems more efficient and easier
to control energy flow.
D. Some New Grid Paradigms
In this subsection, we describe two of the most important
new grid paradigms, which benefit from smart energy subsys-
tem technologies and also further promote the development of
SG. These two paradigms are widely regarded as important
components of the future SG. Note that these two paradigms
also take advantage of other SG technologies as we will
explain in the corresponding sections.
1) Microgrid
Distributed generation promotes the development of a new
grid paradigm, called microgrid, which is seen as one of the
cornerstones of the future SG [68]. The organic evolution
of the SG is expected to come through the plug-and-play
integration of microgrids [70]. A microgrid is a localized
grouping of electricity generations, energy storages, and loads.
In the normal operation, it is connected to a traditional power
grid (macrogrid). The users in a microgrid can generate low
voltage electricity using distributed generation, such as solar
panels, wind turbines, and fuel cells. The single point of
common coupling with the macrogrid can be disconnected,
with the microgrid functioning autonomously [122]. This oper-
ation will result in an islanded microgrid, in which distributed
generators continue to power the users in this microgrid
without obtaining power from the electric utility located in the
macrogrid. Fig. 8shows an example of the microgrid. Thus,
the multiple distributed generators and the ability to isolate the
microgrid from a larger network in disturbance will provide
highly reliable electricity supply. This intentional islanding of
generations and loads has the potential to provide a higher
local reliability than that provided by the power system as
a whole [136]. Note that although these users do not obtain
the power from outside in the islanding mode, they may still
exchange some information with the macrogrid. For instance,
they may want to know the status of the macrogrid and decide
whether they should reconnect to the macrogrid and obtain
power from the electric utility.
Lasseter [135] also pointed out that using microgrids in the
distribution system is straightforward and also simplifies the
implementation of many SG functions. This includes improved
reliability, high penetration of renewable sources, self-healing,
active load control, and improved efficiencies. For example, in
order to realize self-healing during outages, microgrids can
switch to the islanding mode and as a result the users in
microgrids will not be affected by outages.
2) G2V and V2G
An electric vehicle is a vehicle that uses one or more electric
motors for propulsion. As fossil fuels diminish and generally
Fig. 8: An Example of a Microgrid: The lower layer shows a
physical structure of this microgrid, including four buildings,
two wind generators, two solar panel generators, and one
wireless access point (AP). These buildings and generators
exchange power using powerlines. They exchange information
via an AP-based wireless network. The blue (top) layer shows
the information flow within this microgrid and the red (middle)
layer shows the power flow.
get more expensive, fully electric vehicles or plug-in hybrid
electric vehicles will rise in popularity. In the following, we
use EV to represent both fully electric vehicle and plug-
in hybrid electric vehicle. The wide use and deployment of
EVs leads to two concepts, namely Grid-to-Vehicle (G2V) and
Vehicle-to-Grid (V2G).
In G2V, EVs are often powered by stored electricity
originally from an external power source, and thus need to
be charged after the batteries deplete. This technology is
conceptually simple. However, from the perspective of the
grid, one of the most important issues in G2V is that the
charging operation leads to a significant new load on the
existing distribution grids. In the literature, many works have
studied the impact of EVs on the power grid.
Schneider et al. [224] pointed out that the existing distri-
bution grid infrastructure in the Pacific Northwest is capable
of supporting a 50% penetration of EVs with the 120V smart-
charging profile, which equates to approximately 21.6% of the
light duty vehicle fleet. This level of penetration exceeds the
known capability of the existing generation resources, which
is approximately 18%. The authors of [92, 210, 228] further
pointed out that serious problems (e.g.significant degradation
of power system performance and efficiency, and even over-
loading) can arise under high penetration levels of uncoordi-
nated charging.
One solution to mitigate the impact of EVs on the grid is
to optimize their charging profile. In other words, we need
to keep the peak power demand as small as possible, taking
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 9
into account the extra power consumption from the vehicle
charging. This can be done by coordinating the charging
operations of different EVs so that they are not charged at
the same time. For example, Clement et al. [48] show that the
coordinated charging of EVs can improve power losses and
voltage deviations by flattening out peak power.
In V2G, EVs provide a new way to store and supply electric
power. V2G-enabled EVs can communicate with the grid to
deliver electricity into the grid, when they are parked and
connected to the grid. Note that as reported by Kempton et al.
[127], in the U.S. the car is driven only one hour a day on
average. In other words, these cars are parked most of the time
doing nothing. There exist three major delivery setups:
1) A hybrid or fuel cell vehicle, which generates power
from storable fuel, uses its generator to produce power
for a utility at peak electricity usage times. These vehi-
cles serve as a distributed generation system producing
energy from conventional fossil fuels or hydrogen.
2) A battery-powered or plug-in hybrid vehicle uses its
excess rechargeable battery capacity to supply power
for a utility at peak electricity usage times. These
vehicles can then be recharged during off-peak hours
at cheaper rates. These vehicles serve as a distributed
battery storage system to store power.
3) A solar vehicle which uses its excess charging capacity
to provide power to the power grid when the battery is
fully charged. These vehicles serve as a distributed small
renewable energy power system.
Thus far, researchers have focused on the connection between
batteries and the power grid [126,244], the validity of the V2G
system [123], the feasible service [244], its environmental
and economic benefits [186], its new markets [124,125],
and system integration [258]. Utilities currently also have
V2G technology trials. For example, Pacific Gas and Electric
Company tried to convert a number of company-ownedToyota
Prius to V2G plug-in hybrids at Google’s campus [28]. Xcel
Energy performed the nation’s first large test of V2G-enabled
EVs in Boulder, Colorado, as part of its internationally recog-
nized SmartGridCity project [260].
Note that G2V and V2G are not fully separated concepts
in the vision of SG. For example, V2G-enabled EVs are often
used to provide power to help balance loads by “peak shaving”
(sending power back to the grid when demand is high) but also
“valley filling” (charging when demand is low). Therefore, a
key question is how to determine the appropriate charge and
discharge times throughout the day. Hutson et al. [107] studied
this problem and used a binary particle swarm optimization
algorithm to look for optimal solutions that maximize profits
to vehicle owners while satisfying system and vehicle owners’
constraints. Note that particle swarm optimization is an iter-
ative stochastic optimization algorithm. The solution search
is performed in a stochastic nature allowing the algorithm
to overcome nonlinear, non-differentiable, and discontinuous
problems.
E. Summary and Future Research
In this subsection, we have reviewed the work on the
smart energy subsystem, more specifically, power generation,
transmission, and distribution. We have also described two new
grid paradigms: microgrid and G2V/V2G. In the following, we
list several research challenges and possible future research
worth exploring.
1) Effective utilization of intermittent and fluctuant
renewables: It is believed that distributed renewable energy
generation will be widely used in SG. However, the utilization
of distributed renewable energy resources also poses many
challenges and opens up many new research topics.
The key problem is how to model renewable energy source.
The intermittent and fluctuant nature of wind and solar
generation requires much more complicated forecasting and
scheduling. Both long-term and short-term renewable source
patterns and likely behavior must be understood and explored
[198].
For example, He et al. [100] divided a 24-hour period into
Mslots of length T1each; and each T1-slot, in turn, consists
of Kslots, each of length T2. The wind generation can be
modeled as a non-stationary Gaussian random process across
the T1-slots, i.e., the wind generation amount in the kth T2-
slot of the mth T1-slot follows N(θ, σ2), where Nis the
Gaussian distribution, θis the mean, and σ2is the variance.
In addition, finite-state Markov models have also been used as
an effective approach to characterize the correlation structure
of the renewable energy outputs [66, 181]. Since we may
not fully observe the system transition state, hidden Markov
models are also used in modeling renewable energy systems
[32, 33, 175, 230]. Considering that in practice the power
pattern of renewable resources may not follow any simple
distribution or Markov process, Fang et al. [69] further used
non-stochastic multi-armed bandit online learning technique
to learn the evolution of power pattern of renewable energy
source. Note that online learning is a model of induction
that learns the label of one instance at a time. The goal in
online learning is to predict labels for instances. A typical
application could be that the instances are able to describe
the current conditions of the renewable sources, and an online
algorithm predicts tomorrow’s value of a particular source. In
summary, in order to effectively utilize the renewable energy,
more thorough mathematical analysis on modeling renewable
energy is desirable.
Another possible research topic is the optimal deployment
of the additional ancillary services (e.g. energy reserves) to
maintain reliability and meet operational requirements, taking
into account the uncertainty and variability of renewable
energy resources.
2) Utilization of G2V/V2G:
In G2V, the challenge is that vehicle charging will lead to
a significant new load on the existing distribution grids, with
many of these circuits not having any spare capacity. In V2G,
the challenge is the availability of EVs, since an EV can only
deliver power to the grid when it is parked and connected to
grid. As a result, this increases the uncertainty of the power
supplied by EVs.
It is easy to see that the above two challenges lead to
an urgent need for an analysis of large-scale EV stochastic
behavior. More specifically, we can use probability theory or
experiments to model the power request profile for a large
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 10
number of EVs charging operations, and the total available
power profile provided by a large number of EVs. Although
we cannot accurately predict the behavior of each EV, it is
very likely that over a large dataset, the overall profile must
follow some distribution. Let us recall the normal distribution,
one of the most famous distributions. According to the central
limit theorem [63], the mean of a sufficiently large number
of independent random variables, each with finite mean and
variance, follows the normal distribution. This analysis can
help the operator pre-design the system capacity margins.
In addition, queueing theory [87] could also play an impor-
tant role in G2V analysis. Assume that an EV charging station
works as a queue, i.e., serving EVs sequentially. Queueing
theory [87] enables mathematical analysis of several related
processes, including arrival of EVs at the queue, waiting in
the queue for being served, and being served at the front
of the queue. Thus, we can predict the expected number
of EVs waiting or receiving service, and the probability of
encountering the charging system in certain states, such as
empty, full, having an available server, or having to wait a
certain amount of time to be served.
3) Challenges in large-scale deployment: A deployed
large-scale commercial SG may have tens of millions of nodes.
So far we have little experience in large-scale distributed
control approaches to addressing the complex power system
component interactions, and in large-scale deployment of new
technologies, such as batteries, thermal storages, DGs, and
EVs [146]. This challenge requires us to think about how to
organize so many devices in a large-scale SG. Two approaches
may be applicable: top-down and bottom-up.
In the top-down approach, the high-level framework of the
system is formulated by a powerful grid operator, and each
subsystem is then refined in greater details. For example, a
group of users in a microgrid can refine their own connection
structure based on the high-level framework defined by an
upper supervisor. However, this approach needs a powerful
operator to initially design the whole architecture, which is
not an easy task.
The bottom-up approach is the piecing together of systems
to give rise to grander systems. For example, a group of users
first link together to form a system. Then these systems link
together to form a larger system. Although this methodology
does not need a powerful operator to initially design the whole
architecture, the final system grows up from many individually
formed subsystems. Therefore, the performance of the whole
system may not be good enough.
The advantages and disadvantages of both top-down and
bottom-up approaches need to be investigated. Furthermore,
self-organization is a topic worth exploring. For example, in
the bottom-up approach, one question is how a group of users
or devices can be self-organized to form a system.
In addition, open, scalable, and instructive standards will
play an important role in the large-scale deployment of SG. Let
us take the IEEE P2030 guide for SG interoperability [109] as
an example. Using a system of systems approach, this guide
defines three perspectives: power systems, information tech-
nology, and communications. Moreover, these interoperability
architectural perspectives are comprised of domains, entities,
interfaces, and data flows. The domains are the same as the
seven domains mentioned in Section II. Entities (devices, com-
munication networks, computer systems, software programs,
etc.) are generally located inside a domain and are connected
to each other through one or more interfaces. Interfaces are
logical connections from one entity to another that support one
or more data flows implemented with one or more data links.
These data flows are application-level communications from
entities that provide data to entities that consume data.
This guide does not specify which particular technology
should be used. Instead, it aims to establish both entities and
relationships within the environment of the SG and define
interfaces in a technology-agnostic manner. For example,
it defines 20 entities and 81 interfaces among the major
entities in each of the domains. The methodology of this
guide is applicable to all SG implementations. It is general
enough to allow the implementation of newer technologies and
changing conditions in the utility’s operational environment.
This methodology is of great importance for the large-scale
deployment and interoperability of the SG. For example,
each entity can design its own implementation. As long as
its implementation follows the standards, the whole system
will work. In other words, the large-scale deployment task is
hence divided into several small basic tasks, while these small
tasks can be coordinated based on the well-defined interfaces
and relationships. Different functions of SG can be realized
by using different combinations of basic tasks. [109] shows
several such examples. In this way, the complexity of building
a large-scale SG is significantly reduced. Therefore, designing
a highly scalable, open, and instructive standard is of great
importance, albeit difficult.
V. SMART INFRASTRUCTURE SYSTEM II - SMART
INFORMATION SUBSYSTEM
The evolution of SG relies on not only the advancement
of power equipment technology, but also the improvement
of sophisticated computer monitoring, analysis, optimization,
and control from exclusively central utility locations to the
distribution and transmission grids. Many of the concerns of
distributed automation should be addressed from an informa-
tion technology perspective, such as interoperability of data
exchanges and integration with existing and future devices,
systems, and applications [109]. Therefore, a smart infor-
mation subsystem is used to support information generation,
modeling, integration, analysis, and optimization in the context
of the SG.
In this section, we concentrate on the smart information
subsystem. We first explore the information metering and
measurement, which generates information from end entities
(e.g. smarter meters, sensors, and phasor measurement units)
in an SG. This information is often used for billing, grid status
monitoring, and user appliance control. We then explore the
information management, including data modeling, informa-
tion analysis, integration, and optimization. We finally outline
some future research directions and challenges.
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 11
A. Information Metering and Measurement
Study in information metering and measurement can be
classified into smart metering, and smart monitoring and
measurement as shown in Fig. 9.In the following, we describe
this classification in detail.
Information Metering, Monitoring, and
Measurement
Smart Metering
[71,96,251]Smart Monitoring and Measurement
Sensor
[26,139]
[90,91]
[156,157]
[256]
Phasor Measurement Unit
[11,20,22,23,38,56,110]
[137,160,183,184,197]
[196,199,221,237,238]
[243,248,252,270–272]
Fig. 9: Classification of the Work on the Information Metering
and Measurement
1) Smart Metering
Smart metering is the most important mechanism used in the
SG for obtaining information from end users’ devices and
appliances, while also controlling the behavior of the devices.
Automatic metering infrastructure (AMI) systems [96], which
are themselves built upon automatic meter reading (AMR)
systems [208], are widely regarded as a logical strategy to
realize SG. AMR is the technology of automatically collecting
diagnostic, consumption, and status data from energy metering
devices and transferring that data to a central database for
billing, troubleshooting, and analyzing. AMI differs from
traditional AMR in that it enables two-way communications
with the meter. Therefore nearly all of this information is
available in real time and on demand, allowing for improved
system operations and customer power demand management.
Smart meters, which support two-way communications be-
tween the meter and the central system, are similar in many
aspects to AMI meters, or sometimes are regarded as part of
the AMI. A smart meter is usually an electrical meter that
records consumption in intervals of an hour or less and sends
that information at least daily back to the utility for monitoring
and billing purposes [71]. Also, a smart meter has the ability to
disconnect-reconnect remotely and control the user appliances
and devices to manage loads and demands within the future
“smart-buildings.” Fig. 10 shows a typical usage scenario for
smart meters.
From a consumer’s perspective, smart metering offers a
number of potential benefits. For example, end users are able
to estimate bills and thus manage their energy consumptions
to reduce bills. From a utility’s perspective,they can use smart
meters to realize real-time pricing, which tries to encourage
users to reduce their demands in peak load periods, or to
optimize power flows according to the information sent from
demand sides.
2) Smart Monitoring and Measurement
An important function in the vision of SG is monitoring
Fig. 10: An Example of the Smart Metering Structure: The
smart meter collects the power consumption information of
the dishwasher, TV, and the refrigerator, and also sends the
control commands to them if necessary. The data generated
by the smart meters in different buildings is transmitted to a
data aggregator. This aggregator could be an access point or
gateway. This data can be further routed to the electric utility
or the distribution substation. Note that the smart communi-
cation subsystem, described in Section VI, is responsible for
the information transmission.
and measurement of grid status. We review the following
two major monitoring and measurement approaches, namely
sensors and phasor measurement units.
Sensors: Sensors or sensor networks have already been used as
a monitoring and measurement approach for different purposes
[1]. In order to detect mechanical failures in power grids such
as conductor failures, tower collapses, hot spots, and extreme
mechanical conditions, Leon et al. [139] proposed that sensor
networks should be embedded into the power grid and help
to assess the real-time mechanical and electrical conditions of
transmission lines, obtain a complete physical and electrical
picture of the power system in real time, diagnose imminent
as well as permanent faults, and determine appropriate control
measures that could be automatically taken and/or suggested
to the system operators once an extreme mechanical condition
appears in a transmission line.
Wireless sensor networks (WSNs) in particular, given their
low cost, can provide a feasible and cost-effective sensing
and communication platform for remote system monitoring
and diagnosis. Gungor et al. [91] reviewed the application
of WSNs for electric power systems along with their op-
portunities and challenges and presented a comprehensive
experimental study in different electric power system environ-
ments. They concluded that with the help of WSN, a single
system contingency in the power grid could be detected and
isolated before it causes cascading effects and leads to more
catastrophic system-wide breakdowns.
Other research work on applying WSNs for SG are [26,
156,157]. Bressan et al. [26] explored the implementation
of a smart monitoring system over a WSN, with particular
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 12
emphasis on the creation of a solid routing infrastructure
using the routing protocol for low-power and lossy networks,
whose definition is currently being discussed within the IETF
ROLL working group. Lu et al. [156] proposed a closed-loop
energy management scheme with a WSN, which was applied
as the architecture of an industrial plant energy management
system. The importance of the proposed scheme lies in its non-
intrusive, intelligent, and low cost nature. Later, Lu et al. [157]
extended the previous work [156], and studied the overall
system architecture.
However, the use of sensor networks in the SG has many
requirements: [90,91,256]:
1) Quality-of-Service (QoS) requirements: The information
generated by sensor networks may be associated with
some data QoS requirements, such as reliability, latency,
and network throughput. For example, the critical sensed
data related to grid failures should be received by
the controller in a timely manner. The communication
subsystem (described in Section VI) supporting sensor
networks must provide mechanisms to satisfy these QoS
requirements.
2) Resource constraints: Sensor nodes are often low cost
and resource limited devices. Thus the control programs
for sensor networks should be energy efficient.
3) Remote maintenance and configuration: Sensors must be
remotely accessible and configurable, so that the sensor
networks could be maintained remotely, conveniently,
and promptly.
4) High security requirements: Security is very important
for electric power systems. By compromising sensors,
attackers can jeopardize the power grid operation.
5) Harsh environmental conditions: In SG environments,
sensors may be subject to radio frequency (RF) in-
terference, highly caustic or corrosive environments,
high humidity levels, vibrations, dirt and dust, or other
conditions that may cause a portion of sensor nodes
to malfunction. Hence the sensor network design must
consider the survivability requirement, i.e., the sensor
network is still connected or the critical areas are still
monitored if some sensors fail.
Phasor Measurement Unit: Recent developments in the SG
have spawned interest in the use of phasor measurement units
(PMUs) to help create a reliable power transmission and
distribution infrastructure [11]. A PMU measures the electrical
waves on an electrical grid to determine the health of the
system. Technically speaking, a phasor is a complex number
that represents both the magnitude and phase angle of the sine
waves found in electricity. Phasor measurements that occur
at the same time are called synchrophasor, as are the PMU
devices that allow their measurement. Typically, PMU readings
are obtained from widely dispersed locations in a power
system network and synchronized using the global positioning
system (GPS) radio clock. With a large number of PMUs and
the ability to compare shapes from alternating current (AC)
readings everywhere on the grid, system operators can use the
sampled data to measure the state of the power system and
respond to system conditions in a rapid and dynamic fashion.
Refer to [56] for a technical introduction to the PMU.
Phasor measurements using GPS based time synchroniza-
tion were introduced in the mid-1980s [22,196,197,252].
A Virginia Tech research team developed the first prototype
PMU in 1988 [271]. Later, the frequency monitoring network
(FNET) project utilized a network of low-cost, high-precision
frequency disturbance recorders to collect synchrophasor data
from the U.S. power grid [184]. Recently Zhang et al. [271]
presented some of the latest implementations of FNET’s
applications by using PMUs, which are significantly better at
observing power system problems than the earlier implemen-
tations. The current FNET system hierarchy is suitable for
high volume data transfer, processing, storage, and utilization.
A variety of applications, especially with regards to real-time
dynamic monitoring, have been developed and integrated into
the system. FNET is growing into a mature, multifunctional,
and low-cost phasor measurement system with stable perfor-
mance and high accuracy.
Early research on the applications of PMU technology was
mainly focused on validation of system models and accurate
postmortem analysis. However, now with wide-scale real-
time PMU data being obtainable, system operators have the
capability of deploying system state estimation procedures and
system protection functionalities in power grids, with the goal
of making the power system immune to catastrophic failures.
Several countries, such as Brazil, China, France, Japan, South
Korea, Mexico, Norway, and the U.S., have installed PMUs
on their power grid systems for research or are developing
prototypes [270]. The installation of PMUs on transmission
grids of most major power systems has become an important
activity.
In addition, the investigation of PMUs is an exciting area
being explored by both industry and academia. Industry is
investigating how to install the PMUs, collect the data, and
establish communication transfers of this data to the utility
control centers [110,183]. In academia, typical research fields
are the applications of PMU for grid protection functions,
such as providing loss-of-mains protection [137], monitoring
fault event [221,237,238,243,272], locating disturbance
[160], estimating grid state [199,248], studying synchronous
islanded operation [20], monitoring power quality [38], and
devising experimental applications for the monitoring of active
distribution grids [23].
B. Information Management
In SG, a large amount of data and information will be
generated from metering, sensing, monitoring, etc. SG must
support advanced information management. The task of the
information management is data modeling, information anal-
ysis, integration, and optimization.
1) Data Modeling
As stated by IEEE P2030 [109], the goal of SG information
technology data modeling is to provide a guide to creating
persistent, displayable, compatible, transferable, and editable
data representation for use within the emerging SG. In other
words, the objective is to make it as interoperable as possible
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 13
using relevant standards. That is specifically addressing the
data that represents state information about the grid and
individual items in it. This would include nearly all connected
items from generation down to individual consuming devices.
They all have state information that may need to be read,
stored, transmitted, etc.
Why is data modeling important? Let us look at the follow-
ing two reasons. First, the information exchange between two
application elements is meaningful only when both of them
can use the information exchanged to perform their respective
tasks. Therefore, the structure and meaning of the exchanged
information must be understood by both application elements.
Although within the context of a single application, developers
can strive to make the meaning clear in various user interfaces,
when data is transferred to another context (another system),
the meaning could be lost due to incompatible data represen-
tation. Considering that the SG is a complicated system of
systems, design of a generally effective data representation is
very important.
Second, the data modeling is also related to the system
forward compatibility and backward compatibility. On one
hand, a well-defined data model should make legacy program
adjustments easier. We hope that the data representation de-
signed for SG can also be (or at least partially) understood
by the current power system, in order to take advantage
of the existing infrastructure as much as possible. On the
other hand, thus far SG is more like a vision. Its definition
and functionality keep evolving. Suppose that in the current
implementation, all the data is particularly designed to be
stored in an optimized way that can be understood by a
current application X. After some time, a new application Yis
integrated into SG. Data modeling is the key to whether this
new application can understand the historical data and obtain
enough information from the historical data.
IEEE 2030 [109] pointed out that ontology may be a good
option, because it is becoming an increasingly popular way
of providing a data model with formal semantics based on a
shared understanding that is machine-readable. Ontology helps
convey knowledgein a formal fashion, just like a programming
language conveys mathematics in a formal fashion. With ontol-
ogy, one speaks of concepts in a subject area, and relationships
between them. Like a programming language, it helps define,
clarify, and standardize what is being discussed. Another
reason that ontology-based strategies are commonly used with
success in creating and manipulating data models is that
they provide easy export or translation to Extensible Markup
Language (XML) or Unified Modeling Language (UML),
which provides for a great deal of information interoperability.
2) Information Analysis, Integration, and Optimization
Information analysis is needed to support the processing, inter-
pretation, and correlation of the flood of new grid observations,
since the widely deployed metering and monitoring systems
in SG will generate a large amount of data for the utility.
As mentioned in [109], one part of the analytics would be
performed by existing applications, and another part of the
analytics dimension is with new applications and the ability
of engineers to use a workbench to create their customized
analytics dashboard in a self-service model.
Information integration aims at the merging of information
from disparate sources with differing conceptual, contextual,
and typographical representations. In SG, a large amount of
information has to be integrated. First, the data generated by
new components enabled in SG may be integrated into the
existing applications, and metadata stored in legacy systems
may also be used by new applications in SG to provide new
interpretations. IEEE P2030 [109] indicated that data integrity
and name services must be considered in information integra-
tion. Data integrity includes verification and cross-correlation
of information for validity, and designation of authoritative
sources and who are responsible for specific personnel who
own the data. Name service addresses the common issue of
an asset having multiple names in multiple systems.
Second, as stated in [115, 212], currently most utility
companies have limited installed capability for integration
across the applications associated with system planning, power
delivery,and customer operations. In most cases, this informa-
tion in each department is not easily accessible by applications
and users in other departments or organizations. These “islands
of information” correspond to islands of autonomous business
activities. Therefore, the emerging SG calls for enterprise-
level integration of these islands to improve and optimize
information utilization throughout the organization.
Information optimization is used to improve information
effectiveness. The data size in the future SG is expected
to be fairly large as a result of the large-scale monitoring,
sensing, and measurement. However, the generated data may
have a large amount of redundant or useless data. There-
fore, we need to use advanced information technology to
improve the information effectiveness, in order to reduce
communication burden and store only useful information. This
problem has been studied, among others, by [180, 257]. In
order to compress the size of disturbance signals and reduce
sinusoidal and white noise in the signals, Ning et al. [180]
proposed a wavelet-based data compression approach for SG.
The proposed method can be implemented in SG to mitigate
data congestion and improve data transmission and quality.
Wang et al. [257] applied the singular value decomposition
analysis to examine the coupling structure of an electrical
power grid in order to highlight opportunities for reducing the
network traffic, by identifying what are the salient data that
need to be communicated between parts of the infrastructure to
apply a control action. They found that typical grid admittance
matrices have singular values and vectors with only a small
number of strong components.
C. Summary and Future Research
In this section, we reviewed the work on the smart infor-
mation subsystem, especially information metering, measure-
ment, and management in SG. We list the followingchallenges
and possible directions worth exploring.
1) Effective information store: A large amount of informa-
tion, such as the data from smart meters, sensors, and PMUs,
will be sampled in SG, and sent to the control system. One
important problem is what information should be stored in
the control system so that meaningful system or user history
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 14
can be constructed from this data. Note that system history is
important for analyzing system operations, and user history is
important for analyzing user behaviors and bills. Considering
that the amount of information received by the control system
is huge, solving this problem is challenging.
We suggest using data mining, machine learning, and infor-
mation retrieval techniques to analyze the information and thus
obtain the representative data. Furthermore, the correlation
among some data may be high. For example, the smart meter
readings must be similar when no activity takes place at home.
This opens a door for significantly reducing the amount of
information needed to store by using data compression. In
addition, database tools should be used to organize, store, and
retrieve this data.
2) The utilization of cloud computing: Cloud com-
puting has been envisioned as the next-generation comput-
ing paradigm for its major advantages in on-demand self-
service, ubiquitous network access, location independent re-
source pooling, and transference of risk [99]. The basic idea
of the cloud computing is that the cloud providers, who operate
large data centers with massive computation and storage
capacities, deliver computing as a service, whereby shared
resources, software and information are provided to computers
and other devices as a utility over a network. Integrating cloud
computing may improve the information management in SG.
First, since cloud providers have massive computation and
storage capacities, they can design some basic and generic
information management services for electric utilities. There-
fore, electric utilities can focus on more advanced and com-
plicated information management functions, while outsourcing
the basic and generic information management functions to
the cloud. As a result, electric utilities do not have to develop
the information management functions from scratch. This is
especially useful for small utilities or even personal users
who provide power service. Let us consider an example.
Since distributed renewable generations are expected to be
widely used in the emerging SG, a user equipped with a
distributed renewable generator may want to sell his excess
energy to other users nearby. This user can outsource the basic
information storage and management to the cloud, and thus has
no need to design his own information management system.
Second, cloud computing may be able to improve the
information integration level in SG. For example, as mentioned
before in most cases, the information in each department
is not easily accessible by applications and users in other
departments or organizations. These “islands of information”
correspond to islands of autonomous business activities. If all
the information is stored and managed by a cloud, it actually
provides a relatively cost-effective way to integrate these
islands of information. As stated by Rusitschka et al. in [215],
the ease-of-interfacing with the cloud has the potential to
create usable de facto standards while enabling interoperability
and extensibility.
Although using cloud computing may improve the informa-
tion management in SG, it also poses many challenges. First,
information security and privacy must be the major concern
of electric utilities, since the information storage and man-
agement is out of the control of electric utilities. Rusitschka
et al. [215] discussed the confidentiality and privacy issues,
and proposed some solutions, including designing multi-tenant
data architectures, and applying pseudonymization or crypto-
graphic hashes. Second, it is unlikely that an electric utility
outsources all the information management functions to the
cloud. Therefore, we ask two questions:
1) From the cloud provider’s perspective, which informa-
tion management services should be provided to maxi-
mize its own profit?
2) From the electric utility’ perspective, which information
management functions should be outsourced and which
should be operated by itself to maximize its own profit?
VI. SMART INFRASTRUCTURE SYSTEM III - SMART
COMMUNICATION SUBSYSTEM
The third part in the smart infrastructure system is the smart
communication subsystem. This subsystem is responsible
for communication connectivity and information transmission
among systems, devices, and applications in the context of the
SG.
In this section, we first give an overview of the smart com-
munication subsystem in SG. We then describe wireless and
wired communication technologies in Sections VI-B and VI-C,
respectively. In Section VI-D, we describe how to manage end-
to-end communications in this heterogenous communication
system, where various communication technologies, network
structures, and devices may be used. We finally outline some
future research and research challenges.
A. An Overview
The most important question in the communication sub-
system is, “What networking and communication technology
should be used?” While there is a general agreement on the
need for communication networks to support a two-way flow
of information between the various entities in the electric grid,
there is still much debate on what specific technologies should
be used in each SG application domain and how they should
be implemented [231]. One reason why this is still not clear
is that the SG consists of many different types of networks,
including for example
1) Enterprise bus that connects control center applications,
markets, and generators;
2) Wide area networks that connect geographically distant
sites;
3) Field area networks that connect devices, such as intel-
ligent electronic devices that control circuit breakers and
transformers;
4) Premises networks that include customer networks as
well as utility networks within the customer domain.
Although thus far the answer is not clear, since reliable and
effective information exchange is a key to the success of the
future SG, a communication subsystem in an SG must at least
satisfy the following basic requirements:
1) The communication subsystem must support the quality
of service (QoS) of data [144]. This is because the
critical data (e.g. the grid status information) must be
delivered promptly.
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 15
Communication Technologies
Wireless Wired
Wireless Mesh
Network
[2,3]
[81,90]
Cellular
Communications
[3,102]
[106,109]
Cognitive
Radio
[82,161,202]
[203,232]
IEEE 802.15
[3,70]
[177,264]
Satellite
Communications
[57,90,242]
Microwave or Free-Space
Optical Communications
[27,60]
Fiber-Optic
Communications
[90,163]
Powerline
Communications
[16,54,73,76,77]
[78,90,131,149]
[165,195,208,262,273]
Fig. 11: Classification of Relevant Research in Communication Technologies in SG
2) The communication subsystem must be highly reliable.
Since a large number of devices will be connected and
different devices and communication technologies will
be used, guaranteeing the reliability of such a large and
heterogeneous network is not a trivial task.
3) The communication subsystem must be pervasively
available and have a high coverage. This is mandated
by the principle that the SG can respond to any event in
the grid in time.
4) The communication subsystem must guarantee security
and privacy. In Section VIII, we will discuss the security
and privacy issues of information transmission in SG.
In the rest of this section, we focus on the communication
and networking technologies which are applicable in SG. We
describe wireless and wired communication technologies in
Sections VI-B and VI-C, respectively. Fig. 11 shows a classi-
fication of the work related to communication technologies in
SG. Fig. 12 shows an example of a communication network
used in SG.
B. Wireless Technologies
Wireless technologies not only offer significant benefits over
wired technologies, such as low installation cost, rapid deploy-
ment, mobility, etc., but are also more suitable for remote end
applications [193]. Wireless has already been widely used in
our daily life and can be deployed anywhere and anytime.
We list the following important wireless communication and
networking technologies which may be applicable in future
SG.
Wireless Mesh Network: Wireless mesh network (WMN),
which is a communication network made up of radio nodes
organized in a mesh topology, has emerged as a key tech-
nology for next-generation wireless networking [2]. Industrial
standards groups, such as IEEE 802.11 and 802.16, are all
Fig. 12: An example of a communication network in SG:
User devices and smart meters use ZigBee, WiFi, and pow-
erline communications. Wireless mesh networks are used for
information exchanges between users. Communities are con-
nected to their electric utility via free-space optical, satellite,
microwave, or cellular systems. A substation communicates
with an electric utility over the powerline.
actively working on new specifications for WMNs. A WMN
also provides basic networking infrastructure for the commu-
nications in SG. Some of the benefits of using WMNs in SG
are highlighted as follows:
1) Increased communication reliability and automatic net-
work connectivity: Since redundant paths usually exist
in WMNs, network robustness against potential prob-
lems, e.g., node failures and path failures, can be im-
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 16
proved. Furthermore, generally speaking a WMN is self-
organized and self-configured. This feature is crucial
for electric system automation, since it enables electric
utilities to cope with new connectivity requirements
driven by customer demands [90].
2) Large coverage and high data rate: WiMAX mesh
network can enable both long distance and high data rate
communications. In the SG, a large amount of data, such
as the information from smart meters, sensors, and pha-
sor measurement units (PMUs), will be generated and
sent to the control system. A communication network
with large coverage and high data rate is necessary.
Traditionally, electricity power grid and WMN are two
parallel concepts. In the literature, there exists a large amount
of work focused on WMN technologies. An excellent survey
on the general concepts of WMNs can be found in [2]. The
emerging SG will attempt to integrate WMN into power grid.
Recently, some researchers have conducted some studies
along this line. Gharavi and Hu [81] presented a multi-gate
mesh network architecture to handle real-time traffic for the
last mile communication. The multigate routing is based on a
flexible mesh network architecture that expands on the hybrid
tree routing of the IEEE 802.11s. The network is specifically
designed to operate in a multi-gateway structure in order to
meet the SG requirements in terms of reliability, self-healing,
and throughput performance. More detailed studies on wireless
mesh technologies used in electric power system can be found
in [3, 90].
Cellular Communication Systems: A cellular communica-
tion system, such as GSM [174] and 3G (WCDMA [103] and
CDMA-2000 [263]), is a radio network distributed over land
areas called cells, each served by at least one fixed-location
transceiver known as a cell site or base station. It has been
a proven mature technology for data transmission for several
decades. By using the existing 3G (or even 4G [105]) cellular
communication systems, it is quick and inexpensive to obtain
data communications coverage over a large geographic area
[3].
Researchers have also conducted some studies on cellular
communications for the SG. For example, Hung et al. [106]
studied a new network model in which sensor/relay nodes
can also communicate with other back-end nodes using a
wide area network such as the cellular network, and proved
that the delay and cost of transmitting data can be reduced.
Hochgraf et al. [102] proposed to use the GSM network and
Short Message Service (SMS) messages as an option for SG
communications, and presented a system that provides the
control of thousands of mobile electric vehicle (EV) chargers
using a simple SMS message interface. In addition, using 3G
cellular communication system as the backhaul network has
also been recommended by IEEE P2030 [109]. Compared with
WMN, the biggest strength of cellular communication systems
is the pervasiveness of this mature technology.
Cognitive Radio: The communication system in SG needs to
be designed to accommodate the current management require-
ments as well as the potential demand of future applications.
It is likely that unlicensed spectrum will also be used when
SG is in large-scale commercial use.
Ghassemi et al. [82] proposed an application of cognitive
radio for the SG based on the IEEE 802.22 standard. In rural
areas, the standalone option can provide broadband access to
the geographically spread customers. In urban areas, IEEE
802.22 transceivers can be used as secondary radios to handle
high volumes of non-critical data and also act as backup
radios in emergency situations. Ma et al. [161] proposed
to communicate through a cognitive radio link between the
sensors at the consumer side and the control center of the
SG. Therefore, the state estimator needs to adjust to this new
communication link as the link is affected by primary users.
This link is governed by multiple semi-Markov processes
each of which can capture and model one channel of the
cognitive radio system. Sreesha et al. [232] proposed a multi
layered approach to provide energy and spectrum efficient
designs of cognitive radio based wireless sensor networks
at the smart grid utility. Their design provides a reliable
and low-latency routing support for large-scale cognitive SG
networks. Qiu et al. [202] built a real-time CR network testbed,
which can help tie together CRs in the next-generation SG
network. Later, Qiu et al. [203] systematically investigated
the idea of applying CR for SG, studied system architecture,
algorithms, and hardware testbed, and proposed a microgrid
testbed supporting both power flow and information flow.
Wireless Communications based on 802.15.4: Three wire-
less communication technologies based on IEEE 802.15.4
protocol stack are recommended to be used in SG [3]. They
are ZigBee, WirelessHART, and ISA100.11a. ZigBee is a
wireless technology which is designed for radio-frequency
applications that require a low data rate, long battery life, and
secure networking. It might be one of the most widely used
communication technologies in the customer home network.
The ZigBee and ZigBee Smart Energy Profile (SEP) have been
defined as the one of the communication standards for use
in the customer premise network domain of the SG by the
U.S. National Institute of Standards and Technology (NIST)
[177]. It has also been selected by many electric utilities as
the communication technology for the smart metering devices
[70], since it provides a standardized platform for exchanging
data between smart metering devices and appliances located
on customer premises. The features supported by the SEP
include demand response, advanced metering support, real-
time pricing, text messaging, and load control [264].
WirelessHART utilizes a time synchronized, self-
organizing, and self-healing mesh architecture, and supports
operation in the 2.4 GHz band using IEEE 802.15.4 standard
radios. Developed as a multi-vendor, interoperable wireless
standard, WirelessHART was defined for the requirements
of process field device networks. ISA100.11a is an open
wireless networking technology standard developed by the
International Society of Automation. The official description
is “Wireless Systems for Industrial Automation: Process
Control and Related Applications.” For wireless sensor
network applications in SG, such as a substation or a
generation plant, it is recommended to use WirelessHART or
ISA100.11a. These two standards are similar in functionality,
and therefore either standard is suitable for deployment [3].
Satellite Communications: Satellite communication is a
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 17
good solution for remote control and monitoring, since it
provides global coverage and rapid installation [90]. In some
scenarios where no communication infrastructure exists, es-
pecially for remote substations and generation deployments,
satellite communication is a cost-effective solution. For exam-
ple, Deep et al. [57] pointed out that with remote generation
deployments, such as those based on wind energy, a cost-
effective communication system with global coverage using
satellite technology would be advantageous. Such communi-
cation can be easily set up by only acquiring the necessary
satellite communication equipment. Note that some utilities
have already installed such equipment for rural substations
monitoring [242]. Furthermore, a terrestrial-only architecture
is vulnerable to disasters or communication system failures on
the ground. Therefore, in order to ensure the fail-safe operation
and the delivery of critical data traffic in disasters or terrestrial
communication system failures, satellites can be used as a
backup for the existing grid communication networks.
However, we should also note the disadvantages of satellite
communications. There are two major shortcomings. First,
a satellite communication system has a substantially higher
delay than that of a terrestrial communication system. This
makes some protocols (e.g. TCP), which are originally de-
signed for terrestrial communication, unsuitable for satellite
communications [90]. Second, satellite channel characteristics
vary depending on the effect of fading and the weather
conditions. This property can heavily degrade the performance
of the whole satellite communication system [104].
Microwave or Free-Space Optical Communications:
Microwave technologies are widely used for point-to-point
communications, since their small wavelength allows use
of conveniently sized directional antennas to obtain secure
information transmission at high bandwidths. The report by
Donegan [60] pointed out that over 50% of the world’s mobile
base stations are connected using point-to-point microwave
technologies. For over 20 years, microwave has been the
primary solution for rapidly rolling out cost-effective mobile
backhaul infrastructure worldwide [148].
Free-space optical communication is an optical communi-
cation technology that can use light propagating in free space
to transmit point-to-point data. It provides high bit rates with
low bit error rates. Furthermore, it is very secure due to the
high directionality and narrowness of the beams. In addition
to providing long-distance point-to-point communication in
remote or rural areas, these “optical wireless” technologies
also provide point-to-point solutions suitable for use in dense
urban areas where microwave solutions are impractical from
an interference standpoint [27].
Therefore, one important application of microwave or free-
space optical point-to-point communications is to build up SG
communication backhaul networks. This is especially impor-
tant in rural or remote areas, where using other wireless or
wired technologies is costly or even impossible. However, both
microwave communication and free-space optical communica-
tion are line of sight communication technologies. Therefore,
their communication qualities are greatly affected by obstacles
(e.g. buildings and hills) and environmental constraints (e.g.
rain fade).
Remarks: It is clear that wireless communication technologies
are important for future SG, and that different technologies
might be applicable. In order to assess the suitability of
various wireless technologies for meeting the communication
requirements of SG applications, Souryal et al. [231]
presented a methodology to evaluate wireless technologies.
The proposed approach to modeling wireless communications
first identifies the various applications utilizing a specific link.
Second, it translates the requirements of these applications to
link traffic characteristics in the form of a link layer arrival
rate and average message size. Third, it uses coverage analysis
to determine the maximum range of the technology under an
outage constraint and for a given set of channel propagation
parameters. Finally, using the link traffic characteristics and
coverage area determined above, it employs a model to
measure link performance in terms of reliability, delay, and
throughput.
C. Wired Technologies
It is also believed that wired communication technologies
will be integrated into SG. We list the following important
wired communication technologies.
Fiber-optic Communications: There is a long history of the
use of fiber communications by large power companies to
connect their generation network with their network control
facilities. Furthermore, its electromagnetic and radio interfer-
ence immunity make fiber-optic communication ideal for high
voltage operating environment [163].
Due to its high bandwidth capacity and immunity character-
istics, it is believed that optical fibers will play an important
role for the information network backbones in future SG
[90]. Although it is well-known that the installment cost of
optical fibers may be expensive, fiber optic network is still
a cost-effective communication infrastructure for high speed
communication network backbones in future SG, since such
fibers are already widely deployed in today’s communication
network backbones, with a large amount of spare capacity
being unused.
Powerline Communications: Powerline communications
(PLC) is a technology for carrying data on a conductor also
used for electric power transmission. In the last decades,
utility companies around the world have been using PLC
for remote metering and load control applications [73]. The
debate on what is the actual role of PLC in future SG is still
open. Some advocate that PLC is a very good candidate for
some applications [78,208], while others express concerns on
PLC (e.g. the security issue due to the nature of powerlines)
[149,195]. Although the SG could use many different commu-
nications technologies, without a doubt, PLC is the only wired
technology that has deployment cost comparable to wireless
technologies since the lines are already there [78].
Technically, in PLC power electronics are used to ma-
nipulate high-voltage waveforms for signal and information
oriented applications [262]. For example, a thyristor or similar
device is used to create a waveform disturbance such as a
very small but detectable voltage sag. The existence of the
sag implies digital “1” and no sag implies digital “0.”
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 18
Although how to most effectively utilize PLC is still not
clear, Galli et al. [78] predicted that PLC may be more suitable
for the distribution grid. Traditional substations in the medium
voltage distribution grids are not equipped with communica-
tions capabilities. Thus using the existing powerline infrastruc-
ture represents an appealing alternative to the installation of
new communication links. This enables the information about
state and event to flow among substations within a grid. In low
voltage distribution grids (close to the homes), there also exist
a large number of applications of PLC. First, narrowband PLC
is well suited for smart metering infrastructure [208]. Second,
PLC enables the communications between electric vehicles
and power grid via powerline without introducing other wired
or wireless equipments. Third, broadband PLC can provide the
service of transferring data seamlessly from SG controllers to
home networks and vice versa.
The most urgent task for the research on PLC might
be a comprehensive theoretical understanding. Most of the
works are focused on the ultimate performance that can
be achieved over the powerline channel [90]. In order to
further utilize PLC, we need to have a better understanding
of the powerline channel, since it is a complicated and noisy
medium disturbed by noise, external emissions, and frequent
impedance alterations. Some researchers have investigated
channel modeling and analysis methods for PLC, such as
[16,54,76,77,131,165,273].
Barmada et al. [16] studied load-time variation in PLC
systems and analyzed asynchronous impulsive noise and re-
lated channel variations due to switch commutations. Corripio
et al. [54] analyzed the properties of indoor PLC channels
when they are used for broadband transmission. It is shown
that these channels exhibit a short-term variation. Galli [76]
reported for the first time some statistical properties of indoor
powerline channel that exhibit some interesting similarities
to the wireless channel. He also reported for the first time
that both channel gain and root-mean-square delay spread
of indoor channels are lognormally distributed, leptokurtic,
and negatively correlated, thus suggesting that channels which
introduce severe multipath fading are also characterized by
large attenuation. Konat´e et al. [131] studied both frequency
(100 kHz-30 MHz) and time-domain channel modeling in
inverter driven electrical drives. Meng et al. [165] presented
an approach to model the transfer function of electrical pow-
erlines for broadband PLCs. In this approach, the powerline
is approximated as a transmission line, and the two intrinsic
parameters–the characteristic impedance and the propagation
constants–are derived based on the lumped-element circuit
model. Zimmermann and Dostert et al. [273] analyzed and
tried to model impulsive noise in broadband PLC. Refer to
[73] for more work along the research on channel modeling
and analysis methods for PLC.
D. End-to-end Communication Management
One important issue in the communication subsystem is
the end-to-end communication management. More specifically,
in this heterogenous communication subsystem where various
communication technologies, network structures, and devices
may be used, we need to identify each entity (probably by
giving a unique ID for each one), and solve the problem of
how to manage end-to-end communications (perhaps between
any pair of entities).
Recently, there is a growing trend towards the use of
TCP/IP technology (usually based on IPv6 address) as a
common and consistent approach in order to achieve end-to-
end communications [18,46,129,153,154,177,223]. TCP/IP
is an easy solution to the problem of managing systems based
on incompatible lower layer technologies. Therefore electric
utilities can deploy multiple communication systems, while
using TCP/IP technology as a common management platform
for maintaining end-to-end communications, and enjoy the
high rate of innovation (and competition) focused around the
TCP/IP protocol suite.
NIST also indicated that there are a number of benefits that
make TCP/IP an important SG technology, including the ma-
turity of a large number of standards, the availability of tools
and applications that can be applied to SG environments, and
its widespread use in both private and public networks [177].
If an application does not support TCP/IP natively, it may
still be possible to implement encapsulation, gateway, or semi-
transparent tunneling by providing a special communication
interface for this application.
E. Summary and Future Research
In this section, we reviewed the work on the smart com-
munication subsystem, including wireless technologies, wired
technologies, and end-to-end communication management. We
list the following challenges and possible directions worth
exploring.
1) Interoperability of communication technologies: Since
many different communication protocols and technologies will
be used in SG, and each of them probably will use its own
protocols and algorithms, materializing interoperability is not
easy.
Although the framework architecture in the classic layer
model (e.g. the famous Open Systems Interconnection model)
could provide a promising conceptual solution to this problem,
it is well-known that this model suffers in some modern
applications. For example, the performance of the pure TCP
may be very bad in wireless networks since it cannot dif-
ferentiate packet loss due to wireless fading from that due
to a real congestion in the network. In order to improve
the quality of service, under various operational conditions,
some functions or services are not tied to a given layer, but
can affect more than one layer. This often requires cross-
layer design and optimization, which may be essential in the
SG. However, interoperability among different communication
technologies, a precursor to cross-layer approaches, is difficult.
We suggest studying the advantages and disadvantages of
cross-layer design in SG communication subsystem consid-
ering interoperability, especially the trade-offs between cross-
layer optimization and the need for interoperability.
2) Dynamics of the communication subsystem: The com-
munication subsystem underlying an SG may be dynamic, with
topology changes being unpredictable. For example, both the
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 19
operation of connecting (or disconnecting) the electric vehicle
(EV) to (or from) the grid and the motion of vehicle may result
in the change of communication network topology. Note that
these operations may not follow a predictable pattern. The
dynamics of an SG communication subsystem have not been
fully explored.
The following two research directions are worth exploring.
First, systematic protocol designs are needed to support topol-
ogy dynamics. For example, communication protocols should
deal with topology reconfiguration in the connect-disconnect
operation of EVs. Second, dynamic resource allocation algo-
rithms are needed to support topology dynamics. For example,
due to the topology change, the network resources (e.g. band-
width) may need to be reallocated to optimize performance.
3) Smoothly updating existing protocols: The current
power grid has used several protocols to realize simple data
communications. For example, the currently used metering and
Supervisory Control and Data Acquisition Systems (SCADA)
[176] protocols are based on a simple request/response scheme
with their own addressing [223]. One problem is how to
smoothly update existing protocols to the ones which are
applicable in future SG. For example, as stated in Section
VI-D, for the end-to-end communication management, even
if an application does not support TCP/IP natively, it may
still be possible to implement the encapsulation, gateway, or
semi-transparent tunneling for this application. However, en-
capsulating such SCADA and metering protocols into mature
TCP/IP protocols generates an overhead without additional
benefit, and thus is deliberately not considered [223]. How
to smoothly transit the systems using these protocols to the
ones applicable in future SG is still an open question.
For this open question, we may be able to borrow from
industry the idea of how to smoothly update old IPv4 networks
to new IPv6 networks. For example, this transition can be
done in two stages: 1) the system can communicate using
pre-existing old protocols and also TCP/IP and 2) the system
operates by only using TCP/IP. A complete solution along this
line is desired.
VII. SMART MANAGEMENT SYSTEM
In SG two-way flows of electricity and information are
supported, which lay the foundation for realizing various
functions and management objectives, such as energy effi-
ciency improvement, operation cost reduction, demand and
supply balance, emission control, and utility maximization. A
common superficial understanding about SG is that only the
energy, information and communication infrastructure under-
lying the SG is smart. This is not true. The more accurate
assessment is: With the development of new management
applications and services that can leverage the technology and
capability upgrades enabled by this advanced infrastructure the
grid will keep becoming “smarter.”
For example, let us consider demand response, one of the
most important concepts supported by SG. Traditionally, the
electric utilities try to match the supply to the demand for
energy. However, this may be not only expensive but also
impractical, perhaps impossible in the longer run. This is
Fig. 13: Total California Load Profile for a Hot Day in 1999
[214].
because the total amount of power demand by the users can
have a very widespread probability distribution, which requires
spare generating plants in standby mode to respond to the
rapidly changing power usage. The last 10% of generating
capacity may be required in as little as 1% of the time. The
attempts to meet the demand could fail, resulting in brownouts
(i.e. a drop in voltage), blackouts (i.e. electrical power outage),
and even cascading failures. In SG, demand response manages
the customer consumption of electricity in response to supply
conditions. More specifically, by using demand response, SG
does not need to match the supply to the demand, but in
contrast, to match the demand to the available supply by
using control technology or convincing the consumers (such
as through variable pricing) thus achieving better capacity
utilization.
For example, Fig.13 shows the total energy usage pattern
during a twenty-four hour period for a typical hot day in
California (1999) [214]. As we can see, the total demand from
14:00PM to 18:00PM is much higher than the average. In an
SG, smart management by a smart meter can reduce energy
consumptions by turning off non-essential devices during peak
time in a way that the peak total demand can be reduced.
In this section, we explore smart management in SG. We
first classify smart management techniques according to their
management objectives and then according to their manage-
ment methods and tools.
A. Management Objectives
Within the framework of SG, many management goals,
which are difficult and possibly infeasible to realize in con-
ventional power grids, become possible and easy. So far, the
works for smart management mainly focus on the following
three objectives:
1) Energy efficiency and demand profile improvement;
2) Utility and cost optimization, and price stabilization;
3) Emission control.
Fig. 14 shows a detailed classification of the research work
related to these management objectives.
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 20
Management Objectives
Energy Efficiency and
Demand Profile
Demand Profile
Shaping
[14,37,40]
[83,108,130]
[170,171]
[188,236]
Energy Loss
Minimization
[10,12]
[185,228]
Utility, Cost, and Price
Utility and Cost
Individual
User
[32,40]
[53,69]
[95,108]
[120,171]
[170,188]
Multiple
Users
[98,220]
Electricity
Industry
[37,74]
[83,89]
[120,150]
[171,178]
[206,218]
Price
[213]
Emission
[14,32]
[86,151]
[218]
Fig. 14: Classification of Management Objectives
The research on energy efficiency and demand profile
mainly focuses on two topics. The first one can be categorized
as demand profile shaping. It can help match the demand to
the available supply.The usual way to shape demand profile is
shifting, scheduling, or reducing demand in order to reshape
a demand profile full of peaks to a nicely smoothed demand
profile, or reduce the peak-to-average ratio or peak demand of
the total energy demand [14,37,40,83,108,130,170,171,
188,236]. As discussed before, since electrical generation and
transmission systems are generally sized to correspond to peak
demand, lowering peak demand and smoothing demand profile
reduces overall plant and capital cost requirements, and also
increases the system reliability. Next we briefly describe these
research works.
Bakker et al. [14] designed a three step control and opti-
mization strategy and focused on the control algorithms used
to reshape the energy demand profile of a large group of build-
ings and their requirements. Caron and Kesidis [37] proposed
a dynamic pricing scheme incentivizing consumers to achieve
an aggregate load profile suitable for utilities, and studied
how close they can get to an ideal flat profile. Chen et al.
[40] studied two abstract market models for designing demand
response to match power supply and shape power demand.
They characterized the resulting equilibria in competitive as
well as oligopolistic markets, and proposed distributed demand
response algorithms to achieve the equilibria. Ibars et al.
[108] aimed to smooth the electric demand curve and avoid
overloading both the generation and distribution capacity of
the grid, by using a network congestion game, where each user
allocates demand as a response to other users’ actions. Kishore
and Snyder [130] first presented an optimization model for
determining the timing of appliance operation to take advan-
tage of lower electricity rates during off-peak periods. They
then proposed a distributed scheduling mechanism to reduce
peak demand within a neighborhood of homes. They finally
introduced a more powerful energy management optimization
model, based on dynamic programming, which accounts for
electricity capacity constraints. Mohsenian-Rad and Leon-
Garcia [170] proposed an optimal and automatic residential
energy consumption scheduling framework, which attempts to
achieve a desired trade-off between minimizing the electricity
payment and minimizing the waiting time for the operation
of each appliance. Mohsenian-Rad et al. [171] discovered
that by adopting pricing tariffs which differentiate the energy
usage in time and level, the global optimal performance is
achieved at a Nash equilibrium of the formulated energy
consumption scheduling game. O’Neill et al. [188] proposed
an online learning algorithm to reduce residential energy costs
and smooth energy usage. Taneja et al. [236] introduced a
generalized measure of dispatchability of energy (called slack),
identified two classes of dispatchable energy loads, and created
models for these loads to match their consumption to the
generation of energy sources. Ghosh et al. [83] designed an
optimal incentive mechanism offered to energy customers. Ac-
cording to their mechanism, customers who are more willing
to reduce their aggregate demand over the entire horizon,
which consists of multiple periods, rather than simply shifting
their load to off-peak periods, tend to receive higher incentives,
and vice versa.
The second topic of energy efficiency and demand profile
is minimizing energy loss. However, using distributed energy
generation in SG makes this problem more complicated. In
order to minimize the system energy loss, Ochoa and Harrison
[185] proposed to determine the optimal accommodation of
distributed renewable energy generation for minimizing energy
loss by using an optimal multi-period alternating current power
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 21
flow. Atwa et al. [12] aimed to minimize energy loss through
the optimal mix of statistically-modeled renewable sources.
Aquino-Lugo and Overbye [10] presented a decentralized
optimization algorithm to minimize power losses in the distri-
bution grids.
We have reviewed the work on energy efficiency and
demand profile improvement in the above. Improving utility,
increasing profit, and reducing cost are also important manage-
ment objectives. Researchers realize these objectives in various
levels and from various perspectives, such as individual user
cost/bill or profit [32,40,53,69,95,120,170,171,188],
single energy bill or aggregate utility of a group of users
[98,220], cost or utility of electricity industry and system
[37,74,83,89,120,150,171,178,206,218]. Stabilization
of prices is also a research topic in SG, since relaying the real-
time wholesale market prices to the end consumers creates a
closed loop feedback system which could be unstable or lack
robustness, leading to price volatility. Roozbehani et al. [213]
therefore developed a mathematical model for characterization
of the dynamic evolution of supply, demand, and market
clearing (locational marginal) prices under real-time pricing,
and presented a stabilizing pricing algorithm.
Emission control, another important management objective
in the electric power industry, has a significant influence on
environment protection. However, note that minimizing gener-
ation cost or maximizing utility/profit is not directly equivalent
to minimizing emission by utilizing renewable energy as much
as possible. This is because, generally speaking, the cost of
power generation from renewable energy source is not always
the lowest. Therefore, as suggested by Gormus et al. [86],
environmental impact of energy produced from fossil fuels
should be factored into the demand scheduling algorithm as
a cost parameter which may result in more peak loads to be
moved to the periods where renewable sources have a higher
percentage in the generation mix. However, individual users
should be willing to accept their appliances to be scheduled
according to the requirements of low carbon scheduling. In
addition, many researchers have also investigated how to
optimize emission reduction. Saber and Venayagamoorthy
[218] studied how to take advantage of both plug-in hybrid
electric vehicles and renewable resources to reduce emission.
Bakker et al. [14] presented a three step control strategy
to optimize the overall energy efficiency and increase gen-
eration from renewable resources with the ultimate goal to
reduce the C O2emission caused by electricity generation. Bu
et al. [32] modeled the stochastic power demand loads as a
Markov-modulated Poisson process, and formulated the unit
commitment scheduling problem of power generation systems
as a partially observable Markov decision process multi-armed
bandit problem. By adjusting the pollutant emission costs, the
CO2emissions can be reduced. Liu and Xu [151] performed
a mathematical analysis for the effects of wind power on
emission control, and developed a load dispatch model to
minimize the emission.
Considering the importance of microgrids and G2V/V2G,
we particularly list the work related to them.
Microgrids: Guan et al. [89] applied microgrid technology
to minimize the overall cost of electricity and natural gas for
a building operation while satisfying the energy balance and
complicated operating constraints of individual energy sup-
ply equipment and devices. The results showed that through
integrated scheduling and control of various energy supply
sources of the building, significant energy cost saving can be
achieved. Vandoorn et al. [249] presented a method for active
load control in islanded microgrids, which is triggered by the
microgrid voltage level. This is enabled by using the voltage-
droop control strategy and its specific properties. It is shown
that the presented demand dispatch strategy leads to reduction
of line losses and that with the combination of the active power
control and the presented active load control, the renewable
energy can be exploited optimally. A review of challenges to
power management in microgrids can be found in [52].
G2V/V2G: As stated in Section IV-D, in G2V, high pene-
tration levels of uncoordinated electric vehicle (EV) charging
will significantly reduce power system performance and effi-
ciency, and even cause overloading. Coordinated charging has
hence been proposed to mitigate these negative impacts using
stochastic programming [48,49], quadratic optimization [93],
particle swarm optimization [218], and dynamic programming
[95]. Recall that particle swarm optimization can solve com-
plex constrained optimization problems quickly,with accuracy
and without any dimensional limitation and physical computer
memory limit. In the context of coordinated EV charging,
Sortomme et al. [228] studied the relationship between feeder
losses, load factor, and load variance. They showed the benefits
of reduced computation time and problem complexity, when
we use load variance or load factor as the objective function
rather than system losses. Another interesting work is done by
Pan et al. in [192]. Instead of attempting to find an optimized
power dispatching approach, they explored how to aptly place
EV infrastructures like battery exchange stations so that they
can support both the transportation system and the power grid.
In V2G, we often discharge the battery to deliver the power
to the grid and then recharge it later when the price is low.
Therefore, a key question is how to determine the appropriate
charge and discharge times throughout the day, taking into
account the requirements of both vehicle owners and utility.
Hutson et al. [107] studied this problem and used binary
particle swarm optimization to look for optimal solutions that
maximize profits to vehicle owners while satisfying system and
vehicle owners’ constraints. In addition, Lund and Kempton
[159] analyzed the positive influence of V2G on integration
of renewable energy into the electricity sectors. Since today’s
abundant renewable energy resources have fluctuating output,
to increase the fraction of electricity from them, we must learn
to maintain a balance between demand and supply. V2G tech-
nology can provide storage, matching the time of generation
to time of load. They found that adding V2G to these national
energy systems allows integration of much higher levels of
wind electricity without excess electric production, and also
greatly reduces national C O2emissions.
B. Management Methods and Tools
In order to solve management objectives, researchers have
adopted various methods and tools. Thus far, researchers
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 22
Management Methods and Tools
Optimization Machine Learning Game Theory Auction
Convex
Programming
[120,170]
[220,228]
Dynamic
Programming
[8,95,130,172]
Stochastic
Programming
[48,49,150]
[178,192,194]
Robust
Programming
[53]
Particle Swarm
Optimization
[107,168,217,218]
Machine Learning
[69,188]
Game Theory
[108,171]Auction
[94,169]
Fig. 15: Classification of Management Methods and Tools
mainly use optimization, machine learning, game theory, and
auction as tools to solve various management problems.
Fig. 15 shows a classification of the related work.
For optimization approaches, the commonly used mathe-
matical tools are convex programming [120,170,220,228]
and dynamic programming [8,95,130,172]. Since the
renewable energy supply is often a time-varying process,
other optimization techniques such as stochastic programming
[48,49,150,178,192,194] and robust programming [53] are
also widely used. In addition, since the particle swarm opti-
mization can solve complex constrained optimization problems
quickly, with accuracy and without any dimensional limitation
and physical computer memory limit, it is also a widely used
optimization tool [107, 168, 217, 218].
Machine learning focuses on the design and development
of algorithms that allow control systems to evolve behaviors
based on empirical data, such as from sensor or phasor
measurement unit (PMU) data. O’Neill et al. [188] used online
learning application to implicitly estimate the impact of future
energy prices and consumer decisions on long term costs,
and thus schedule residential device usage. Fang et al. [69]
used online machine learning to analyze the renewable energy
resource use strategy in islanded microgrids. More specifically,
a customer tries to decide among multiple renewable energy
sources which should be used to maximize profit. Although the
power pattern of the renewable energy source is not known in
advance, they proved that when the time horizon is sufficiently
large, on average the upper bound on the gap between the
expected profit obtained at each time slot by using the optimal
renewable energy source and that by following their strategies
is arbitrarily small. Considering that a large number of smart
meters, sensors, and PMUs will be deployed, we believe that
machine learning will play an important role in analysis and
processing of user data and grid states.
Game theory is also a strong analysis tool for SG man-
agement. One reason is that we cannot always expect and
require all the users to be cooperative. Game theory can
help us design effective schemes to cope with this case. For
example, Ibars et al. [108] proposed a distributed solution
based on a network congestion game to guarantee that the
optimal local solution of each selfish consumer is also the
solution of a global objective. In [171], by adopting pricing
tariffs that differentiate the energy usage in time and usage
level, the global optimal performance is achieved at the Nash
equilibrium of the formulated energy consumption scheduling
game. Another reason why game theory is desirable is that the
emerging SG will lead to the emergence of a large number
of markets, which will be akin to multi-player games. For
example, the consumers within a microgrid can create a market
for trading energy. Game theory can help us analyze the
resulting market. For example, Chen et al. [40] characterized
the resulting equilibria in competitive as well as oligopolistic
markets, and proposed distributed demand response algorithms
to achieve the equilibria.
Auction may also be popular in the emerging SG. Dis-
tributed energy generation and microgrid will be widely used
in SG. As mentioned above, the consumers within a microgrid
can create a market for trading energy. Bidding and auction
can be used for energy sale within a local microgrid market.
Auction is not limited to the microgrid market. The authors of
[94,169] proposed demand reduction bid, a kind of demand
response programs. Recall that demand response can be used
to reduce the system peak load. At the peak load period,
customers send demand reduction bids to the utility with the
available demand reduction capacity and the price asked for.
This program encourages customers to provide load reductions
at prices for which they are willing to curtail.
C. Summary and Future Research
Within the framework of SG, many management goals,
which are difficult and even infeasible to realize in conven-
tional power grids, become easy and possible.
As reviewed in this survey, we found that most of the works
on the smart management aim to improve energy efficiency,
shape demand profile, increase utility, and reduce costs and
emissions based on the advanced SG smart infrastructure.
We believe that this advanced infrastructure will lead to an
explosion of functionalities, and that more and more new
management services and applications will emerge. We present
two possible services in the following.
1) The integration of pervasive computing and Smart
Grid: Pervasive computing is a post-desktop model in which
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 23
information processing has been thoroughly integrated into
everyday objects and activities. Using the information pro-
vided by pervasive computing, SG is able to serve users
more effectively and smartly, and eventually revolutionize
the consumers’ lives. Let us consider a simple example. In
summer, the average high temperature in Phoenix (Arizona,
USA) area can be over 100◦F(37.8◦C) and has spiked over
120◦F(48.9◦C). People hope that when they get home, the
temperature at home is within 60 −80◦F. Therefore, the smart
meter connected to the air conditioner can periodically inquire
the position of the house owner by sending the inquiry to the
owner’s smart phonewhich can obtain the owner’s position via
GPS. If the smart meter finds the owner coming back home,
it will decide to turn on the air conditioner in advance so
that when the owner gets home, the temperature is within the
comfortable zone.
2) Smart Grid Store: Like “Apple’s Application Store”[9],
many management applications and services are available
online. Users can choose their expected services and download
them to the local system (e.g. the smart meter). For example, a
user, who needs a management program supporting the smart
control of air conditioner mentioned above, can buy such a
program from the Smart Grid Store online and download it.
This Smart Grid Store provides an integrated platform, which
can drive the third party to develop new management programs
and meantime help users easily customize their management
services.
Although a smart management system in SG is promising
and encouraging, we still face many challenges. We summarize
the important challenges and the possible research directions
worth exploring in the following.
1) Regulating emerging markets: As mentioned before,
microgrids could lead to the emergence of new markets among
users within a microgrid for trading energy. Due to the lack
of supervision of conventional utilities, it is a challenge to
regulate such new markets.
For example, in an auction market where the consumers
within a microgrid trade energy, how to guarantee truthful
auction is a challenge, since some users could make untruthful
bids to cheat the seller in order to obtain the benefit which
they cannot get in truthful bidding. Note that truthful auction
has been studied for several years. One of the most well-
known auction schemes is the Vickrey-Clarke-Groves (VCG)
scheme [47,88,253], which is a type of sealed-bid auction
where multiple items are up for bid, and each bidder submits
a different value for each item. This system charges each
individual the harm they cause to other bidders, and ensures
that the optimal strategy for a bidder is to bid the true
valuations of the objects. The truthful auction problem in SG
may be solved along this line.
2) Effectiveness of the distributed management system:
The SG is expected to utilize distributed management sys-
tems more often, since distributed generation and plug-and-
play components will be widely used to form autonomous
and distributed subgrids (e.g. microgrids). However, generally
speaking, more amount of timely information means smarter
management decisions in the management process. There-
fore, considering the limitation on accessible information,
distributed management systems usually cannot compute a
globally optimal decision. We thus need to consider how
to determine the optimal size of a subgrid controlled by a
distributed management system and how to obtain necessary
timely information, so that the local decision is good enough.
3) Impact of ad-hoc organization of SG: In an SG many
parts (such as solar panels) are deployed in a distributed
manner, and are even required to work as plug-and-play
components. More flexibility from the users’ perspective also
leads to more difficulty in system design and management.
This actually opens up many possible research topics.
For example, we need to study the self-configuration of the
power grid, and further how to manage the power dispatching
in such a self-configured system. Let us consider a more spe-
cific example. In a power system containing many functioning
microgrids, when some components stop functioning in one
microgrid, it could be interesting to study how to optimally
allocate some resources (e.g. distributed energy generators)
from other microgrids to this one to improve the overall
performance of the whole power system.
4) Impact of utilization of fluctuant and intermittent
renewables: The utilization of the renewable resources, such
as wind and solar, also makes management more difficult due
to their fluctuant and intermittent nature. The management
system should maintain reliability and satisfy operational
requirements, meanwhile taking into account the uncertainty
and variability of energy sources.
In order to solve the optimization problems related to
such renewables, we suggest that stochastic programming and
robust programming [53] play more important roles as math-
ematical tools, since they can model optimization problems
that involve uncertainty. As discussed in Section IV-E, Markov
process and online machine learning technology may also be
applicable for the analysis of renewable source performance.
Based on the prediction of renewable source performance, we
can try to achieve our management objectives.
VIII. SMART PROTECTION SYSTEM
The smart protection system in SG must address not only
inadvertent compromises of the grid infrastructure due to
user errors, equipment failures, and natural disasters, but also
deliberate cyber attacks, such as from disgruntled employees,
industrial spies, and terrorists.
In this section, we explore the work targeted the smart
protection system in SG. We first review the work related to
system reliability analysis and failure protection mechanisms,
and then the security and privacy issues in SG.
A. System Reliability and Failure Protection
Reliability is the ability of a component or system to
perform required functions under stated conditions for a stated
period of time. System reliability is an important topic in
power grid research and design. In the U.S., the annual cost
of outages in 2002 was estimated to be in the order of $79B,
while the total electricity retail revenue was $249B [173].
Furthermore, cascading blackouts could happen. For example,
in the infamous 2003 East Coast blackout, 50 million people in
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 24
the U.S. and Canada lost power for up to several days [182].
An initial review of methods for cascading failure analysis
in power grid systems can be found in [15]. The future SG
is expected to provide more reliable system operation and
smarter failure protection mechanism.
1) System Reliability
It is expected that distributed generation (DG) will be widely
used in SG. While using some fluctuant and intermittent
renewables may compromise the stability of the grid [61,132],
the authors of [43,173] stated that innovative architectures and
designs can offer great promise to connect DGs into the grid
without sacrificing reliability.
Chen et al. [43] proposed to take advantage of new ar-
chitectures such as microgrid to simplify the impact of DG
on the grid. Intuitively, as loads are being served locally
within a microgrid, less power flows within the entire grid
infrastructure. Thus, the reliability and stability of the SG can
be enhanced. They found a very encouraging result that local
power generation, even when only introducing a small number
of local generators into the grid, can reduce the likelihood
of cascading failures dramatically. Moslehi and Kumar [173]
observed that an ideal mix of the SG resources (e.g. distributed
renewable sources, demand response, and storage) leads to a
flatter net demand that eventually further improves reliability.
However, realizing this requires a systematic approach –
developing a common vision for cohesive gridwide integration
of necessary information technologies. Thus, they proposed an
architectural framework to serve as a concrete representation
of a common vision.
Furthermore, the reliability and stability of an SG also
depends on the reliability of the measurement system which is
used to monitor the reliability and stability of the SG. Recently,
the wide-area measurement system (WAMS) based on phasor
measurement units (PMUs) is becoming an important com-
ponent for the monitoring, control, and protection functions
in SG. In order to analyze the reliability of WAMS, Wang
et al. [255] presented a quantified reliability evaluation method
by combining Markov modeling and state enumeration tech-
niques. This method can be used for evaluating the reliability
of the backbone communication network in WAMS and the
overall WAMS from a hardware reliability viewpoint.
Another research topic is using simulation for system relia-
bility analysis. The more accurately a simulation platform can
emulate the behavior and performance of an SG architecture,
the better we will understand its advantages and potential
shortcomings. However, the question is how to build up a
simulation system which is accurate, flexible, adaptable, and
scalable enough. Bou Ghosn et al. [25] utilized an incremental
method, beginning with simulating a local microgrid, but with
a scalable design that can grow hierarchically into a more
complete model. Such a simulator can help us understand
SG issues and identify ways to improve the electrical grid.
Godfrey et al. [85] proposed to model both the communication
network and the power system in SG using simulation. This
model provides means to examine the effect of communication
failures as a function of the radio transmission power level.
2) Failure Protection Mechanism
In this subsection, we first review two topics related to failure
protection mechanism. First, failure prediction and prevention
play important roles in the smart protection system since they
attempt to prevent failures from happening. Second, once the
system does fail, failure identification, diagnosis, and recovery
are required to make the system recover from the failure
and work normally as soon as possible. Fig. 16 shows a
classification of the work related to these two topics. In
addition, since the microgrid is regarded as one of the most
important new components in the SG vision, we review the
work focused on the microgrid protection.
Failure Prediction and Prevention: For an SG, one effective
approach to preventing failures from happening is predicting
the weak points or the region of stability existence in its energy
subsystem. Chertkov et al. [44] developed an approach to
efficiently identify the most probable failure modes in static
load distribution for a given power network. They found that if
the normal operational mode of the grid is sufficiently healthy,
the failure modes are sufficiently sparse, i.e., the failures
are caused by load fluctuations at only a few buses. Their
technique can help discover weak links which are saturated at
the failure modes, and can also identify generators working
at the capacity and those under the capacity, thus providing
predictive capability for improving the reliability of any power
system. Vaiman et al. [248] proposed to utilize PMU data
to compute the region of stability existence and operational
margins. An automated process continuously monitors voltage
constraints, thermal limits, and steady-state stability simulta-
neously. This approach can be used to improve the reliability
of the transmission grid and to prevent major blackouts.
Failure Identification, Diagnosis, and Recovery: Once a
failure occurs, the first step must be quickly locating and
identifying the failure to avoid cascading events.
Due to the wide deployment of PMUs in SG, the authors
of [237,238,272] proposed to take advantage of the phasor
information for line outage detection and network parameter
error identification. Tate and Overbye [237] developed an
algorithm which uses known system topology information,
together with PMU phasor angle measurements, to detect
system line outages. In addition to determining the outaged
line, their algorithm also provides an estimate of the pre-
outage flow on the outaged line. Later, Tate and Overbye
[238] studied how double line outages can be detected using
a combination of pre-outage topology information and real-
time phase angle measurements that are obtained from PMUs.
Zhu and Abur [272] showed that the identification of certain
parameter errors based on conventional measurements, no
matter how redundant they are, may not always be possible.
They hence described the need for phasor measurements to
overcome this limitation.
Other works on failure identification and diagnosis include
[34,35,101,216]. Sometimes we need to select proper
features to identify the root cause. Cai et al. [34] reviewed
two popular feature selection methods: 1) hypothesis test,
2) stepwise regression, and introduced another two: 3) step-
wise selection by Akaike’s Information Criterion, and 4)
LASSO/ALASSO. Those algorithms are used to help engi-
neers to find out the information that may be buried under
the massive data. Considering that many events in power
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 25
Failure Protection Mechanism
Failure Prediction and Prevention
[44,248]Failure Identification, Diagnosis, and Recovery
Failure Identification
and Locating
[34,35,101,216]
[237,238,272]
Grid Self-healing
[145,205]Smart Meter
Data Recovery
[39]
Methodology of
Failure Processing
[189]
Fig. 16: Classification of Failure Protection Mechanisms
systems are stochastic, He and Zhang [101] proposed to
use probabilistic graphical models for modeling the spatially
correlated data from PMUs, and to use statistical hypothesis
testing for the task of fault diagnosis. Calderaro et al. [35]
carefully designed a Petri Net to capture the modeling details
of the protection system of the distribution grid, and presented
a method based on this Petri Net to detect and identify the
failures in data transmission and faults in the distribution grid.
Note that the Petri net is one of several mathematical modeling
languages for the description of distributed systems. Russell
and Benner [216] presented some examples of the types of
incipient failures that can be detected from substation electrical
waveforms, and described the significant data and analysis
requirements to enable detection and classification.
The ability to “self-heal” in the event of failure is expected
to be an important characteristic of SG according to the
standards [177] from the National Institute of Standards and
Technology (NIST). An effective approach is to divide the
power grid into small, autonomous islands (e.g. microgrids)
which can work well during normal operations and also
continue working during outages [205]. By appropriately con-
trolling the system reconfiguration, the impact of disturbances
or failures can be restricted within the islands or can be
isolated. Cascading events and further system failures can
hence be avoided. Therefore the overall efficiency of system
restoration can be improved [145].
Failures could also occur on smart meters so that load
data could contain corrupted or missing data. Processing or
even recovering such data is important since it contains vital
information for day-to-day operations and system analysis.
Chen et al. [39] hence presented B-Spline smoothing and
Kernel smoothing based techniques to cope with this issue
and automatically cleanse corrupted and missing data.
In addition, for the methodology of making decision on how
to process a failure, Overman and Sackman [189] suggested
that the decision-making ability should be distributed to the
substation and/or field devices; or at the minimum, to preload
these distributed devices with sufficient information such that
they can take corresponding automatic actions in the event
of a system failure without having to wait for instructions
from the central controller. They found that when coupled
with a distributed rather than hierarchical communications
architecture, preloading substation and field devices with a
set of next-actions-to-be-taken instructions can significantly
increase grid reliability while simultaneously reducing real-
time impact from loss of reliable control.
Microgrid Protection: Protection of microgrids during nor-
mal or island operations is also an important research topic
since microgrids will be widely used in SG. Note that pro-
tection of a microgrid is strongly related to its control and
operation issues [133]. For example, traditional protection for
distribution grids is designed for high fault-current levels in
radial networks. However, during an islanding operation of
the microgrid, high fault-currents from the utility grid are
not present. Moreover, most of the DG units which will
be connected to the low voltage microgrid will be con-
verted/interfaced with limited fault-current feeding capabili-
ties. This means that the traditional fuse protection of low
voltage network is no longer applicable for microgrid, and that
new protection methods must be developed. Feero et al. [72]
also examined several protection problems that must be dealt
with to successfully operate a microgrid when the utility is
experiencing abnormal conditions, and pointed out that there
are two distinct sets of problems to solve. The first is how
to determine when an islanded microgrid should be formed in
the face of abnormal conditions that the utility can experience.
The second is how to provide segments of the microgrid with
sufficient coordinated fault protection while operating as an
island separated from the utility.
These new issues drive the research on new protection
methods. Various methods for microgrid protection have been
proposed in [4,31,62,133,179,219,229,229,246]. Al-
Nasseri and Redfern [4] described a relay that uses distur-
bances in the three phase voltages to provide reliable and fast
detection of different types of faults within the microgrid.
Brucoli and Green [31] investigated the fault behavior of
inverter-supplied microgrids. They showed that the response
of the system in the event of a fault is strongly dependent
FANG et al.: SMART GRID – THE NEW AND IMPROVED POWER GRID: A SURVEY 26
Security and Privacy
Metering and Measurement
Smart Metering
Security
[7,19,50]
[162,164,250]
Privacy
[45,64,79,177]
[121,140,143,147]
[162,207]
Monitoring and
Measurement
[21,55,152]
[222,261,268]
Information Transmission
[59,158,241]
[128,142]
Fig. 17: Classification of the Work on Security and Privacy for SG
on the inverter control which actively limits the available
fault current. Therefore, the choice of an alternative protection
scheme for an islanded microgrid is strongly dependent on
the type of control implemented. Driesen et al. [62] discussed
protection issues concerning DG. Their simulations concern
the effect of local induction generators on protection selectivity
in a system with parallel distribution feeders. Nikkhajoei and
Lasseter [179] indicated that the philosophy for microgrid
protection is to have the same protection strategies for both
islanded and grid-connected operations, and that it is important
that the protection functions have plug-and-play functionality.
Laaksonen [133] presented the protection issues of low-voltage
microgrids and developed extensions to the low-voltage mi-
crogrid protection concept based on simulations with PSCAD
simulation software [201]. Tumilty et al. [246] aimed at de-
veloping practical protection, control, and automation schemes
for microgrids, and outlined several schemes covering both
urban and rural applications. Sortomme et al. [229] proposed
a protection scheme using digital relays with a communication
network for the protection of the microgrid system, explored
the increased reliability of adding an additional line to form a
loop structure, and demonstrated a novel method for modeling
high impedance faults to show how the protection scheme
can protect against them. Salomonsson et al. [219] proposed
a low-voltage DC microgrid protection system design. They
presented the operating principles and technical data of low-
voltage DC protection devices, both available and in the
research stage. They also discussed different fault-detection
and grounding methods.
B. Security and Privacy
Security is a never-ending game of wits, pitting attackers
versus asset owners. SG security is no exception to this
paradigm. Cyber security is regarded as one of the biggest
challenges in SG [17,166,177]. Vulnerabilities may allow an
attacker to penetrate a system, obtain user privacy, gain access
to control software, and alter load conditions to destabilize the
grid in unpredictable ways [166]. Fig. 17 shows a classification
of the work on security and privacy for SG. We must note
that the advanced infrastructure used in SG on one hand
empowers us to realize more powerful mechanisms to defend
against attacks and handle failures, but on the other hand
opens up many new vulnerabilities. Thus in the following,
we also discuss many new security and privacy issues due to
the deployment of smart meters, sensors, and PMUs, together
with some solutions.
1) Information Metering and Measurement
Security in Smart Metering: One of the security issues
comes from the newly deployed smart meters. Smart meters
are extremely attractive targets for malicious hackers, since
vulnerabilities can easily be monetized [162]. Hackers who
compromise a smart meter can immediately manipulate their
energy costs or fabricate generated energy meter readings
to make money. A common consumer fraud in traditional
power grid is that customers turn a traditional physical meter
upside down in the electrical socket to cause the internal u