Reliability data for smart grids: Where the real data can be found

Full text

SMART CITIES SYMPOSIUM PRAGUE 2018
Reliability Data for Smart Grids:
Where the Real Data can be Found
Stanislav Chren, Bruno Rossi, Member, IEEE, Barbora B¨
uhnova, Member, IEEE, and Tom´
aˇ
s Pitner
This is the accepted version of the paper Chren, S., Rossi, B., Buhnova, B., and Pitner, T. (2018). Reliability Data for Smart Grids: Where the Real Data
can be Found. In Smart Cities Symposium Prague (SCSP), 2018. IEEE
Abstract—Smart Grids play an important role in modern
society and for the sustainability of its wellbeing. However, the
undoubted advantages come at the cost of higher complex-
ity, especially at the level of information and communication
technologies that enhance the physical grid infrastructure. As
such, software quality requirements, such as reliability, resilience,
safety, security, privacy, and performance assume a more func-
tional facet. In this paper, we focus on software reliability
as one of the key qualities of a Smart Grid infrastructure,
which is however not yet well defined and understood. We
formulate relevant definitions of software reliability in the Smart
Grid context, categorize information necessary to quantify the
identified reliability views, and explore existing literature and
online resources to assess what datasets, necessary for reliability
quantification, are available to make the reliability assessment
possible.
Index Terms—Smart Grids, Power systems reliability, data
analysis.
I. INTRODUCTION
Information & Communication Technology (ICT)-enhanced
power grids, so called Smart Grids, are becoming the backbone
of modern society and also one of the key critical infrastruc-
tures. Although many quality perspectives are becoming the
center of attention during Smart Grid design and deployment
(including reliability, resilience, safety, security, privacy, and
performance [1], [2]), the concept of reliability still remains
bound mainly to the power distribution and the physical
infrastructure [3], [4], while software reliability is receiving
very little or no attention.
In this paper, we investigate the role of software reliability
in the Smart Grid domain by examining the Smart Grid
architecture, layer by layer, and identifying different views of
Smart Grid reliability that shall become the focus of software
reliability engineering in this domain. In this way, we also
provide insight into the specifics of Smart Grid reliability,
which is surprisingly strongly linked to the fulfillment of other
non-functional (rather than functional) requirements (such as
safety, security, performance or measurements accuracy) [5].
Such a study of reliability in the context of Smart Grid pro-
vides the basis for the main contribution of this paper, which
is the examination of types of datasets that carry the necessary
information for reliability engineering studying the identified
reliability directions, and the survey of the availability of such
datasets to reliability engineers, with the emphasis on real data.
All authors are with the Lab of Software Architectures and Information Sys-
tems (Lasaris), www.lasaris.cz, Faculty of Informatics, Masaryk University,
Brno, Czech Republic, e-mail:{chren,brossi,buhnova,tomp}@mail.muni.cz
In particular, we present the classification of information needs
for Smart Grids reliability, and identify information needs and
open datasets to support Smart Grids reliability analysis.
The paper is structured as follows. Section II provides an
overview of the Smart Grid Architecture. Section III discusses
different levels of reliability for the Smart Grid. Section IV
presents information needs for reliability analysis in the Smart
Grid and available datasets for the different needs. Section V
provides the discussion with the main findings, assumptions,
and limitations of the study. Section VI concludes the paper.
II. SM ART GRID ARCHITECTURE
According to the Smart Grid Reference Model (SGAM) [6],
the Smart Grid architecture consists of multiple layers, namely
the business layer defining the market policies and business
models, the functional layer describing provided services, and
the communication layers detailing the protocols and standards
for information exchange between components. The lowest
layer consists of a variety of hardware and software com-
ponents, which cover the whole electrical energy conversion
chain comprising five parts (bulk generation,transmission,
distribution,distributed electrical resources (DER),customer
premises).
Bulk generation represents generation of energy in large
quantities, for example by fossil, nuclear or hydro power plants
that are connected to the transmission system. Transmission
represents the infrastructure and organization for long-distance
energy transportation. Distribution represents the infrastructure
and organization that distributes the electricity to customers.
DER describes the distributed electrical resources connected
to the public distribution grid. Customer premises include the
consumers of electricity and also the local producers.
The services of the Smart Grid are enabled by several
systems (Fig. 1). The most significant system is the Advanced
Metering Infrastructure (AMI) System [8]. The major AMI
components are: smart meters, meter data concentrator and
data central server [9]. A smart meter is a replacement for a
traditional electric meter which ensures the collection of de-
tailed data about customer power consumption and production
and also supports remote control of specific appliances as part
of demand-response program. A demand-response program,
which governs the strategies of power demand control and
management, is used for dynamic billing of customers and
load optimization in a power grid during peak hours. Data
concentrators are responsible for collection of data from smart
meters and they can be also used for temporary storage,
analysis or encryption of the data, and they can offer early
978-1-5386-5017-2/18/$31.00 c
2018 IEEE

Fig. 1. Smart Grid data and information flow. Adapted from [7]
information about outages. A data central server (also called
Demand-Response server) manages the AMI operation by
storing and analysing the data and handling the demand-
response program.
However, there are other systems that are needed to ensure
Smart Grid reliability:
•Blackout Prevention System (WAMPAC) whose objec-
tive is to protect the grid from instabilities and failures.
It covers the whole power grid. It utilizes Phasor Mea-
surement Units (PMUs) to obtain relevant information
from the grid. PMU is a device which measures electrical
waves in a grid and helps to detect anomalies and failures.
•Supervisory Control and Data Acquisition System
(SCADA) is one of the core systems of a Smart Grid
that provides support to operation activities and functions
in transmission automation, dispatch centers and control
rooms. In a SCADA system, a remote terminal unit (RTU)
collects data from devices in a substation and delivers
the data in packets, on command, to a central Energy
Management System (EMS).
•Flexible Alternating Current Transmission System
(FACTS) is responsible for reliable and secure trans-
mission of power. It allows dynamic voltage control, in-
creased transmission capability and capacity, and supports
fast restore of the grid after failure.
•Feeder Automation System is responsible for operation
of medium-voltage (MW) networks including fault detec-
tion.
Smart Grids use variety of communication technologies
for information exchange between components (Fig. 2). The
used technologies need to take into account communication
requirements of various usage scenarios, such as maximum
allowed latency, payload size, or frequency of data transfer.
In the smaller scope of Home Area Networks (HAN),
each device is expected to require bandwidth from 10 to 100
Kbps, conventionally adopting WiFi, ZigBee, and HomePlug
technologies. In Building/Business Area Network (BAN), a
wired technology named BACnet is the prominent communica-
tion protocol. The communication between consumer premises
and aggregation points is currently handled by Power-Line-
Communication (PLC). The Advantages of PLCs are their
low cost and expansion, and penetration in utility providers
territory. Their disadvantages include low bandwidth (up to
20Kbps), and data distortion around transformers which re-
quires bypassing transformer points using other techniques.
PLC is especially useful in remote locations where the num-
ber of nodes (consumers) is relatively low and no wireless
(cellular, GPRS) coverage is available. For cases when higher
bandwidth is required (e.g. in urban areas for advanced
demand-response services), Mesh networks are considered. In
Mesh networks, each node is responsible for collecting its own
data, as well as relaying information from other nodes in the
network [8].
III. RELIABILITY IN SMA RT GRID
Reliability is generally considered one of the most important
quality attribute in Smart Grids [1]. Due to the potentially
serious impact of a Smart Grid failure on our lives, the
electrical power grid is being considered the most critical in-
frastructure of modern society. Critical infrastructures require
specific treatment to ensure their reliability, which includes
also the legislation point of view. The handling of critical
infrastructures is described by the European Programme for
Critical Infrastructure Protection (EPCIP) [11], in the Eu-
ropean context. In the US, there is an organization—The
North American Electric Reliability Corporation (NERC)—
that supervises the reliability of the power grid.
The understanding of the reliability concept might vary
depending on the Smart Grid layer, namely communication,
power distribution, bulk power generation and transmission,
and software layers.
A. Communication layer
In the communication layer, reliability is typically defined as
a fraction of unsuccessfully delivered data to the whole volume
of the transfered data (i.e. probability of data loss). Moreover,
it can be specified in terms of other Quality of Service (QoS)
parameters. These include latency, payload size, bandwidth or
frequency of data transfers [1]. Any violation of the specified
QoS can be perceived as a failure in the Smart Grid system.
The exact requirements also depend on the individual use cases
and the communicating components.
B. Distribution layer
At the distribution level, reliability is expressed with a
variety of performance indicators representing the availability
of the service from the consumer point of view. The most
common indicators are [12]:

Fig. 2. Communication infrastructure of the AMI [10]
•System Average Interruption Frequency Index
(SAIFI): this indicator is estimated by dividing the
number of customer interruptions by the total number
of customers served. Thus, it is a measure of the num-
ber of outages experienced by users. In UK, Customer
Interruption is used instead, computed as 100*SAIFI.
•System Average Interruption Duration Index
(SAIDI): this indicator is computed by dividing the sum
of long duration interruptions (i.e., longer than 3 minutes)
by the total number of customers. Thus, it is a measure
of the average amount of time when customers encounter
interruptions in electricity supply. In UK, it is called
Customer Minutes Lost (CML).
•Customer Average Interruption Duration Index
(CAIDI): this indicator is computed as the ratio of
SAIDI to SAIFI and is expressed in terms of minutes
per interruption.
These three reliability indicators are often prescribed in
the legislation and are monitored by the Energy Regulatory
Authority (ERA). The companies that provide power genera-
tion/distribution services need to obtain a license from ERA
which binds them to maintain the minimum values for SAIFI,
SAIDI and CAIDI.
Further indicators include system availability, number of
incidents, excursions beyond set voltage limits or excursion
beyond set frequency limits. Additionally, within the DIS-
CERN project the Key Performance Indicators (KPI) related
to individual Smart Grid use cases were surveyed [13]. Those
related to reliability are, for example, improvement of SAIDI,
improvement of voltage quality, peak load reduction, or suc-
cess index in meter reading.
C. Generation and transmission layer
On the transmission grid, reliability can be described by the
following [12]:
•Energy Not Supplied (ENS): total amount of energy
that would have been supplied if there were no interrup-
tions.
•Average Interruption Time (AIT): amount of time that
the electricity power supply is interrupted.
The indicators on the distribution and transmission layers
are useful when benchmarking the reliability performance
of the power grid, for example when we want to compare
multiple systems or when we want to determine whether
conditions of the service contract have been met. On the other
hand, they are not suitable to make predictions about future
system behaviour [14]. For example, the occurrence of failures
is stochastic in nature. Therefore, for prediction of system
failures, the probabilistic indicators, such as probability of
loss of load are recommended. Probability of loss of load
represents the probability that there will not be enough power
load to satisfy the demand, i.e. the outage occurs.
D. Software layer
In literature, the reliability of Smart Grids focuses on the
hardware components [1]. However, since a Smart Grid also
relies on various software systems, the potential failures in
their software components might also have a significant impact
on the Smart Grid operation [1], [15]. For example, there might
be a failure during the reading of the smart meter, resulting
in obtaining incorrect consumption data, and thus wrongly
calculating the price for power consumption. Although such
failure does not cause a power outage, it might have negative
impact on the associated stakeholders and possibly on other
use-cases as well (e.g. wrong information from smart meters
might influence the load management algorithms).
In software engineering, reliability can be defined as the
probability of successfully running software operations, or its
intended functions, for a specified period of time in a specified
environment [31]. There are several factors which influence
software reliability, such as inadequate testing, operations

Dataset Year Source Type Characteristics URL
REFIT [16] 2017 UK LOLP house aggregate loads / 9 individual appliance measure-
ments at 8-second intervals per house / 20 houses / 2
years
http://dx.doi.org/10.15129/31da3ece-f902
-4e95-a093-e0a9536983 c4
Cadiz-PQ [17] 2017 ESP PD signals recordings from the power network of the Uni-
versity of Cadiz / 5 years
https://ieee-dataport.org/documents/
real-life- power-quality-sags
µPMU [18] 2016 US LOLP data from 3 PMUs / voltage and current on 3 phases,
magnitude and phase angle / 3 months
https://pubarchive.lbl.gov/islandora/object/ir%
3A1006408
EPRI-DMD [19] 2015 US LOLP,
HW
aggregates multiple datasets / health index for grid de-
vices / power load disturbances
http://smartgrid.epri.com/DMD-DMI.aspx
UK-Dale [20] 2015 UK LOLP electricity usage data at a sample rate of 16 kHz for the
whole-house and at 1/6 Hz for individual appliances / 5
houses, 655 days
http://jack-kelly.com/data/
PLAID [21] 2014 USA LOLP 1094 current and voltage records from 11 different ap-
pliance types collected at 30 KHz
http://plaidplug.com
GREEND [22] 2014 ITA/AUT LOLP electricity usage data / 1 year data / 8 houses / 9 sensors
/ 1Hz resolution
https://sourceforge.net/projects/greend/
AMPds [23] 2013 USA LOLP electricity usage data / 1 year data / one house / 21 smart
meters
http://ampds.org
BERDS [24] 2013 USA LOLP energy usage & climate data Campus buildings UC
Berkeley
https://people.eecs.berkeley.edu/∼maasoumy/
(not online)
iAWE [25] 2013 IND LOLP energy & water usage, climate information for 3 houses,
73 days, 33 sensors
https://github.com/nipunbatra/Home
Deployment
Smart* [26] 2012 USA LOLP electricity usage data / high resolution dataset from 3
homes and low resolution dataset from 400 anonymous
homes / 1Hz resolution /multiple sensors & weather
http://traces.cs.umass.edu/index.php/Smart/
Smart
BLUED [27] 2012 USA LOLP electricity usage data / 8 days / 1 house / 12 Khz
resolution
http://inferlab.org/publications/blued/
ECO [28] 2012 SWI LOLP electricity usage data & occupancy data / 8 months / 6
houses / 1Hz resolution
https://www.vs.inf.ethz.ch/res/show.html?what=
eco-data
Tracebase [29] 2012 GER LOLP energy usage 1,000 event traces / 31 appliances https://www.tracebase.org
REDD [30] 2011 USA LOLP electricity usage data / 10 homes, 119 days http://redd.csail.mit.edu
TABLE I
SMA RT GRIDS RELIABILITY DATASETS. LOLP = LO SS OF LO AD PRO BAB IL ITY,PD=POWE R DISTRIBUTION,HW=HARDWARE
errors, different operating conditions, failures of underlying
hardware, low quality source code, interaction with external
services or applications, random events, and issues related to
the operational environment [32].
IV. DATA SE TS FOR SMART GRI DS
In this section, we describe the types of data that might serve
as the basis for deriving the reliability parameters discussed
in Section III. Additionally, we provide insights into our
methodology for dataset search, with the summary overview
of the datasets found, summarized in Table I.
A. Information Needs
In order to obtain network quality data at the communication
layer—such as latency or probability of data loss—we can
use the captured packet data (PCAP data). This kind of data
is often collected for SCADA systems. However, the main
purpose in that context is the detection of cyber-security
violations rather than reliability analysis.
At the distribution layer, the performance indicators de-
tailed in Section III.B are available in regular reports that
are required from the distribution companies. Other relevant
information could be collected from the sensor components
monitoring low level grid parameters, such as voltage or signal
disturbances.
The transmission layer indicators can be found in the
company reports. However, unlike the distribution reports, they
are typically not publicly available.
The data about failures in the individual software modules,
as well as hardware components, can be collected from event
logs. Software components in the Smart Grid systems log
their changes in state and behaviour into event records. Other
useful information can be timeseries of failure occurrences or
mean-time-between-failures (MTBF). They are often available
during the testing periods of the software and hardware
components and they can be used to derive the components
probability of failure.
To calculate the loss of load probability, data about the
actual load of the network, that is the volume of power
production and consumption, are necessary. The load profiles
can differ in the level of granularity and scope. They can range
from per minute to hourly, weekly or monthly values and
they can represent the load profiles of individual appliances,
households or full geographic areas.
B. Search Method
To set-up a search for datasets related to Smart Grid reliabil-
ity and to make it more systematic, we used the following strat-
egy. First, we used Google Scholar and Google search engines
with the keywords from Section III and IV followed by the

string “(Smart Grid Case Study) OR (experience report) AND
dataset”. Given the type of search, we privileged more gray
literature as a source of relevant datasets. Second, we searched
in open dataset repositories for Smart Grid analysis for existing
available datasets (e.g. OpenEI1). Next, we looked at resources
of real Smart Grid projects, since some of them provide related
datasets on their web pages2. Finally, since there is a high
demand for real datasets among researchers, we monitored
scientific community portals, such as Researchgate3, Zenodo4
and Figshare5for responses to dataset availability requests.
C. Datasets for Smart Grids
We report the datasets according to different type of infor-
mation needs (Table I). We divided the datasets in three main
categories, depending on the information that is available: Loss
of Load Probability (LOLP), Power Distribution (PD) and
Hardware (HW). Datasets range from year 2011 to 2017, and
the general motivation for the release of the datasets was due
to research needs, to allow for replicability of research results.
For this reason, there is a variety of very heterogeneous LOLP
datasets, in terms of number of residential units available,
appliances monitored, sensors, environmental information, but
also in terms of the measurement interval. The table summa-
rizes the main characteristics of the datasets according to these
aspects.
These datasets can be used for reliability prediction in loss
of load probability. Information about the load profiles has
been used in past research in several studies as input to various
analytical approaches, such as regression [33] or Bayesian
models [34].
V. DISCUSSION OF FINDINGS
The first aspect we noted during the study is the scarce
availability of specific failure information for Smart Grids
components (for example, in form of events logs). Our as-
sumption is that event logs analysis is very common in
software/system engineering, but such datasets are likely to
be available for general reliability analysis, non-specific to the
Smart Grid. Finding openly available failure event logs for
different Smart Grids components is challenging.
Another aspect worth noting is that gathering data about
Smart Grids networking communication was very demanding,
generally raw datasets are very rare. However, there is good
availability of datasets from generic SCADA systems, aimed
often more at security concerns, such as intrusion detection.
There are many datasets that can be used mainly to test
different intrusion detection algorithms.
We reported that there are vast amounts of datasets for load
control with different levels of granularity, and with sources
from disparate number/types of sensors. As highlighted in this
paper, such datasets can be relevant for reliability analysis, for
example for the prediction of over-current situations.
1https://en.openei.org/datasets/dataset?sectors=smartgrid
2e.g. http://traces.cs.umass.edu/index.php/Smart/Smart
3https://researchgate.com
4https://zenodo.org
5https://figshare.com
For distribution/transmission, the indicators we reported are
generally available in public reports, while the only ones
we found, covered in some datasets, are related to Phasor
Measurements (PM).
Finally, when reviewing literature for sources, a vast number
of papers dealing with data analysis in Smart Grids is using
more synthetic data from simulations, rather than using real
datasets. However, the availability of many datasets (Table I)
brings the possibility of better comparative research, allowing
for better research replicability and reproducibility, as all
datasets are openly available.
A. Assumptions and Limitations
A first limitation of the study is given by the fact that the
reliability in Smart Grid is not clearly defined, which is why
we needed to define different reliability views and categories.
As reported in the architecture overview, the complexity
and different layers of the Smart Grids pose challenges in
the definition of reliability, and very often aspects such as
resilience, availability, security are used interchangeably.
The second limitation is related to the exhaustiveness of the
search approach. However, although we might have missed
relevant resources due to their industrial rather than academic
context, we combined different sources with the aim of im-
proving the quality of the search result. As such, Table I is
meant more as a starting point for reliability engineers to
gather insights about datasets according to different reliability
categories, rather than a definitive table listing all existing
resources. In some of the categories, however, it is close
to being exhaustive. These are namely information about
software failures and detailed software and hardware event
logs, where hardly any work is publicly available, due to the
potential sensitive data included. On the other hand, there are
likely more industry-based datasets from companies available,
which have not been included as we have mainly focused on
datasets that have already been part/used in existing research.
VI. CONCLUSION
The complexity of Smart Grids renders software quality
requirements such as reliability, resilience, safety, security,
privacy, and performance of key relevance. Software reliability
is one of the key qualities of Smart Grid infrastructures.
However, the different layers of the Smart Grid make the
concept more complex than in traditional software reliability.
With the aim of increasing the understandability of relia-
bility in the Smart Grid context, we categorized information
needs for different reliability views, and explored different
open datasets available for reliability analysis. Overall, we
found many datasets that can be used for reliability analysis.
These can be useful to improve comparability across research
studies, as they represent an openly available source of data to
improve replicability and reproducibility of current research.
ACKNOWLEDGMENT
The work was supported from ERDF/ESF ”CyberSecurity,
CyberCrime and Critical Information Infrastructures Center of
Excellence” (No. CZ.02.1.01/0.0/0.0/16 019/0000822).

REFERENCES
[1] P. Hines, J. Veneman, and B. Tivnan, “Smart grid: Reliability, security,
and resiliency,” 2014.
[2] K. Sharma and L. M. Saini, “Performance analysis of smart metering for
smart grid: An overview,” Renewable and Sustainable Energy Reviews,
vol. 49, pp. 720 – 735, 2015.
[3] D. Niyato, P. Wang, and E. Hossain, “Reliability analysis and redun-
dancy design of smart grid wireless communications system for demand
side management,” IEEE Wireless Communications, vol. 19, no. 3, 2012.
[4] J. W¨
afler and P. E. Heegaard, “A combined structural and dynamic mod-
elling approach for dependability analysis in smart grid,” in Proceedings
of the 28th Annual ACM Symposium on Applied Computing. ACM,
2013, pp. 660–665.
[5] S. Lehnhoff, S. Rohjans, R. Holzer, F. Niedermeier, and H. de Meer,
“Mapping of self-organization properties and non-functional require-
ments in smart grids,” in International Workshop on Self-Organizing
Systems. Springer, 2013, pp. 163–168.
[6] S. G. C. G. CEN-CENELEC-ETSI, “Smart grid reference architecture,”
2012.
[7] H. Daki, A. El Hannani, A. Aqqal, A. Haidine, and A. Dahbi, “Big data
management in smart grid: concepts, requirements and implementation,”
Journal of Big Data, vol. 4, no. 1, p. 13, 2017.
[8] R. R. Mohassel, A. Fung, F. Mohammadi, and K. Raahemifar, “A survey
on advanced metering infrastructure,” International Journal of Electrical
Power & Energy Systems, vol. 63, pp. 473–484, 2014.
[9] S. Chren, B. Rossi, and T. Pitner, “Smart grids deployments within eu
projects: The role of smart meters,” in Smart Cities Symposium Prague
(SCSP), 2016. IEEE, 2016, pp. 1–5.
[10] B. Al-Omar, A. Al-Ali, R. Ahmed, and T. Landolsi, “Role of information
and communication technologies in the smart grid,” Journal of Emerging
Trends in Computing and Information Sciences, vol. 3, no. 5, pp. 707–
716, 2012.
[11] “European programme for critical infrastructure protection,” http:
//eur-lex.europa.eu/legal-content/EN/TXT/?uri=URISERV%3Al33260,
accessed: 2017-01-20.
[12] R. Nepal and T. Jamasb, “Security of european electricity systems: Con-
ceptualizing the assessment criteria and core indicators,” International
Journal of Critical Infrastructure Protection, vol. 6, no. 3, pp. 182–196,
2013.
[13] D. G. Energy, “Project DISCERN - KPI questionnaire,” 2014.
[14] R. N. Allan, Reliability evaluation of power systems. Springer Science
& Business Media, 2013.
[15] M. Schvarcbacher and B. Rossi, “Smart grids co-simulations with low-
cost hardware,” in Software Engineering and Advanced Applications
(SEAA), 2017 43rd Euromicro Conference on. IEEE, 2017, pp. 252–
255.
[16] D. Murray, L. Stankovic, and V. Stankovic, “An electrical load mea-
surements dataset of united kingdom households from a two-year
longitudinal study,” Scientific data, vol. 4, p. 160122, 2017.
[17] O. F.-O. M. J. E.-G. J.-J. G. de la Rosa; A. Agera-Prez; Jos Carlos
Palomares-Salas; J. M. Sierra-Fernndez, “Real-life power quality sags,”
2017. [Online]. Available: http://dx.doi.org/10.21227/H2K88D
[18] E. Stewart, A. Liao, and C. Roberts, “Open µpmu: A real world refer-
ence distribution micro-phasor measurement unit data set for research
and application development,” IEEE Power and Energy Society, 2016.
[19] “Epri distribution modernization demonstration (dmd) data mining
initiative,” 2015. [Online]. Available: http://smartgrid.epri.com/
DMD-DMI.aspx
[20] J. Kelly and W. Knottenbelt, “The uk-dale dataset, domestic appliance-
level electricity demand and whole-house demand from five uk homes,”
Scientific data, vol. 2, p. 150007, 2015.
[21] J. Gao, S. Giri, E. C. Kara, and M. Berg ´
es, “Plaid: a public dataset of
high-resoultion electrical appliance measurements for load identification
research: demo abstract,” in proceedings of the 1st ACM Conference on
Embedded Systems for Energy-Efficient Buildings. ACM, 2014, pp.
198–199.
[22] A. Monacchi, D. Egarter, W. Elmenreich, S. D’Alessandro, and A. M.
Tonello, “Greend: an energy consumption dataset of households in italy
and austria,” in Smart Grid Communications (SmartGridComm), 2014
IEEE International Conference on. IEEE, 2014, pp. 511–516.
[23] S. Makonin, F. Popowich, L. Bartram, B. Gill, and I. V. Bajic, “Ampds:
A public dataset for load disaggregation and eco-feedback research,” in
Electrical Power & Energy Conference (EPEC), 2013 IEEE. IEEE,
2013, pp. 1–6.
[24] M. Maasoumy, B. Sanandaji, K. Poolla, and A. S. Vincentelli, “Berds-
berkeley energy disaggregation data set,” in Proceedings of the Workshop
on Big Learning at the Conference on Neural Information Processing
Systems (NIPS), 2013.
[25] N. Batra, M. Gulati, A. Singh, and M. B. Srivastava, “It’s different:
Insights into home energy consumption in india,” in Proceedings of
the 5th ACM Workshop on Embedded Systems For Energy-Efficient
Buildings. ACM, 2013, pp. 1–8.
[26] S. Barker, A. Mishra, D. Irwin, E. Cecchet, P. Shenoy, and J. Albrecht,
“Smart*: An open data set and tools for enabling research in sustainable
homes,” SustKDD, August, vol. 111, p. 112, 2012.
[27] A. Filip, “Blued: A fully labeled public dataset for event-based non-
intrusive load monitoring research,” in 2nd Workshop on Data Mining
Applications in Sustainability (SustKDD), 2011, p. 2012.
[28] C. Beckel, W. Kleiminger, R. Cicchetti, T. Staake, and S. Santini, “The
eco data set and the performance of non-intrusive load monitoring
algorithms,” in Proceedings of the 1st ACM Conference on Embedded
Systems for Energy-Efficient Buildings. ACM, 2014, pp. 80–89.
[29] A. Reinhardt, P. Baumann, D. Burgstahler, M. Hollick, H. Chonov,
M. Werner, and R. Steinmetz, “On the accuracy of appliance identifi-
cation based on distributed load metering data,” in Sustainable Internet
and ICT for Sustainability (SustainIT), 2012. IEEE, 2012, pp. 1–9.
[30] J. Z. Kolter and M. J. Johnson, “Redd: A public data set for energy
disaggregation research,” in Workshop on Data Mining Applications in
Sustainability (SIGKDD), San Diego, CA, vol. 25, 2011, pp. 59–62.
[31] J. D. Musa, A. Iannino, and K. Okumoto, “Software reliability,” Ad-
vances in Computers, vol. 30, pp. 85–170, 1990.
[32] W. Ahmed and Y. W. Wu, “A survey on reliability in distributed
systems,” Journal of Computer and System Sciences, vol. 79, no. 8,
pp. 1243–1255, 2013.
[33] A. N. Celik, “Effect of different load profiles on the loss-of-load
probability of stand-alone photovoltaic systems,” Renewable Energy,
vol. 32, no. 12, pp. 2096–2115, 2007.
[34] D. C. Yu, T. C. Nguyen, and P. Haddawy, “Bayesian network model for
reliability assessment of power systems,” IEEE Transactions on Power
Systems, vol. 14, no. 2, pp. 426–432, 1999.