Redefining Smart Grid Architectural Thinking Using Stream Computing

• Cognizant 20-20 Insights

Redefining Smart Grid Architectural
Thinking Using Stream Computing
Executive Summary tecture will help utilities transform their power
grids into adaptive and intelligent infrastructures
After an extended pilot phase, smart meters
that inform continuous improvements in opera-
have moved into the mainstream for measuring
tional efficiency and business effectiveness.
the performance of a multiplicity of business
functions across the power utilities industry. This white paper explores the challenges and
Moving forward, the next objective is to create new benefits of Smart Grid creation and offers concrete
ways of handling large data sets for constructing thinking on new architectural approaches built
actionable responses to smart-meter-generated on emerging software standards that more
data, particularly when it comes to processes effectively leverage established forms of stream
such as validation estimation and evaluation, computing.1 It examines new thinking on ways to
demand response and load management. capture and analyze data generated by smart
meters that can help power utilities achieve new
As smart meters proliferate across power grids,
thresholds of performance over the near- and
energy utilities are dealing with huge packets of
long-term, while building better relationships with
data coursing through their IT networks. More and
consumers. We examine how stream data2 aids
more granular data holds the promise of enabling
usage forecasts (predicted by converting historic
faster and more informed decision making that
data coupled with real-time events into opera-
drives operational improvements and, perhaps,
tional KPIs) and identifies anomalies and patterns
enables consumers to better manage their own
in an ever-changing and high-transaction environ-
power consumption. To get there, however,
ment. In our view, when operational data is trans-
utilities must first conquer growing network
ported on a pervasive communication infrastruc-
latency challenges caused not only by the huge
ture (and coupled with two-way communication
profusion of smart-meter-generated data but
between utilities and consumers) data integration
also by processing inefficiencies created by their
challenges can be overcome, setting the stage for
dependence on more centralized models.
a brighter and more energy-efficient future.
Forward-thinking utilities need more distributed
and virtual complex event processing models that Using Cloud Platforms for Smart Meter
transform real-time operational data into applied Infrastructure
insights. Creating real-time operational knowledge One way to unlock the data treasure trove
can drive better demand response management, enabled by smart meters is by tapping virtual
improve quality of service and preempt fraud and data processing infrastructure delivered via
service outages before they inflict reputational cloud computing. Clouds offer the advantages of
damage. Rethinking their basic information archi- scalable and elastic resources to build software

cognizant 20-20 insights | june 2011

Consumers and Smart Meters: Interactions on a Cloud Stream

Active feedback of pricing
Load curtailment signals

Pow
er co
nsum
ption Residential
data Consumption
strea
m Hourly Consumption
Prediction
Pattern
Recognition
d ata
Weather

ata Commercial
ond Consumption
ucti
prod Historian Power

w er Generation

Po

Figure 1

infrastructure that support such dynamic, shows, Smart Grid applications that span smart
always-on applications. But the unique needs of meters (distributed at the consumer level),
energy informatics applications also highlight the cloud platforms (for data integration by service
challenges of using cloud platforms, such as the providers) and clusters (at energy utilities)
need to support efficient and reliable streaming, introduce systems heterogeneity, which efficient
low-latency scheduling and scale-out, as well as streaming is positioned to rationalize.
effective data sharing.
The need to perform complex processing with
Cloud platforms are an intrinsic component in minimal latency over large volumes of data has
creating a software architecture to drive more led to the evolution of various data processing
effective use of Smart Grid applications. The paradigms. Industry majors such as IBM, Oracle,
primary reason: Cloud data centers can accom- Microsoft and SAP have developed event-oriented
modate the large-scale data interactions that application development approaches to decrease
take place on Smart Grids and are better archi- the latency in processing large volumes of data.
tected than centralized systems to process the These efforts reveal the following:
huge, persistent flows of data generated across
the utility value chain. Figure 1 shows how this • Since smart meters generate interval data
might work within a power utilities company. that is time-series in nature, companies need
efficient ways of processing queries incremen-
The computational demand for demand-response tally and via in-memory technologies. They
applications will be a function of the energy then need a way to apply the results to their
deficit between supply and demand. This typically emerging dynamic business process models.
oscillates based on the time of the day and
possible weather conditions. This translates into a
• Since this buffered data is also stored offline
for static analysis, mining, tracing and back-
need for compute- intensive, low-latency response
testing, companies need a means of managing
at midday and limited analysis in off-peak evening
and accessing these stores efficiently.
hours. The elastic nature of cloud resources makes
it possible for utilities to avoid costly capital As Smart Grids proliferate, businesses require
investment for their peak computation needs. greater data availability rates. Companies can no
longer afford to collect all time-series data, load it
This results in information sharing on real-time into a database and then build database indexes
energy usage and power pricing. As Figure 1 for query efficiency. Instead, businesses need

cognizant 20-20 insights 2

these queries to be continuously and incremen- Ease of Management
tally computed and updated as new relevant data
To effectively deploy smart meters and the data
arrives from smart meter sources. This will avoid
they generate, a number of factors need to be
the need to re-process existing data. Incremental
addressed, including query composability and
computation is necessary to create a low-latency
ease of deployment over a variety of environ-
response to continuously flowing time-series data.
ments, such as single servers and clusters. Query
Complex event processing (CEP) is a widely used composability requires the ability to “publish”
technique in smart meter data processing, where query results, as well as the ability for Continuous
data is continuously monitored, verified and acted Query (CQ) to consume results of existing CQs
upon, given ongoing and changing conditions. and streams.
With this approach, data, including the event
Typical meter streaming queries entail rules such
streams from multiple sources, is processed based
as:
on a declarative query language. Importantly, all
of this is accomplished with near-zero latency. • Present the top three values every 10 minutes.

Event-Driven Data Processing • Compute a running average of each sensor
value over the last 20 seconds.
Challenges
The key attributes of complex event processing • Filter out sensor readings when the device was
in a maintenance period.
include:

• Express fundamental query logic: Incorpo- • Illustrate when event “A” was followed by event
“B” within three minutes.
rate windowed processing and time progress
as a core component for query logic. OSIsoft’s PI System provides power utilities
• Handle error or delayed data: Delayed with the leading operation data management
processing until guaranteed, with no late-arriv- infrastructure for Smart Grid components that
ing events. This increases latency; otherwise, encompass power generation, transmission and
aggressively process event and produce distribution. This software provides capabilities
tuples.3 for energy management, condition-based mainte-
• Extensibility: Given the complexity of meter nance, operational performance monitoring, cur-
data and event operations, there is a need tailment programs, renewable energy monitoring
to support custom-built streaming logic as and phasor monitoring of transmission lines,
libraries. among other functionalities. `

• Universal specification: Semantics of query OSIsoft MDUS integrates a utility’s meter system
logic need to be independent of when and how and SAP’s AMI Integration for Utilities to perform
programmers physically read and understand tasks such as billing. It also provides the ability to
events. Applications time, rather than system integrate meter data with other operational data.
time, is used to enable a universal time zone. It serves as a real-time data collector, which is
head-end system vendor-independent.
These attributes ensure that with complex event
processing, query logic is kept generic regarding Integration of meter data into business systems
how events are read and how their output is inter- such as billing requires data validation and other
preted. Tuples should follow universal time, which forms of aggregations. OSIsoft has chosen to
can be read and processed on any system. leverage CEP to accomplish this task. CEP provides
the scalability required by SAP AMI and utilizes a
Performance Implications
SQL-based language for defining the validation
In-stream processing doesn’t allow data to be rules. OSIsoft uses Microsoft’s StreamInsight
written back to the disk for processing later from CEP engine, which enables utilities to define and
internal state in main memory. With smart meter implement required meter data validation. This
data, an event queue is filled to capacity once puts this important facet of regulatory compliance
the arrival rate is greater than the processing requirements into the hands of the utility’s IT and
capability of the system. The metrics used to business specialists.
manage the data stream are latency, throughput,
correctness and memory usage.


PI Interface Node
Foreign
Device
System
PI Server
Input Adapter(s) Output Adapter(s)
Data
Source Queries
Stream Insight Engine (vs .NET-
LINQ)
There are two ways streaming can be adopted in
Complex Event Processing Engine and deployed on the Eucalyptus4 private cloud,5
current meter data systems: shows 50% bandwidth savings, resulting from
adaptive stream rate control.
• Server-side streaming: The stream is pro-
cessed on the (OSIsoft) PI snapshot and
Low-latency stream processing is a key com-
streamed with the processed results back to
ponent of the software architecture required
the PI server (see Figure 2).
to support demand-response applications. The
stream processing system ingests smart meter
data arriving from consumers and acts as a first
PI Server
responder to detect local and global power usage
skews and to alert the utility operator. At 1KB per
event generated each minute, 2TB of data will
Queries
Stream Insight Engine (vs .NET-LINQ) stream each day. Processing such large-scale
streams can be compute- and data-intensive;
public or private cloud platforms provide a scal-
Figure 2 able and flexible infrastructure for building such
Smart Grid applications.
• Edge processing: In this model, the CQs
are applied at the data source (and at the PI However, computational and bandwidth con-
interface level), where the results are only straints at the consumer and utility levels mean
stored as processed data (see Figure 3). that power usage data streamed at static rates
from smart meters to the utility can either be
at too high a latency to detect usage skews in a
PI Interface Node timely manner or at too high a rate to computa-
Foreign
Device tionally overwhelm the system. Smart meters
System connect to the utility using heterogeneous
PI Server
Data
Queries
networks and range from low bandwidth power
Source
Stream Insight Engine (vs .NET-
LINQ)
line carriers at ~20Kbps, to 3G cellular networks
Complex Event Processing Engine
at ~2Mbps, as well as ZigBee at ~250Kbps.
Network bandwidth can thus be a scare resource
Figure 3
at the consumer end. In the case of smart meters,
traffic can be bursty, since data is sent indepen-
dently, causing instantaneous bandwidth needs
Cloud and Adaptive Rate Control
to spike.
The growing importance for utilities to process
and analyze thousands of meter data streams In the case of high power demand, meters emit
PI Server
suggests that they should a large volume of information, which requires a
The growing consider the adoption of
throttle controller to respond to these events and
control latency.
importance for utilities cloud platforms.NET-LINQ)
Stream Insight scalable,
to achieve
Queries
Engine latency-sensitive
(vs
to process and analyze stream processing. One Applying InfoSphere Streams
thousands of meter approach to consider is IBM InfoSphere Streams is a stream processing
data streams suggests adaptive rate control, which system that continuously analyzes massive
is the process of restrict- volumes of streaming data for business activity
that they should ing the stream rate to meet monitoring and active diagnostics. It consists
consider the adoption accuracy requirements for of a runtime environment that contains stream
of cloud platforms Smart Grid applications. instances running on one or more hosts. Within
This approach consumes InfoSphere is a Stream Processing Application
to achieve scalable, less bandwidth and com- Declarative Engine (known as SPADE), which is
latency-sensitive putational overhead within a stream programming model (executed by the
stream processing. the cloud for stream runtime environment) that supports stream
processing. The experi- data sources that continuously generate tuples
mentation of the Smart Grid stream processing containing typed attributes.
pipeline, modeled using IBM InfoSphere Streams


Tracking Energy Consumption

A stream processing pipeline is used to continuously monitor energy usage. Processing elements
in dotted lines show the addition of throttle logic.

Notify Notify
DB/File
1 max
if(u 1 >U ) if(u 1 >.136*u avg)
1 1
Update u1sum Update u1avg
(m1,t1,u11)
Store Running AMI’s 15-min
Condition Condition average
daily sum R1++ Increase
AMI rate
(mn,t1,un 1) if(c1-u1avg < accept)
Condition Utility’s 15-min Condition
average
Network

Update u avg
1
Decrease
Superscript = Meter ID AMI rate
Subscript = Time R1

Figure 4

Figure 5 shows the smart meters present on the performed for each smart meter stream (shaded
public Internet that generate power usage data in brown in Figure 4.
streams accessible over TCP sockets. Here, the
Next, the pipeline aggregates smart meter tuples
InfoSphere streams run on a cluster that doesn’t
across all streams using a tumbling window to
support out-of-box deployment on a cloud plat-
calculate the cumulative consumer energy usage
form. To instantiate a stream processing environ-
within a 15-minute time window. This stream
ment on a Eucalyptus private cloud, a customized
operator (shaded blue in Figure 4) calculates the
VM image must be built that supports InfoSphere
total load on the utility. It can be used to alert the
streams. Communication to the stream instance
utility manager in case, say, the total consumption
is activated when the VM instances are online.
reaches 80%, 90% and >100% of available power
This communication, however, is initiated exter-
capacity at the utility. Operators shown in dotted
nally by a SPADE application started on a stream
lines (Figure 4) are not part of the application
instance and configured with a list of named
logic and form the adaptive throttling introduced
stream instances on specific hosts.
next. This core model could be used in demand
Each smart meter is a stream source whose response management.
tuples have the identity of the smart meter,
power used within a time duration, as well as the
SAP Event Insight
timestamps of the measurement interval. Addi- The emergence of smarter grids powered by
tional meta data about the smart meter and con- stream computing has made clear the need for
sumer is part of the payload but will be ignored more robust processing at the enterprise systems
for the purposes of this discussion. Each tuple level. These systems typically struggle to keep
is about 1KB in size. The pipeline first checks if pace with high data volume and a large number
each individual power usage tuple reports usage of heterogeneous and widely dispersed data
that exceeds a certain constant threshold, Umax sources and changing data requirements. This is
m defined by the utility. being resolved by enterprise software systems
such as mySAP ERP, which have begun to adapt
Crossing this threshold will trigger a critical in-memory processing algorithms for this new
notification to a utility manager. Next, a relative architectural proposition. The result is that SAP
condition will check to see if the user’s consump- can now deliver an event insight application that
tion increases by more than 25% since his/her understands the impact of operational events
previous consumption. This will trigger a less in real time. In-memory processing not only
critical notification. The pipeline then archives brings just-in-time rhyme and reason to real-time
the tuple into a sink file and proceeds to compute business events, but it can also do so with signifi-
a running sum of the daily usage by the consumer. cantly less effort, a reduction in reporting, oper-
Subsequently, the running average over a ational and opportunity costs, which can power
tumbling window is updated. These operations are competitive advantage.


Architecture of Stream Processing and the Throttle Controller

Control Feedbacks
Throttle Controller
InfoSphere Streams
Input Streams Streams Processing Response

TCP/IP
Electric AMI
Gas
Industrial/Commercial Data Files

Electric AMI
Gas
Data Files
Residential Building

Figure 5

Looking Down the Road network optimization and intelligent processing,
saving money by automating their demand
We see stream computing as a key element of the
response program and load management
future of work that could be applied broadly by
processes. Standardizing these processes saves
the power utilities industry. Our view is that its
IT maintenance expense, freeing capital to be
deployment would minimize network latency and
invested in other core business activities.
function as a key component for demand response
management. Moreover, we are planning to inves- In a business context, this approach will help
tigate stream computing on the cloud platform. utilities with energy efficiency programs and
Our research will appraise the throughput of grid management. It does this by providing a
a stream processing system and its latency in mechanism to convert dollars saved by eliminat-
processing each tuple as the stream rates adapt. ing inefficient energy generation and distribution
toward more effective asset management.
This approach will help utilities that are adopting
Smart Grids in their mainstream business with

Footnotes
1
Stream computing is a high-performance computer system that analyzes multiple data streams from
many sources, live. Stream computing uses software algorithms to analyze data in real time, which
increases speed and accuracy when dealing with data handling and analysis.
2
Stream data is a sequence of digitally encoded coherent signals (packets of data or data packets) used
to transmit or receive information.
3
Tuple is an ordered pair of energy data to be processed and is an effective way of representing
in-stream computing.
4
Eucalyptus Cloud is a software platform for the implementation of private cloud computing on
computer clusters.


5
Private clouds are internal clouds that, according to some vendors, emulate cloud computing on private
networks. These (typically virtualization automation) products offer the ability to host applications or
virtual machines in a company’s own set of hosts. They provide the benefits of utility computing,
such as shared hardware costs, the ability to recover from failure and the ability to scale up or down
depending upon demand.

References
“IBM Infosphere Streams Version 1.2.1: Programming Model and Language Reference,” IBM, Oct. 4,
2010, http://publib.boulder.ibm.com/infocenter/streams/v1r2/topic/com.ibm.swg.im.infosphere.streams.
product.doc/doc/IBMInfoSphereStreams-LangRef.pdf.

D. J. Abadi, Y. Ahmad, M. Balazinska, U. Cetintemel, M. Cherniack, J. H. Hwang, W. Lindner, A. Maskey,
A. Rasin, E. Ryvkina, N. Tatbul, Y. Xing and S. B. Zdonik, “The Design of the Borealis Stream Processing
Engine,” Proceedings of the Second Biennial Conference on Innovative Data Systems Research, pp
277-289, January 2005.

D. J. Abadi, D. Carney, U. Cetintemel, M. Cherniack, C. Convey, S. Lee, M. Stonebraker, N. Tatbul and S.
Zdonik. “Aurora: A New Model and Architecture for Data Stream Management,” The VLDB Journal, Vol
12, Issue 2, August 2003.

A. Arasu, S. Babu and J. Widom. “The CQL Continuous Query Language: Semantic Foundations and
Query Execution.” The VLDB Journal, Vol 15, Issue 2, June 2006.

A. M. Ayad, J. F. Naughton. “Static Optimization of Conjunctive Queries with Sliding Windows Over Infinite
Streams,” Proceedings of the International Conference on Management of Data, SIGMOD 2004, ACM.

C. Ballard, D. M. Farrell, M. Lee, P. D. Stone, S. Thibault and S. Tucker, “IBM InfoSphere Streams Harnessing
Data in Motion,” IBM, September 2010.

A. Biem, E. Bouillet, H. Feng, A. Ranganathan, A. Riabov, O. Verscheure, H. Koutsopoulos and C. Moran,
“IBM InfoSphere Streams for Scalable, Real-Time Intelligent Transportation Services,” Proceedings of
the International Conference on Management of Data, SIGMOD 2010, pp 1,093-1,104, ACM.
S. Chandrasekaran, O. Cooper, A. Deshpande, M. J. Franklin, J. M. Hellerstein, W. Hong, S. Krishnamurthy,
S. Madden, V. Raman, F. Reiss and M. A. Shah, “TelegraphCQ: Continuous Dataflow Processing for an
Uncertain World,” SIGMOD 2003, ACM.

StreamBase, http://www.streambase.com/

D. Abadi et al., “The Design of the Borealis Stream Processing Engine.”

“Why IP is the Right Foundation for the Smart Grid,” Cisco Systems, Inc., January 2010.

“The Role of the Internet Protocol (IP) in AMI Networks for Smart Grid,” National Institute of Standards
and Technology, NIST PAP 01, Oct. 24, 2009.

D. Zinn, Q. Hart, B. Ludaescher and Y. Simmhann, “Streaming Satellite Data to Cloud Workflows for
On-Demand Computing of Environmental Products,” Workshop on Workflows in Support of Large-Scale
Science (WORKS), 2010.

Arvind Arasu, Shivnath Babu, Jennifer Widom, ”CQL: A Language for Continuous Queries over Streams
and Relations,” Database Programming Languages, 9th International Workshop, DBPL 2003, Potsdam,
Germany, Sept. 6-8, 2003.
“Pattern Detection with StreamInsight” Microsoft StreamInsight blog, Sept. 2, 2010, http://tinyurl.
com/2afzbhd
“InfoSphere Streams,” IBM, http://www.ibm.com/software/data/infosphere/streams


About the Author
Ajoy Kumar is a Senior Architect within Cognizant’s Manufacturing and Logistics Practice, where he
is working on the Smart Grid program that focuses on Smart Grid architecture, design performance,
demand response, enterprise integration and meter data management. Before joining Cognizant, he
worked with OSIsoft, Inc. where he led numerous initiatives, including one in which he spearheaded
the development of a meter data unification system integrating OSIsoft and SAP AG. Ajoy has also
worked extensively in the energy, pharma, chemical and mining and steel industries and has spent over
17 years focused on information technology. Ajoy holds a Master’s Degree in Computer Science. He can
be reached at ajoykumar.arumugam@cognizant.com.

About Cognizant

Cognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process out-
sourcing services, dedicated to helping the world’s leading companies build stronger businesses. Headquartered in
Teaneck, New Jersey (U.S.), Cognizant combines a passion for client satisfaction, technology innovation, deep industry
and business process expertise, and a global, collaborative workforce that embodies the future of work. With over 50
delivery centers worldwide and approximately 111,000 employees as of March 31, 2011, Cognizant is a member of the
NASDAQ-100, the S&P 500, the Forbes Global 2000, and the Fortune 500 and is ranked among the top performing and
fastest growing companies in the world. Visit us online at www.cognizant.com or follow us on Twitter: Cognizant.

World Headquarters European Headquarters India Operations Headquarters
500 Frank W. Burr Blvd. Haymarket House #5/535, Old Mahabalipuram Road
Teaneck, NJ 07666 USA 28-29 Haymarket Okkiyam Pettai, Thoraipakkam
Phone: +1 201 801 0233 London SW1Y 4SP UK Chennai, 600 096 India
Fax: +1 201 801 0243 Phone: +44 (0) 20 7321 4888 Phone: +91 (0) 44 4209 6000
Toll Free: +1 888 937 3277 Fax: +44 (0) 20 7321 4890 Fax: +91 (0) 44 4209 6060
Email: inquiry@cognizant.com Email: infouk@cognizant.com Email: inquiryindia@cognizant.com

© Copyright 2011, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is
subject to change without notice. All other trademarks mentioned herein are the property of their respective owners.

Redefining Smart Grid Architectural Thinking Using Stream Computing

More Related Content

What's hot (14)

Similar to Redefining Smart Grid Architectural Thinking Using Stream Computing (20)

More from Cognizant (20)

Recently uploaded (20)

Redefining Smart Grid Architectural Thinking Using Stream Computing