Uploaded by PETER SAWYER

Evaluation of WiMax (802.16) Standard for Smart Grid Last Mile Communications. MSC Thesis

advertisement
Evaluation of WiMax (802.16) Standard for Smart Grid
Last Mile Communications.
Sawyer E. Peter
S11737214
A THESIS SUBMITTED IN PARTIAL FULFILMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
Master of Science MSc.(Telecommunications)
Faculty of Technology, Engineering & the Environment
School of Computing, Telecommunications and Networks
BIRMINGHAM CITY UNIVERSITY, UK.
October, 2013
i
Declaration: I have read and I understand the MSc dissertation guidelines
on plagiarism and cheating, and I certify that this submission fully complies
with these guidelines.
ii
ACKNOWLEDGMENTS
Firstly, I want to express my gratitude in a very special way to almighty God
for the grace of completion he gave to me for the success of this project.
Thank you Lord.
I also want to thank my supervisor, Shane mcMordie for his guidance and
supervision, in the development of this thesis, without twhose remarks and
observations this project couldn’t have been successfully completed.
My very special gratitude goes out to my family, especially my dearly
beloved parents whose encouragement and support have made the success
of this project posssible. To them I express my deepest love.
In a very special way ,I also wish to acknowledge my boss and colleagues
at work whose patience and understanding have contributed by no small
means to the success of this project. God bless you guys abundantly.
iii
ABSTRACT
A smart grid is the traditional electricity distribution network incorporated with
ICT and telecommunications technologies. Telecommunication is required
inorder to facilitate the exchange of information generated by SGLM
applications between customer’s smart meters and other substation
distribution devices, supporting their operational and functional
requirements.
Utilities recognize that with Advanced Metering Infrastructure (AMI) and
Distributed Automation Systems(DAS) applications on their last mile, the
SGLM interfaces need to be application aware, provide granular Quality of
Service (QoS) provisioning for mission critical data and posses fast
response times , with pervasive coverage.
Wireless technologies are well suited to these requirements but WiMax is
especially investigated due to its granular QoS support and wide range
coverage ability. Thus, this research work is concerned with the investigation
of the performance of WiMax in supporting the smart grid’s last mile AMI and
DAS application traffic. A WiMax communication architecture for the
simulation using OPNET modeler is advanced based on the development of
a traffic model developed with traffic classification and the use of a simple
priority scheduling algorithm.
Simulation results of delay, throughput and packet delivery success with
varied node densitiesare compared with the previously established
communication requirementsin the literature reviewof smart grid applications
of smart metering and distribution automation in determining the
performance of the proposed WiMax network architecture.
The results from the simulation reveal that the WiMAX network’s architecture
proposed efficiently and reliably meets the QoS requirements of smart grid
applications in the last mile of the SG.
iv
TABLE OF CONTENTS
Acknowledgments ..........................................................................................i
Abstract ..........................................................................................................iii
Table Of Contents..........................................................................................v
List Of Figures ...............................................................................................vii
List Of Tables ................................................................................................xi
List Of Acronyms ...........................................................................................xiii
CHAPTER 1.....................................................................................................1
Introduction ......................................................................................................1
1.1. Motivation ..................................................................................................2
1.1.1 Scope of Thesis.......................................................................................4
1.2. Aims and Objectives .................................................................................6
1.3. Research method and design ...................................................................6
1.4. Thesis outline.............................................................................................7
CHAPTER 2. ...................................................................................................8
Smart Grids last mile communication..........................................................8
2.1. Introduction................................................................................................8
2.2. Smart Grid definition functions and objectives...........................................9
2.3. Smart grid reference models ...................................................................10
2.3.1. Existing communication architecture.......................................... ..........10
2.3.2 NIST interoperability conceptual framew................................................10
2.3.2.1 Bulk generation domain.......................................................................12
2.3.2.2 Transmission ......................................................................................13
2.3.2.3 Distribution domain..............................................................................13
2.3.2.4 Operations domain...............................................................................14
2.3.2.5 Markets domain....................................................................................14
2.3.2.6 Customer domain.................................................................................14
2.3.2.7 Services domain...................................................................................15
2.4. Smart Grids Last Mile applications and functions.....................................15
2.4.1. Advanced metering Infrastructure (AMI).................................................16
2.4.1.1 Smart meter..........................................................................................17
2.4.1.2 Customer Premises Network (CPN).....................................................17
2.4.1.3 Meter Data Management System.........................................................18
2.4.1.4 SGLM Communication Infrastructure...................................................18
2.4.2. Distribution Automation Systems(DAS)...................................................20
2.4.2.1 Dynamic feeder reconfiguration............................................................20
2.4.2.2 Conservation voltage & dynamic VAR control.......................................21
2.4.3. Distributed energy resources .................................. ...............................21
2.4.4. Electric Vehivles.......................................................................................21
2.4.5 Demand Response ..................................................................................21
2.5 Smart grid communication requirement.......................................................22
2.5.1 Quality of service(QoS).............................................................................22
2.5.1.1 Latency....... ..........................................................................................22
2.5.1.2 Bandwidth requirements........................................................................23
v
2.5.2 Interoperability..........................................................................................24
2.5.3 Reliability...................................................................................................25
2.5.4 Security.....................................................................................................27
2.6 SGLM communication technologies.............................................................27
2.6.1 Wireline technologies................................................................................28
2.6.1.1 Power Line Communications(PLC) .......................................................28
2.6.1.2 GPON.....................................................................................................29
2.6.1.3 DOCSIS..................................................................................................29
2.6.2 SGLM wireless communication technologies............................................29
2.6.2.1 RF mesh 802.15.4..................................................................................30
2.6.2.2 3G/4G/LTE/GPRS/EDGE/HSPDA...........................................................30
2.6.2.3 WLAN 802.11n/g....................................................................................31
2.6.2.4 Zigbee 802.15.4.....................................................................................31
2.6.2.5 Smart Utility Network(SUN)802.15.4g ...................................................32
2.6.2.5 WiMax 802.16........................................................................................33
2.7. Summary ....................................................................................................34
CHAPTER 3.......................................................................................................32
Research method and design.......... ...............................................................32
3.1. Introduction .................................................................................................35
3.2 Simulation as a research tool........................................................................39
3.3 Smart grid last mile traffic model...................................................................46
3.4 Network model.................................................... .........................................46
3.5. Summary......................................................................................................49
CHAPTER 4. .....................................................................................................50
Simulations and Results ...................................................................... ..........50
4.1. Introduction........ .........................................................................................50
4.2. Simulation model ........................................................................................50
4.3. Presentation of results ................................................................................52
4.3.1 Latency results presentation..... ................................................................52
4.3.1.1 UL latency results presentation...............................................................53
4.3.1.2 DL latency results presentation...............................................................54
4.3.2 Througput simulation results......................................................................55
4.3.2.1 DL throughput simulation results.............................................................55
4.3.2.1 UL throughput simulation results.............................................................56
4.3.3 Relibility results presentation......................................................................57
4.4 Summary of findings.....................................................................................59
CHAPTER 5. ......................................................................................................60
Discussion of results............................................................................................60
5.1 Introduction....................................................................................................60
5.2 Packet delay analysis ...................................................................................61
5.3. Throughput analysis ....................................................................................62
5.4. Reliability results analysis ............................................................................63
5.5. Summary .....................................................................................................63
vi
CHAPTER 6. ....................................................................................................65
Conclusions and future work ........................................................................ ...65
6.1. Thesis summary ........................................................................................65
6.2. Recommendations .............................................................................. ......66
References .......................................................................................................67
Bibliography.......................................................................................................72
vii
LIST OF FIGURES
Figure 1.0. Overview of electric power system. .............................................11
Figure 2.1 Smart grid conceptual flow of information.....................................12
Figure 2.2 AMI network overview...................................................................17
Figure 2.3 Smart grid communication networks.............................................19
Figure 2.4 SGLM communication technologies mapping...............................28
Figure 2.5 WiMax MAC & PHY layer architecture..........................................33
Figure 3.1 Architecture of proposed WiMax network.....................................47
Figure 4.2 UL average delay results from SGLM traffic service class............54
Figure 4.3DL average delay results from SGLM traffic service class.............55
Figure 4.4Throughput in the DL of SGLM traffic service class. .....................56
Figure 4.5 Throughput in the UL of SGLM traffic service class. ....................57
Figure 4.6 Packet delivery ratio for SGLM service classes............................58
viii
LIST OF TABLES
Table 2.0 Adopted SGLM applications and their network requirements..............16
Table 3.1 traffic flows mapped to service classes ...............................................20
Table 3.2Scheduling type mapping to SG applications.........................................20
Table 3.3Mandatory QoS parameters for each scheduling service......................20
Table 3.4 IP ToS mapping to scheduling type......................................................32
Table 3.5 Service class configuration for SG application....... ..............................33
Table 3.6 Mapped reliability parameters to service classes........................... ......33
Table 4.0(a) Simulation Parameters.......................................................................34
Table 4.0(b) QoS Service Classes Configuration Parameters...............................34
Table 4.1 UL & DL traffic source classification................. ....................................35
Table 4.2 UL average delay results for SGLM traffic service classes....................35
Table 4.3 DL average delay results for SGLM traffic service classes....................35
Table 4.4 Throughput in the DL for SGLM service classes....................................38
Table 4.5 Throughput in the UL for SGLM service classes....................................38
Table 4.6 Packet delivery ratio%............................................................................52
Table 5.1 Comparison of simulation results and literature review packet latency
values....................................................................................................................54
ix
x
CHAPTER 1
1.0 INTRODUCTION
The smart grid has been widely credited for revolutionizing the conventional
electricity grid as we know it. Incorporating information technology (IT) and
telecommunications, it has changed the model of unidirectional energy flow
in the electricity grid enabling a two way flow of energy including data
between generation sites and customer’s premises and this is arguably its
most significant advantage (Lima , 2010).
The two-way flow of energy and communication data has empowered the
electricity grid to efficiently and cost effectively integrate renewable
distributed energy resources(DERs) or microgrids both at the distribution
and customers premises levels with the use of Advanced Metering
Infrastructure (AMI) and Distribution Automation Systems (DAS).
The European Union(EU) with its 20-20-20 vision which targets a 20%
increase in energy efficiency,20% reduction of CO2 emissions, and 20%
renewables inclusion by 2020, is at the forefront of the global call for the
reduction of dependence on local or imported fossil fuels. This mitigates the
effect of greenhouse gas (GHG) emissions and encourages the use of
renewable Distributed Energy Resources( DERs).The vision seeks to
enhance the reliability of the electricity grid and energy security through
diversification of energy sources {Barosso, 2010} {VanderDrift,2011}. The
intermittency and uncertainty of these renewable energy sources constitutes
a problem {Camacho2011}, creating a need for the smart grid solution that
incorporates communications, computations and controls.
Additionally, the electricity grid is being remodelled to a demand response
system through the smart grid’s AMI functionality, where actual generation
and consumption of energy which is its demand and supply is determined by
real time pricing signals. As Haerick (2012) puts it, demand response is the
voluntary, temporary adjustment of power demand by end-user or
counterparty in response to market signal (e.g. price, emergency, etc.)
1
Demand response systems are vital to the smart grid actualization as they
reduce the peak generating capacity requirement.
AMI and DAS enables these functionalities through bi-directional
communication of mission critical metering data between the customer’s
smart meters and the rest of the grid. Communication technologies that
connect consumers to the smart grid (SG) have been reffered to as
Advanced Metering Interfaces (AMI) Gomez-Cuba (2011). The designing
and provisioning of this access network/last mile of the smart grid is critical
in actualizing the goals and benefits of smart metering. Because AMI
represents only one possible use of the last mile network ,Gomez-Cuba
(2011), we adopt the term last mile for the sake of this dissertation.
1.1 MOTIVATION
The overlay network of the smart grid’s last mile is called the Neigbourhood
Area Network(NAN), and has been adjudged the most critical subnetwork of
the smart grid.(Gomez-cuba, 2011). The NAN communication technology to
be used, wireless or wired must scalably accomodate a large number of
smart meters numbering in their tens of thousands(Gomez-cuba2011) and
support the astronomical growth in data (Claudio,2010) the future holds for
the smart grid.
Wireless communication represents the final step for smart grids to reach
the widest coverage. They insist that though most of the nodes of the grid
will be connected using high-end wired communication, it is neither cost nor
energy efficient when applied to billions of devices belonging to millions of
users. (Zori, et al 2011). For access to remote locations, some form of
wireless technology is required in bridging the communication gap. An
access point must aggregate traffic from the tens of thousands of smart
metering households making wired solutions unattractive. More urgently, it
must meet the minimal bandwidth, data latency, QoS and throughput
requirements of the varied AMI and DAS applications sent through the
communication channel.
2
The ITU Focus Group on Smart Grid (2011) iterates that the choice of what
type of network is needed to support a particular smart grid function shall be
driven by the requirements of that function. The IEEE 2030 delivery time
standard on SGLM traffic and the IEEE Power Engineering Society IEEE
Standard 1646 define delivery-time performance requirements for substation
and Intelligent electronic devices (IED) telecommunications.
These standards also specify broad targets for the communication network
parameters, such as bandwidth, latency, bitrate, QoS etc that vary by
application. (Al-omar,2012),(Claudio, 2010). According to Arnold( 2013),
IEEE 2030 specifies the communications infrastructure necessary for smart
grid, from high-speed synchrophaser data to in-premise meter and customer
notification systems.Bui (2011) succintly states that constrained wireless
networks of the future will be able to schedule the energy expenditure of
every single appliance in the households enabling the distributed monitoring
and control of recent renewable energy sources, such as photovoltaic
panels.
Utilities and energy distributors must now implement a reliable, scalable
secure wireless communications technology to efficiently support the
bidirectional flow of energy on the grid, meeting the bandwidth, throughput,
latency, bitrate, QoS and security requirements of the AMI and DA
applications. Third party or commercial cellular telecommunications
providers cannot meet the reliability and security needs of utilities, and
hence utilities must own their own infrastructure.
Pavlovski (2010),iterates the factors that have informed utilities deploying
their own wireless infrastructure to support grid transformation as:
1.The need to support several thousand simultaneous devices
2.Higher quality of service( application awareness)
3. Priority for mission critical traffic
3
The communications technology employed in the last-mile of the smart grid,
must reliably and securely meet the requirements of the AMI and DAS
applications that it would be conveying. These requirements in terms of data
latency, bandwidth, reliability and security on the network must be rigidly met
in order to acheive the smart grid objectives. These objectives according to
Hammoudeh (2012) includes; the integration of renewable energy
resources, demand-side load management, self healing mechanism,
improved security and reliability etc. In the words of (Gao, 2012) “Before a
communication technology is chosen for a particular power system
application, a thorough analysis is required to match the application
requirements with the technology properties”.
In this thesis, the WiMax 802.16 standard is investigated to ascertain its
ability to effectively meet the bandwidth, latency, bitrate , QoS and reliability
concerns of utilities’ smart grid last mile applications.
1.1.1 Scope of the Thesis.
The electric power grid in its entirety comprises of a High Voltage (HV)
transport network,transporting power from power generation plants to a
Medium Voltage(MV) tier via HV/MV transmission substations as seen in fig
1.0 below.The MV distribution network takes power from the HV/MV
substation to Low Voltage(LV) consumers that are fed from a MV/LV
distribution substation.
(Kirkham,2010), (Hammoudeh,2012) both opine that the MV/LV distribution
system is the least automated, monitored and close-loop controlled
subsystem of the entire electric power system.
4
Fig 1.0 Overview of the Electric Power System
Source:Hammoudeh
(2012)
The communications infrastructure is critical to the smart grid’s
implementation. Hammoudeh (2012), insists that without a robust
communication system in the MV/LV distribution network, only a small part
of the Smart Grid vision could be realized.
As described by (Al-omar2012), (Claudio2010) the Smart Grid
communications network can be divided into the following four domains:
Core or Metro Segment – This segment connects transmission and
distribution substations to the utilities’ headquarters.
Backhaul Segment – Here, data aggregators connectsto the distribution
substation
automation at broadband speeds.
Neighborhood Area Network (NAN) or last-mile – This subnetwork
connects the customer’s smart meter or gateway to the data aggregators.
Home Area Network (HAN) – This represents the customer’s home or
building automation, connecting household smart meters to intelligent
electronic devices (IEDs).
5
According to (Ran,2012),(Laverty, 2010) the power network from the MV/LV
distribution stations to the consumer houses and industries constitutes the
‘last-mile’ connection in the power grid. The scope of this thesis is confined
to the last-mile of the electric power system which is referred to as the Smart
Grid Last Mile(SGLM) in this work, and its corresponding communication
layer.
1.2 AIMS AND OBJECTIVES
In this thesis, the WiMax 802.16 standard is investigated to ascertain its
ability to effectively meet the bandwidth, latency and reliability concerns of
utilities’ smart grid last mile applications.
The objective of this thesis is to:
1. Identify and model SGLM applications traffic, defining their bandwidth,
latency and reliability requirements.
2. Analyse the SGLM WiMax interface through simulations of a traffic model
and evaluation of its performances against established SGLM application
requirements.
3. To propose a WiMax (802.16) interface communication architecture through
simulations in meeting the QoS and reliability concerns of the SGLM.
1.3 RESEARCH METHOD AND DESIGN
OPNET Modeler, a discrete event network simulator is employed in this
work.
I would develop a Smart Grid Last Mile traffic model which is based on a
QoS classification and mapping of traffic classes that occur and generated
by AMI and DAS applications on the last mile performing an applicationbased QoS packet classification. Consequently a SGLM WiMax network
model would be designed and adopted for simulations with the traffic profile
model specified above.
6
Research articles, journals and academic references would be cited in the
body of this thesis. These would be the main sources of secondary data to
be used in establishing the requirements of SGLM communications for
analysis and evaluation of the adopted WiMax SGLM model’s performance.
OPNET Modeler is offered free for academic use by the developers. No
ethical or professional challenge has arisen or is anticipated as a result of
this research work.
1.4 THESIS OUTLINE:
This chapter provides an overview, specifying the rationale, objectives and
research method employed. Chapter 2 of this thesis would review earlier
works and literature on the problem our thesis is trying to address. Chapter
three would define the research method and design.While chapter four
would contain the simulation and presentation of results with discussions
and analysis of the results as obtained from the simulation presented in
chapter 5.
7
CHAPTER 2: SMART GRID LAST MILE COMMUNICATIONS
2.1: INTRODUCTION
Within the electric power system, the power distribution subsystem is the
least automated and poses the greatest challenge in the acheivement of the
smart grid vision.The distribution subsystem consists of transformers,
substations and devices that provide low voltage connections to industrial,
commercial and private customers. This low voltage connection constitutes
what is known as the ‘last mile’ of the electricity network. A major source of
concern within the smart grid literature has been the adequate provisioning
of a communications network in the last mile (Laverty et al 2010), to meet
the latency-intolerant and bandwidth requirements of SG applications such
as Advanced Metering Infrastructure (AMI) and Distributed Automation
Systems(DAS).
Wireless communication technology has been widely considered as a
feasible option for reliable bidirectional networking along the power
distribution subsystem(Clark2010), (Hammoudeh2012),{Gao2012}. The
wireless communications technology to be overlaid ontop of the power
distribution network is required to possess the transmission capacity to
exchange information across the distribution subsystem devices. Though
Zhang opines that many power distribution companies have already utilized
public cellular wireless networks such as (GPRS/CDMA/3G)to monitor the
power distribution devices, issues of adequate QoS and security have
questioned their implementation in very recent times.
NIST’s framework asserts that the capabilities and weaknesses of specific
wireless technologies must be assessed in all possible conditions of Smart
Grid operations due to their differing requirements which limit their suitability
for certain applications. The assertion failed to note that on the flip-side, the
parameters of the SG applications need to be identified and assessed to
determine the suitability of wireless technologies whose parameters have
earlier been investigated and established within the literature.
8
Thus, a thorough review of the literature is required in establishing firstly, the
SGLM applications and then defining their exact data latency, bandwidth and
reliability requirements. According to Ran(2012), there is not much known
about the communicationrequirements for smart grid applications in low
voltage grids which translates to unknown communication requirements of
the smart grid’s last mile.
Hence, the subsequent segments within this chapter has reviewed the
literature to identify the SGLM applications QoS, and reliability requirements.
This it acheives by firstly defining a SG conceptual model as envisioned by
the National Institute of Science and Technology(NIST). Thereafter the
strengths and weaknesses of the identified SGLM communication
technologies was reviewed and analyzed with particular focus on WiMax
whose performance would be investigated in chapter 4 via the use of
simulation vis-a-vis the established requirements for SGLM communications
in this chapter.
2.2 SMART GRID DEFINITION
The smart grid can be defined as the traditional electric power system that
has been enabled with two-way power flow of energy from power generation
plant to customer’s plug and back through the use of advanced sensors,
meters, intelligent electronic devices(IED) and communications
technologies, Lima, 2010).
Therefore, the Smart Grid Last Mile(SGLM) is defined as the smart grid’s low
voltage connection from the distribution subsystem to the customer’s
premises.
In essence, it is expected that the smart grid would provide solutions to
many of the problems encountered by the traditional electric power system.
Its new architecture would enable the bi-directional flow of power, supporting
integrated distributed power generation from renewable energy resources.
9
2.3 SMART GRID REFERENCE MODELS
This section discusses the architecture of the traditional electricity grid briefly
and then goes ahead to introduce and elaborate on the smart grid’s
conceptual model as proposed by the NIST. Lima (2010), acknowledges that
the NIST architecture provides the “big picture/generic approach” which
breaks the system into smaller interoperable and modular components. This
indeed corroborates Hammoudeh’s(2012) assertion that “the first step of
designing a robust communications network is establishing an architecture
that outlines data flow among various parts of the system” and this the NIST
conceptual framework does very well.
2.3.1 Existing Communication Architecture
The existing communication architecture involves control centers,
substations and Supervisory Control and Data Acquisition (SCADA) systems
in a star topology. It has been argued that the communications system never
grew in par with the electric grid. Iyer(2011) calls the present
communication architecture archaic and unable to meet the demands of the
ever growing and complex power grid. According to {Satyendra2012} “There
are many necessities in electrical distribution in the 21st century which
cannot be addressed by the traditional grid, that’s why the modernization in
traditional grid is necessary”.
The above challenges plaguing the conventional electric power grid has
birthed the need for integrating ICT and telecommunications technologies in
automating the electric distribution processes. The modality of integrating
these disparate systems and technologies to ensure scalability and
sustainability has given rise to the NIST conceptual framework.
2.3.2 NIST Interoperability Conceptual Framework
The Energy Independence and Security act of 2007 mandated an agency of
the US department of Commerce, the National Institute of Standard and
technology (NIST) with the coordination and development of a framework of
protocols and standards which will ensure smart grid interoperability. NIST
10
posits that its conceptual reference model drives three major advantages;
being a reference for use cases, identify interfaces for which standards
would be developed and lastly, the development of a cyber security strategy.
The conceptual model presented in this chapter is aimed by the NIST to
direct on-going development work as well as to guide decision making on
how to achieve a functional fit within the modernized electric power
infrastructure. In other words, the NIST framework as insisted by its authors
is descriptive and not presrciptive.
The NIST conceptual framework characterizes the smart grid into seven
domains, these consists of;customers, markets, service providers,
operations, bulk generation, transmission, and distribution. Interdomain
interactions are made possible through actors and applications. Actors are
the devices, systems, or programs that make decisions and exchange
information necessary for use by the applications and they include: smart
meters, solar generators, and control systems while applications on the
other hand, are the tasks performed by one or more actors within a domain.
The Smart grid conceptual model reveals all the communications and
electricity flows which connect each domain and their interrelation. All the
applications in the smart grid generate basically two types of traffic.
Monitoring information traffic, which provides real-time or non real-time
status of devices, components and systems. Secondly, command and
control traffic. As seen in Figure 2.1, the diagram illustrates the conceptual
flow of information in the NIST smart grid vision. Monitoring information is
generated in the customer premises, the transmission, distribution, and the
bulk generation domain{Jeon et al2011}. The information is forwarded to the
management systems in the operations domain. Meter data management
systems(MDMS), Wide Area Monitoring Systems(WAMS) are examples of
management systems in the operators distribution and transmission
subdomain respectively. Command and control traffic originates in the
operations domain and flows through the communication network to the
transmission, distribution, and residential facilities.
11
Fig 2.1 Smart Grid Conceptual flow of Information
2.3.2.1
Bulk Generation Domain
The bulk generation domain is characterized by electricity generation using
renewable or nonrenewable resources like oil, coal, nuclear fission, flowing
water, sunlight, wind, tide, etc. It is responsible for managing the variability of
renewable resources in a way that surplus electricity generated is stored-up
for redistribution at times of resource scarcity. As depicted in fig 2.1 it
communicates with the market domain and the operations domain through a
market services interface gateway actor . Other gateway actors of Plant
control systems and generators communicate with substation LANs and
devices of the transmission domain.It enjoys the most automation and
communication technologies(Hammoudeh2012). Wang (2011) states that
its major functions includes communicating key parameters like generation
capacity and scarcity to the other domains.
12
2.3.2.2
Transmission Domain
The transmission domain transports generated electricity from the
generation stations to the distribution domain. The regional transmission
office maintains the stability of regional transmission lines by balancing
between the demand and supply. This domain also accomodates small
scale energy generation and storage.Self-healing functions and wide area
situational awareness and control is executed in this domain via information
captured from the grid and sent to the control centers. The control centers
will also send responses to the devices in remote substations. The
bidirectional communications between control centers and substations are
handled in the transmission domain too.
2.3.2.3
Distribution Domain
This domain is also known as the SGLM and is responsible for connecting
the smart meters in the customer premises network and all intelligent field
devices, providing management and control through a bi-directional wireless
or wireline communications network. The dispatch of electricity to end users
in the customer domain is implemented by making use of the electrical and
communication infrastructures that connect the transmission and customer
domains. Distribution level storage facilities and distributed energy resources
(DER) are also connected at this domain.
While the original distribution network had no provision for two-way power
flows (Hammoudeh, 2012),the NIST framework provisions for bidirectional
power flows for the integration of DERs. Wang(2011) opines that it is the
responsibility of the distribution domain to deliver electricity to energy
consumers according to the user demands and the energy availability. Wang
further states that the stability of this domain needs to be monitored and
controlled for an overall efficient system thus the SGLM and is the focus of
this work.
13
2.3.2.4 Operations
In this domain NIST describes the efficient and optimal operations of the
transmission and distribution domains. Using SCADA systems, it primarily
relies on field area and wide area networks in the transmission and
distribution domains to obtain information of the power system activities like
monitoring, control, fault management, maintenance, analysis and metering.
As seen in fig 2.1 the operations domain may be further subdivided into subdomains for transmission, distribution, and RTO/ISO operations. These subdomains may be controlled by different organizations.
2.3.2.5
Markets
The market domain strikes the balance between the demand and supply of
electricity. As described by NIST, it is a form of clearinghouse consisting of
both retail and bulk suppliers, traders who buy electricity from suppliers and
sell it to retailers, and aggregators who combine smaller DER resources for
sale. As depicted from fig 2.1 the communication requirements of this
domain is between bulk producers of electricity, and the DERs in order for it
to fulfill its primary role of matching production to demand.
2.3.2.6
Customer
Customers within this domain includes home, commercial or industrial
buildings which consume, generate (using DERs), or store electricity.
Electrically, it is a wired low voltage connection to the distribution domain
and communicates with the distribution, operation, service provider and
market domains. Demand response process which allows customers to
actively participate in the grid, is present here including remote load control.
ESI with a two-way communication interface between the customer
premises and the distribution domain is present at the customer premises.
The customer domain is of significant importance due to the fact that it is the
most visible part of the Smart Grid to the consumer.The communication
network within the customer premises which exchanges data and control
commands between the utility and the smart customer devices is referred to
as a home area network (HAN).
14
2.3.2.7
Services
It communicates with the operation domain to get the metering information
and for situational awareness and system control. Electricity is provided to
customers and utilities
through service providers. It is a domain that caters for services like billing
and customer account management for utility companies.Gomezcuba(2011).
The section above has introduced the smart grid and its conceptual vision as
proposed by NIST. The sections that follow discusses SG technologies and
establishes AMI and DAS as the SGLM applications. Thereafter
applications which make-up the DA systems are discussed extensively, as a
grasp of the knowledge of their operation assists the reader in better
appreciation of WiMax’s desired communication requirements in the SGLM.
2.4
SMART GRID LAST MILE (SGLM) APPLICATIONS AND
FUNCTIONS
In the development of standards to foster interoperability NIST focused its
priorities towards 8 key functionalities for the smart grid with cyber security
and network communications included. These functionalities makeup the
enabling technologies of the entire smart grid(Hammoudeh 2012) and they
include:

Wide area situational awareness

Demand response and consumer energy efficiency

Energy storage

Electric transportation

Advanced Metering Infrastructure(AMI)

Distribution Automation System

Cyber security

Network communications
15
Within the Smart Grid’s distribution subsystem,its enabling technologies and
applications have largely been referred to as Distribution automation and
AMI systems (Gomez-cuba 2011). This is so because, AMI and DAS
provides the infrastructure to integrate the other above listed applications as
would be discussed in the section to follow.
2.4.1 Advanced Metering Infrastructure(AMI)
Automated metering Infrastructure, arguably the greatest enabler of the
smart grid vision provides the means to remotely read customer’s electrical
usage, empower utilities with load control mechanism, integration of DER,
monitoring for electrical faults, and appliance-level reporting. As seen in fig
2.3 AMI provides a critical link between the grid, consumers and their loads,
and generation and storage resources.NETL (2008) describes this
advantage enabled by AMI as a fundamental requirement of the modern
grid.
Like the OSI model, AMI consists of an application layer and transport layer
where, the application layer collects data, analyzes, controls operations and
real-time monitoring. The transport layer is the communication media
responsible for a two-way information transfer between utility and customer.
16
Fig 2.2 AMI network overview
Source:NETL (2008)
Additionally, the AMI isnt just a stand-alone system, it utilizes a combination
of sytems and technologies that connect consumers to system operators
enabling a smarter grid.These components according to (NETL,2008),
(Hammoudeh,2012) includes:
1. Smart meters
2. Customer Premises Network(CPN)
3. Communications Infrastructure
4. Meter Data Management System(MDMS)
2.4.1.1 Smart Meters
Within the hermit of the NIST smart grid vision, thousands of smart meters
act as gateway for data generated within the respective customer premises
networks(CPN), while the AMI communication infrastructure used ensures it
is transported to a meter data management system (MDMS) via the AMI
Head-end as seen in fig 2.2 above.
17
2.4.1.2 Customer Premises Network (CPN)
Based on the consumer’s consumption profile, the CPN network can be
further divided into three different networks namely; Home Area Networks
(HAN), Business Area Networks (BAN) and Industrial Area Networks (IAN).
The NIST conceptual reference model refers to the CPN as being special as
it provides the interface for the customer to the utility. Therefore, a main
challenge is how to build ascalablecommunication architecture to handle the
huge amount ofdata generated by customer’s smart meters, which delineate
the AMI network from the CPN. As would be noted below, CPN in-home
applications can leverage AMI networks, but can optionally also exist
separately from utility-driven systems(DOE 2010).
A combination of wired and wireless technologies are candidates for the
CPN such as PLC, Zigbee, Xbee, Bluetooth, Z-Wave, Wi-Fi, Home Plug, 6lowPAN which possess different requirements from SGLM communication
technologies (Clark 2010).
2.4.1.3 Meter Data Management System(MDMS)
MDMS is a database which collates data for long term storage and analytical
management for the vast quantities of usage data and events. MDMS
interfaces with other smart grid operation management systems, such as
geographic information system(GIS), consumer information system (CIS),
and distribution management systems(DMS) etc. The architecture used in
deploying MDMS would define the scalability of the AMI system(Hu,2012).
2.4.1.4 SGLM Communication Infrastructure
SG communications network are demarcated by the IEEE P2030 standard
into Wide Area Networks, Backhaul, Last Mile, and CPN. The
communication link that overlays the power distribution subsystem
constitutes the last mile network and the backhaul demarcation of data
concentrators as seen in fig 2.3 below.But the scope of this thesis is
concerned with the last mile communication infrastructure.
18
The SGLM communication architecture as depicted in fig 2.3 is made up of
neighborhood and field area networks(NAN/FAN).
a) Neighborhood Area Network (NAN)
The NAN is located in the power distribution subsystem of the entire grid,
connected to the backhaul of the utility and CPN at both ends as seen in fig
2.3 below. The SGLM communication network requires local concentrators
or data aggregation points(DAP)to collect data from groups of meters and
transmits that data to a central MDMS server. The collector deviceor
concentrator in each NAN is connected via backhaul communication
systems to the enterprise AMI head-end system, providing network
connectivity to endpoints such as smart meters and distribution automation
(DA) devices{Cupp 2008}.
Fig 2.3 Smart Grid Communication Networks
source : IEEE 2030
19
b) Field Area Networks(FANs)
The power system devices operating in the distribution domain utilize field
area networks to share and exchange information. Al-omar et al (2012}
enlists some devices whose communications are handled by the FAN as:

Voltage Regulators, re-closers and remotely operable switches.

Capacitors banks and active power factor connection, distance to fault relays
and line fault indicators.

Phasor measurement and distributed RTUs, line sag indicators and
maximum demand indicators.

Renewable energy resources/ generators
According to Weng(2011) FANs add to the reliability of the grid by ensuring
extensive monitoring for the integration of DERs.The monitoring process
involves gathering of information from distribution feeders, transformers
equipped with electrical sensors and communication capability, PEV
charging stations, DER sites and customer premises.
2.4.2 Distribution Automation Systems(DAS)
Distribution automation refers to the control applications of the medium- and
low-voltage part of the power distribution system. Architecturally,the DAS is
made up of a central control system, such as a Distribution Management
System(DMS) working in conjunction with distribution automation field
equipments located in the FAN that can be remotely monitored, and
probably remotely controlled or operated.
Technologies/ devices that comprise DA systems have been identified and
discussed below so as to fully appreciate the requirements of
communication technologies in the SGLM.
2.4.2.1 Dynamic Feeder Reconfiguration
Is the automatic isolation and notification of faults in the power distribution
system and requires close orchestration between the DMS system and
automated feeder / re-closer switching devices deployed inthe distribution
network. Wireless technologies are argued to be well suited for this function
20
as they provide an independent path so that information can reach beyond
the fault.
2.4.2.2 Conservation voltage control and Dynamic VAR control
Within the distribution system authors have mainly identified voltage rise
constraints as the main technical barrier to the connection of Distributed
Generation in present medium and low voltage networks.{Gulich2010}.
Conservation voltage control and dynamic VAR control addresses this
concern by removal of distributed generation voltage via onload tap
changers(remotely reducing the output volt of a transformer while in service)
at HV or MV substations. The communication requirements here would be
critically delay-intolerant.
2.4.3 Distributed Energy Resources
One of the promises of the Smart Grid is better and more uniform integration
of distributed energy resources (at the customer or distribution level) into the
grid, most notably on-grid renewable energy sources. The energy flow thus
will be bi-directional, from utility to home, home to utility requiring a robust
communications technology to attend to the delay-intolerant nature of the
protective and control switching it requires.
2.4.4 Electric Vehicles
EVs present new opportunities in that they offer the potential to function as
an energy storage device, thus playing a unique role in balancing demands
on the Smart Grid.(DOE, 2010)(Wang 2011).The absorption of energy from
the grid by EVs during periods of low demand and reintroduction of
excessive supply back into the grid in periods of high demand is automated
by the DAS.
2.4.5 Demand Response (DR)
The DR system functions by temporarily changing the electricity
consumption by loads on the distribution grid in response to market forces to
maintain the reliability on the grid. Implementation of a DR system is
beneficial for both the utility companies and the customer. The DR systems
21
use the AMI infrastructure and field area networks of the customer domain to
implement their functionalities. (Weng 2011).
2.5
SMART GRID COMMUNICATION REQUIREMENTS
SGLM applications discussed above in section 2.4 are dependent on a
number of parameters whose requirements when met, would ensure that the
smart grid acheives its objectives as envisioned in the NIST conceptual
framework. The detailed communication requirements established in this
section would provide a basis for comparative analysis with results from
simulation in chapter 4 of this thesis.
QoS(latency and bandwidth),and reliability have been the major parameters
adopted within the literature(Castellanos, 2012)(Jeon, 2011) and are also
adopted by this work for investigation as the communication media
requirements of SG applications in the last mile.
The IEEE P2030 framework’s classification of application parameters
according to certain value ranges is at best qualitative and isn’t inclusive of
throughput parameters. As such, quantitative specifications by Department
Of Energy(2010) and the more recent UCA International Users
Group(UCAIug 2013) publications for each SGLM application have been
used in establishing the SGLM communication requirements of applications
discussed in section 2.4, except cited otherwise.
2.5.1 Quality of Service (QoS)
2.5.1.1 Latency
With reference to the smart grid operations, WiMAx Forum(2013) describes
the latency of the communication system as the transfer time for the
complete transmission of a message from one automation device to another.
22
AMI application’s latency is measured as the difference between the
moment instantaneous energy use is measured and when it is reported, and
is adjudged non mission-critical. The different means of DR implementation
has generated differing views with regards to latency requirements of
demand response programs. Honeywell(2010) have argued that the NAN
was not originally designed for load control purposes and will be unable to
provide the latency required for these applications on a broad base.This is
not entirely true as it is the DR capability that basically differentiates AMI
from its predecessor AMR. The least latency tolerant DR implementation
would be the use of instantaneous values for load control purposes and the
DOE recommendation sets it as 500ms. Signals implementing the
integration of DERs are by far the most latency intolerant application within
the entire SG literature(Hammoudeh 2012)(Castellanos 2012), 20ms have
been recommended. A bulletized summary of the DOE latency range
recommendations is provided below .

DER latency-will need to be in the 300 milliseconds to 2 second range.
latency of 20 milliseconds would be needed during faults when protection
devices are switching.

AMI Latency:As with on-premises communications, UTC and Verizon
suggest that the required latency will be in the range of 2 to 15 seconds for
some types of data traffic.

Demand Response Latency Requirements- Estimates of the latency
requirements of DR fall into a wide range, from as little as 500ms to 2
seconds and up to several minutes.

PEV Latency requirements range from 2 seconds to five minutes

DAS Latency Requirements-1 second of latency for alarms and alert
communications and sub-100 milliseconds for messaging between peer-topeer nodes.
2.5.1.2 Bandwidth Requirements.
With regards to AMI, the amount of data being transferred at any instant
would mainly be the instantaneous electricity use of each device which is
23
required to be transmitted from the inhouse premises smart meters to the
collectors/aggregation points. Greater bandwidth demands is envisioned by
DOE (2010) if appliance-level data points as opposed to whole-home data
are transmitted to the aggregation point. AMI for obvious reasons posseses
the highest bandwidth requirement due to the high number of smart meters
compared to other application’s devices. DR bandwidth recommendations
have generally been considered lower than AMI’s requirement but with the
NIST vision of DR systems to work in tandem with AMI, bandwidth
requirements as recommended by DOE would be at least as high as AMIs’.
Other SG application bandwidth requirements have been established to be
lower than AMI, and bulletized below as recommended by the DOE as:

AMI Bandwidth requirementsof 10-100kbps have been regarded for the
AMIapplication

DER Bandwidth Requirements - bandwidth required for DER will be same
as that required for AMI, i.e. 9.6 kbps to 56 kbps.

PEV Bandwidth required for both load balancing and billing purposes will be
between 9.6 kbps and 56 kbps.

DAS Bandwidth requirements will be in the range of 9.6 kbps – 100 kbps,

DR Bandwidth-14 kbps to 100 kbps per node/device,
2.5.2 Interoperability
Interoperability is simply the ability of two or more systems or components
to exchange information and to use the information that has been
exchanged.NIST is the chief coordinator of interoperability activities for the
smart grid, its goal is to ensure integration, effective cooperation, and twoway communications among the many interconnected elements of the smart
grid. Standards are the primary vehicle it uses in driving interoperability.
Jeon (2011) states categorically that what is required is an open
communications architecture that fully support the interoperability based on
universally accepted standards. It is misleading as was deduced from Jeon
24
assertion that it is when standards-based network with the right performance
characteristics, such as QoS, reliability etc are utilized, that new
technologies are easily and seamlessly incorporated onto the network.
2.5.3 Reliability
Reliability is the parameter which describes the ability of a network to
perform within its normal operating parameters and still provide a specified
level of service. WiMax Forum (2013) list reliability criteria in the SG as :

Availability of electricity regardless of the environment.

Asset management ability to monitor, control, and forecast the usage.

The communications network being stable, predictable and robust.
Reliability is often used when measured as a minimum performance rating
over a specified duration of operation. In general, availability for
communication networks reliability values ranges from 99.9% to 99.999%
Hammoudeh(2012). Yan et al (2012)describe demand response
resources,PHEV, and storage resources from reliability viewpoint, saying, an
appreciable amount of load growth can also aggravate the variability of
demand and its associated reliability problems depending on the charging
schemes and consumer behavioural patterns. Reliability figures as deduced
from the DOE(2010) includes:

AMI reliability- will need to be in the 99 percent to 99.99 percent range.

DAS reliability will be 99 percent to 99.999 percent, as these functions are
only required when there is electricity available, backup power is not
required.

Electric Transportation reliability would be in the 99-99.99% range.

DER reliability would be in the 99-99.99% range
25
Applications
Bandwidth
Latency
Reliability Security
Demand
14kbps-
500 ms-
99-
Response
100kbps per
several
99.99%
node/device
minutes
9.6-56 kbps
20 ms-15 sec
Distributed
Energy
99-
High
High
99.99%
Resources and
Storage
Electric
9.6-56 kbps
2 sec- 5 min
Transportation
Distribution
9.6- 100kbps
100 ms-2 sec
Automation
AMI
99-
Relatively
99.99%
High
99-
High
99.999%
10-100kbps
2- 15 sec
per
99-
High
99.99%
node/device
Direct Load
14 kbps to
500 ms-
99-
Control
100 kbps per
several
99.99%
node/device,
minutes
Automatic
2
Feeder
sec/switching
Switching
5
sec/telemetry
Voltage Control
2
sec/operation
5
sec/telemetry
Table 2.0 adopted SGLM applications and their network requirements
source: (DOE 2011)
26
2.5.4 Security
Smart grid security is of extreme importance within the power distribution
subsystem. There isaneed to ensure all customers data remain private and
there is no unauthorized access to the network. Security plagues of
conventional wireless media are no different from wireless communication
technologies in the SGLM. These attacks includes,but are not limited to
spoofing, jamming, man in the middle attack, denial of service (DOS) etc.
Within the literature, techniques involvingMAC and PHY layer user
authentication, access control authorization and data encryption are
employed to ensure network security is not breached. Because a wireless
network uses broadcast medium, it must be resistant to tampering of
messages, preserving confidentiality and integrity of information preventing
unauthorized access.
2.6
SMART GRID LAST MILE(SGLM) COMMUNICATION
TECHNOLOGIES
A number of communication technologies are implemented in the SG.
Though the choice of communication technology should be based on
latency, bandwidth requirement, topology of network, security and reliability
as established earlier, a review of the communications technology is
required in knowing a best match for the established requirements in the
preceeding subsection.
The IEEE P2030 framework as captured by Lima(2010),recommends how
network architectures should be mapped todifferent wireline and wireless
protocols as seen in fig 2.5 below. Staying within the scope of the project’s
objectives, communications technologies mapped to the last-mile /AMI/NAN
of the SG would be discussed further below.
27
Fig 2.4 SGLM communication technologies mapping
Source: Lima (2010)
2.6.1 Wireline Technologies
2.6.1.1 Power Line Communication (PLC)
Power Line Communication(PLC) has traditionally been utilized in the lastmile low voltage section of the grid. DOE (2010) makes a case for PLC and
argues that it is the most common conduit for AMI functions in rural, lowdensity areas where wireless coverage is less available relaying meter data
and other internal communications over a utility’s power lines. Power line
communication(PLC) is used in some AMI but has been said to have
bandwidth and scalability problems.
The bandwidth provided by PLC is not adequate to meet the requirements of
real-time AMI at the per-device level which is about to 100 kbps per
device(DOE 2010).Industry stakeholders contend that traditional PLC and
wireless mesh may well be replaced by broadband communications such as
28
the IEEE 802.16 standard. PLC’s biggest drawback yet is the loss of the
communication link as soon as there is an outage or loss of potential on the
line(Hammoudeh, 2012) . Obviously,it falls short in meeeting LM bandwidth
and reliability requirement.
2.6.1.2 GPON
GPON is a fibre to the premises (FTTP) network technology that uses a
point-to-multipoint scheme to serve multiple buildings and is arguably the
most expensive wireline media available to install. It posseses superior
characteristics like extremely high data rates, long distance reach and
immunity from electromagnetic interference(Hammoudeh 2012). Due to the
large number of smart meters to be deployed, it is cost prohibitive and more
complex installation as compared to wireless technology options.
2.6.1.3 Data Over Cable Service Interface (DOCSIS)
Data Over Cable Service Interface (DOCSIS) is another wireline media type.
Typically used by cable companies to provide video services and in recent
years voice and data. The architecture is a shared resource in a single
neighbourhood network and as such bandwidth is shared amongst devices.
Two fundamental issues affecting DOCSIS have been raised within the
literature and these are: lower data rates during peak usage hours in
densely populated areas especially for those customers located at the end of
the cable route in a neighborhood and security concerns for customer’s
confidential data since the main cable is a shared medium in the last mile
segment(Hammoudeh 2012).
2.6.2 SGLM Wireless Communications Technologies
Ease and speed of deployment has been the major advantage of wireless
technologies over wired solutions in the literature.The NAN communications
requirements of QoS, reliability security and interoperability were discussed
in the earlier segment. This segment reviews the literature to discuss SGLM
wireless solutions and advances both WiMAX and 3G/4G technologies as
being most favourable for LM communications with WiMAx having the
advantage due to its rich QoS features.
29
2.6.2.1
RF Mesh-802.15.4
Radio frequency (RF) mesh grids can provide a flexible, reliable and secure
network that can span a considerably large area through hopping. Its mesh
architecture ensures the network is self-forming and self-healing. For the
last-mile/NAN, Mesh networks are not considered a practical option for
reasons of uncertainty, due to the some-what proprietary nature of the
current solutions and the lack of access to run these networks in licensed
spectrum allocations. {VanderDrift2011}{Clark2010).
2.6.2.2
3G/4G LTE/ GPRS/EDGE/HSDPA
In obtaining data communications coverage quickly and inexpensively over a
large geographical area, Kirkham(2010) recommends 3G/4G Networks.
WiMax is said to possess significant bandwidth and latency advantage over
3G cellular networks, but cellular network’s 4G LTE evolution is believed
would make the WiMax advantage short-lived.
Cellular Networks including 3G/4G LTE/ GPRS/EDGE/HSDPA though well
proven and stable, with regards to the SG last mile, Pavlovski et al (2010)
iterated application aware issues to be considered regarding its use as:

High quality of service guarantees to ensure that remote sensing and control
data is given higher priority over general voice and data traffic.

Households require 100 per cent coverage, and while urban areas are
generally covered by these networks rural areas often pose a difficulty.

Most cells are configured for a limited number of cell connections, which
may be insufficient for large-scale household deployments. In some cases, a
base station may need to cater for up to 10,000devices.

A high degree of security to prevent malicious attack, including encryption,
mutual authentication, and data integrity is required but not guaranteed.
30

Response time is critical, from idle to active, granting the ability to establish
connection with the network and transmit data rapidly.
2.6.2.3
WLAN-802.11n/g
IEEE 802.11 wireless local area networks (WLAN) are the second most
successful and widely deployed wireless data networks in the world (second
to only cellular networks) (Kirkham et al, 2010).802.11g variant operates in
the 2.4GHz frequency band, offering support up to 54Mbps. Its major
disadvantages lies in interference with appliances on the unregulated signal
frequency.
2.6.2.4
Zigbee-802.15.4
802.15.4/Zigbee operates in the unlicensed Industrial Scientific and
Medicine band of 2.4 GHz and has been realized as the most suitable
communication standards for Smart Grid residential network domain by
NIST (Hammoudeh 2012). However, they are not suitable for deployment in
the Distribution Network due to short reach and serious security
issues.(Kirkham 2010).
2.6.2.5
Smart Utility Network-802.15.4g
It is an outdoor short-range networking standard , a PHY layer ammendment
to the IEEE 802.15.4. The standard will be for outdoor long range leaps
within the 902-928 MHz band. It was designed to facilitate very large scale
control applications, especially the smart grid. For use within the SGLM, its
major disadvantage is that, its yet to be accepted as a standard,(Maxim
2010).
2.6.2.6
WiMax- 802.16d/e
Worldwide Interoperability for Microwave Access is an Institute of Electrical
and Electronics Engineers (IEEE) standard which is identified by the
designation of 802.16-2004 for fixed wireless applications and 802.16e-2005
standard focusing on mobile broadband access. Subscriber stations within
the context of the SG are stationary smart meters, the 802.16d.
31
Castellanos(2012) argues that amongst all the features of the WiMAX
standard, its strong support for multiple QoS classes makes it particularly
suited for SGLM communications as its enabled in its PHY and MAC layers
described below.
a. WiMax MAC and PHY Layers
The 802.16 standard consists of MAC and PHY layers, whose functions
make it particularly suited in meetinq QoS demands of SGLM applications.
The MAC portion consists of three sublayers as shown in fig 2.5 :
Convergence sublayer(CS), the common part sublayer(CPS) and security
sublayer. The primary functions of system access, bandwidth allocation,
connection establishment and connection maintenance rests on the MAC
CPS receiving data through the MACSAP from the various CSs already
classified to particular MAC connections. QoS is applied to the transmission
and scheduling of data over the PHY.
Scheduling is done both ways in the standard: the usage of the air link
among the SSs is scheduled and scheduling of individual packets at the BSs
and SSs.The bandwidth scheduler determines the bandwidth requirements
of the individual SS based on the service classes of the connections and on
the status of the traffic queues at the BS and SS.
The PHY layer is responsible for controlling when and how the wireless
shared channel is accessed. Its technique for radio access employs
Orthogonal Frequency Division Medium Access (OFDMA) which divides
radio resources into both time and frequency slots. In Time Division
Duplexing(TDD) mode, a frame is divided into downlink (DL) and uplink(UL)
slots. In Frequency Division Duplexing mode, there is a channel used for
downlink and a second paired channel used for uplink communications.
32
Fig 2.5:WiMaX MAC and PHY Layer Architecture
The MAC layer also contains a separate security sub-layer providing
authentication, secure key exchange, and encryption.
b. WiMaX QoS
WiMax supports five types of service (TOS). These TOS employ resources
based on the traffic classification and QoS.These classes of service
includes:

Unsolicited Grant Services (UGS): is used to minimize resource
assignment overhead for constant bit rate (CBR) applications.Subscriber
stations using UGS will make a single request and geta reserved allocation
for the duration of the connection. The BS is aware of the rate and
periodically assigns resources without control messages.
33

Real-Time Polling Services (rtPS): Real-Time Polling Services supports
time-sensitive, variable bit rate traffic such as compressed video and voice
communications.

Non-Real-Time Polling Services (nrtPS): Non-Real-Time Polling Services
is designed to support services that require variable size data allocation on a
regular basis.

Best Effort (BE) Services: BE services are used for internet traffic.

Extended real-time variable rate service (ertPS): This service is defined in
IEEE 802.16e-2005 and is suitable for applications that have variable data
rates but still require minimum guaranteed data rate.
2.7
SUMMARY
This chapter basically introduces the reader to the communication
challenges in the last mile of the smart grid within the ambits of the
conceptual model defined by NIST.The conceptual model discussed divides
the SG as envisioned by the NIST into seven important domains of
generation, transmission, distribution, customers, markets, services and
operations. The model breaks the system into smaller interoperable and
modular components describing their interdomain electricity and
communications flows and requirements.
With the interdomain interactions of the conceptual model,communication
requirements of the last mile have been defined as being very critical to the
overall success of the envisioned model of the SG and wireless technologies
with adequate QoS capability and robust security features were
subsequently established the most suitable in meeting those requirements.
Finally this chapter presents wireless comunication technologies reviewed in
the light of the above quantified requirements, but with a focus on the WiMax
802.16d standard in order to determine its suitability as a SGLM
communication technology.
34
CHAPTER 3: RESEARCH METHOD AND DESIGN
3.1 INTRODUCTION
SGLM applications of AMI and DA identified in the preceeding chapter had
their communications requirements defined. Hence, to evaluate the
performance of WiMax in meeting the SGLM requirements,this work has
adopted the use of literature review in the preceeding chapter,traffic and
network modelling methodology established in this chapter foruse in
discrete event simulator, OPNET Modeler in the next, in meeting the
research objectives.
Simulation is a research method which is widely utilized in modelling large
systems for performance or efficiency evaluation purposes. It is arguably
the most common quantitative modelling technique used and the methods
are either event based or continous state. A discrete event simulator,
OPNET Modeler, has been employed. My first port of call in this chapter
would be to appraise simulation as a research tool and thereafter formulate
the traffic and network models to be utilized in the simulations for evaluating
the performance of WiMax.
The modelling methodology dictates the success of the discrete event
simulator as a problem solving technique. Hence in this chapter, a traffic
modelbased on a further classification of the SGLMapplication traffic
established in chapter 2 has been advanced. These were mapped to service
classes and scheduling algorithms which also utilize the IP ToS field in the
IP header to prioritize traffic.Shah et al agrees it is essential to classify the
heterogeneous traffic in order to support data delivery according to the
priority of each traffic class.
Two challenges arose in the course of developing the traffic model bordering
on the inability to classify mission critical application to the UGS scheduling
type due to their non-periodic and variable sized nature and a theoretical
35
limitation due to delays on rtPS traffic. The adopted SS scheduling algorithm
of the proposed model addresses these challenges through the use of
deadline timestamps.
Once the traffic model was established, the network model is advanced
which describes the radius of coverage, number of smart meter/ESI nodes
to be adopted for simulation.Based on the IEEE P2030 framework of
characteristics discussed in chapter 2 where each application could be
classified according to certain values range or qualitative consideration.
Qualitative reliability classification has been mapped to the above service
classes for evaluation purposes.Low reliability is allocated to the non real
time service class applications, medium reliability requirements for price
signals and meteringapplications in the soft real-time service class and very
high for alarm command ,alarm control, and network joining traffic in the
mission critical service class.
Lastly parameters to be analyzed across each service class to be used in
the performance evaluation of the WiMax interface were outlined.
3.2 SIMULATION AS A RESEARCH TOOL
Within the telecommunications industry and across others such as medical,
military, aeronautic etc, simulation has been employed as a problem solving
technique for many of the research problems and questions as they have
arisen over the years. According to (Page,1994) simulation is described as
the use of a mathematical/logical model as an experimental vehicle to
answer questions about a referent system. This section will provide reasons
for discrete event simulation as our choice of evaluating the performance of
WiMax and why OPNET Modeler in particular.
The choice of a research method should basically depend on how closely
such a method answers or addresses the problems or research questions
raised. When real life data and conditions are correctlymodelled into a
simulator, the results provides answers that can be used in basing future
36
decisions. Arriving at the correct decision is thus the highest goal of any
simulation model.
The challengelies in ensuring the correct decision is arrived at. Five factors
have been advanced as investigated previously within the literature by
(Ziegler 1976), (Page, 1994) that ensures the validity and effectiveness of
simulation as a research tool.

Firstly, as was succintly noted earlier, the efficacy of the simulation result is
very dependent on the modelling methodology. The modelling methodology
in essence is defined by a theory of modelsand the relationship of the
processes to supporting the decisions made .Errors made during modelling
could lead to the acceptance of results based on an invalid model. This can
be catastrophic when significant investments are involved.

For proper modelling of the problem to be made its outright understanding
cannot be overemphasized. When a research problem cannot be accurately
and articulately defined it is common sense little hope exists for the desired
answers to be provided.

The need for the models discussed above to be precise resurfaces here.
The simulator programs are designed with these models , thus errors are
transferred into the programs when done incorrectly.

The right questions must be enquired of the programs to acquire the right
answers.

Lastly, interpretation of the results by the researcher dictates the efficiency
of the simulation as a research tool. It is to his advantage, the researcher,
who possess good statistical skills which would lend themselves to the
accurate interpretation of the simulation results.
Traffic models are used in two fundamental ways: (1) as part of an analytical
model or (2) to drive a Discrete Event Simulation (DES), OPNET Modeler
adopted for the cause of this work is a DES. It has been adopted for the
cause of this work for a number of reasons and these include:
37
1. The availability of correctly modelled traffic to be inputed into the program,
makes the results valid for performance evaluation and decision making.
2. It is the most versatile methodologyfortesting scheduling algorithms to
actually simulate the designed algorithm with real life data and conditions.
3. Other programming languages such as C, C++, Java, MATLAB and many
othersthough strong and featuring rich languages,do not come with a model
of a specific system. But OPNET does and has been adopted for the sake of
accuracy.
4. OPNET Modeler has been researched to be one of the best candidates for
Wimax system model simulation (Kaur et al, 2011). It is a Discrete Event
Simulation(DES) program where traffic processes are characterized by
slotted time intervals and events are handled in a chronological manner.
5. A friendly GUI which adopts a drag and drop approach which simplifies
selection and configuration of objects on its pallete.
The use of discrete event simulator,OPNET modeler, has been informed
mainly by its accuracy, simplicity and wide acceptance within the industry.
(Castellanos 2012), (Iyer, 2011) and (Leclerc 2010) utilize traffic models in
simulations to analyze the performance of communication networks.
Secondary data from the DOE, WiMax forun and UCaig has been employed
in conducting WiMax’s performance evaluation in SGLM. These sources are
at the forefront of advancing and standardizing smart grid communications
and thus possess credible and latest information which is critical in
conducting a good performance evaluation.
In the light of the above factors, the next section is dedicated to the accurate
modelling of the traffic based on its classification, assignment to service
classes and flows, BS and SS scheduling and mapping to the IP ToS for
prioritization, which constitute the basic techniques for QoS implementation
in the network.
38
3.3 SMART GRID LAST MILE TRAFFIC MODEL
The SGLM traffic though originally designed for the conveyance of metering
data has evolved beyond the scope of its initial purpose. Safe and optimal
integration of premises and distributed generation and storage facilities as
envisioned by the NIST and also demand response load control systems
have been facilitated by smart grid last mile communications networks.
Having a system with traffic of differing importance and characteristics, traffic
prioritization must be employed for QoS to be implemented on the WiMax
communications link. It is established prior to now that by prioritizing traffic,
the lowest possible latency is ensured for mission-critical traffic.
Adas et al,(1997) posits that in order to evaluate wireless communications
technologies based on goodness-of-fit metrics, parameters should be
defined that allow for quantifying how close the traffic model is to the actual
data.The parameters of latency, data rate and reliability have been chosen
as the basis of priority classification.
For the purpose of this work and the development of the traffic model, a
preclassification ofSGLM traffic established in the preceeding chapterbased
on their generic naturehas been proposed in the following manner :

AMI’s Metering data: generated at the ESI, its traffic flows towards the
DAP, reporting energy usage.

Price signals: generated at the DAP and flows towards the ESI, reporting
variable energy prices.

Firmware updates : would flow in the direction from the DAP to the smart
meter.
Price signals represent demand response schemes which actuate load
control automatically or designed to cajole the customers to cooperate
based on certain price incentives.
39
DA applications traffic have been established in the preceeding chapter as
being mission critical due to their latency intolerant nature, these
includescontrol and protection signals for DER integration, capacitor
banks,feeder reconfiguration, voltage and dynamic VARcontrol. These traffic
are response-type in nature and have been further classified generically
into:.

Alarm signals- these originate from the ESI flowing towards theDAP

Alarm commands- traffic fromDAP towards the ESI.

Network joining-session initiation requests sent by ESI to DAP when they
want to join the network
Network joining messages are session initiation messages relayed by the
ESI/smart meter when they wish to connect to the grid.(Gomez-cuba 2011),
describes these as mission critical with a low latency requirement, as their
delay would consequently delay other grid tasks, regardless of whether they
have high or low latency requirements.
The next step in the development of the traffic model is to ensure each
application and its associated traffic flows are assigned to a unique service
class. Service classes are a traffic management method for grouping similar
traffic types e.g streaming video, voice, email, file transfer etc which is based
on the scheduling techniques/WiMax QoS types captured in section 2.5
each assigned with their level of priority.
For the purpose of simplicity,three service classes of mission critical, soft
real-time and non real-time traffic have been identified and used in this
model, therafter mapped accordingly to above preclassified applications
traffic.

Mission critical -alarms and alarm response messages raised by users and
providers respectively and network joining traffic.)
40

Soft real time traffic (soft real-time interactive maintenance commands,
periodic meter readings and other measurements, and the dissemination of
energy pricing and other policies.

Non-real time traffic:( these includes firmware update traffic, which is
envisioned to originate from the DAP).
Applications
Service Class
Alarming signals
Mission Critical
Alarming Commands
Network Joining
Price Signals
Soft real time
Metering Data
Firmware updates
Non Real time
Table 3.1 Traffic flows mapped to service classes
The next step in the development of the traffic model lies in the association
of the created service classes to the scheduling algorithm’s service flows
which guarantees differentiated traffic flow and provides different levels of
QoS for each traffic class.
Scheduling types are used by the SS in defining service flows with uplink
and downlink QoS parameters. A Service Flow (SF) is a unidirectional traffic
connection offeringspecific QoS levels described by the application traffic
attached to it. Section 2.5 of this work specifies all service flows available to
the WiMax ESI/SS. For the sake of the proposed model UGS, rtPS and
nrtPS have been adopted and mapped to the service classes above as
described below and illustrated in table 3.2.
41
Applications
Alarm Commands,
Service Class
Scheduling Type
Mission critical
rtPS
Soft real time
UGS
Firmware Updates
Non real time
nrtPS
Rest(0)
Default
Best effort(BE)
Alarm Control,
Network Joining
Metering Data,
Price Signals
Table 3.2 scheduling type mapping to smart grid applications
Alarm traffic(signal and command) of the mission critical service class are of
an irregular nature andwouldposses long periods of silence followed by
periods of activity. Due to the fact that this traffic needs to be processed in
real-time, the scheduling algorithm rt-PS is selected as the scheduling
type.Metering data and demand response signals traffic of the soft real-time
service class, though of less criticality than alarm trafficare often fixed sized
and periodic transmissions in nature and are thus alloted the UGS
scheduling algorithm which doesnt use priority bits. Non real time and
default service class traffic are allotted the nrtPS and BE scheduling types
respectively.
Mandatory QoS parameters exists but vary for each scheduling scheme.
These parameters are shown in table 3.4 and include: maximum sustained
traffic rate (MSTR), minimum reserved traffic rate (MRTR), Maximum
latency, Tolerated Jitter, and Traffic priority and interpolling time.
42
Maximum Sustained traffic
UGS
rtPS
rtPS nrtPS
BE
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
rate
Minimum Reserved Traffic
Rate
Maximum Latency
Y
Tolerated Jitter
Y
Traffic priority
Y
Y
Table 3.3 mandatory QoS parameters for each scheduling service.
For the UGS class, the mandatory parameters are packet latency, jitter and
MSTR which is equal to the MRTR, due to its constant bit rate requirement.
rtPS class is defined by the combination of delay, MRTR and MSTR and
traffic priority. Delay in this class is the base station’s processing time. The
last class in our model nrtPS has mandatory parameters of MSTR, MRTR
and traffic priority.
These QoS parameter values are not specified by the 802.16 standard but
are assumed in this work for the sake of the performance evaluation.
In addition to alarm traffic being assigned the rtPS class, a theoretical
limitation persists for rtPS of alarm traffic in the proposed model.The BS is
required to store the above flow information and QoS parameters which
handle traffic between packets. Considering the large smart metering
population, each BS is required to permanently store large KB of information
that is infrequently used and this does not constitute a practical solution to
be adopted by vendors. Consequently with rtPS alarm traffic being bypassed by the scheduler, DSA dynamic service allocation is requested via a
DSA Request (DSA-REQ) control packet while it waits for a response. This
delay caused by waitingis envisioned to be the most restraining factor on
timely delivery of alarm commands and signals.
Elazouzi et al. propose that due to constraints reflecting in general the QoS
requirements of different service classes, the scheduling solution for the
43
above problem should be modelled as an optimization problem whose
objective is to maximize the system performance.
For this work, the proposed SS scheduler for this model would utilize a
simple fixed priority scheme—1, 2, 3 and 4 for BE, nrtPS, rtPS and UGS
scheduling services respectively. Bandwidth is firstly guaranteed for UGS
connections, while rtPS packets possess deadline timestamps with which
they would be scheduled—deadline timestamps are a function of arrival time
+ tolerated delay (which addresses the delay problem indicated above).
Each nrtPS packet is associated with a virtual time calculated to guarantee
the minimum reserved bandwidth and hence maintain an acceptable
throughput. A simple FIFO mechanism is adopted for BE queues.
Table 3.3 shows the ToS marking assigned to each packet and the
associated service class mapping.
IP ToS
Service Class
Scheduling
Type/Service Flows
Excellent effort(2)
Non real time
nrtPS
Streaming media(4)
Soft real time
UGS
Interactive media(3)
Mission critical
rtPS
Rest(1)
Default
Best effort(BE)
Table 3.4 IP ToS mapping to scheduling type
When packets don’t match with any of the IP ToS classes tabularized above,
then the BE allocation is used, which for the sake of this work is the default
service class. Any traffic not directly assigned to any service class is sent
using BE service.
Service flows are responsible for selecting the modulation and coding
scheme (MCS) values, the size of the queue, the average service data unit
(SDU) or the packet size, the idle timer applicable for nrtPS connections and
the retransmission methods like automatic repeat request (ARQ).
44
The QPSK ½ rate coder is the MCS adopted and employed across all
service classes due to its characteristic as the strongest modulation and
coding scheme to support long distance transmissions. The average value of
packets generated by the application layer, its packet size is set to 1Kb. The
buffer size of 128Kbytes is configured, this permits at least up to 128
connections within the same frame.
Retransmission parameter in non real time and default service classes are
set to zero due to their latency tolerant characteristic. ARQ is used for
mission critical and soft real time service classes, which reacts to losses by
retransmitting the PDUs.
ARQ is used for FTP packets but not for UDP. Hence the transport layer
protocol configured for the mission critical and soft real time service classes
is FTP while UDP is used for non real-time and default service class.
45
Applicatio Service
Scheduli
n
ng Type
Class
IP(ToS)
MCS
Averag
Buffer
Retransmi
e SDU
size
ssion
parameter
Alarm
Mission
rtPS
Interactiv
QPSK
Command critical
e
½
s,
media(3)
Rate
Alarm
1Kbyte
128kbyt
ARQ
e
coder
Control,
Network
Joining
Metering
Soft-
Data,
Price
UGS
Streamin
QPSK
real
g
½
time
media(4)
Rate
Signals
1Kbyte
128kbyt
ARQ
e
coder
Firmware
Non-
Updates
real
nrtPS
Excellent
QPSK
effort(2)
½
time
1Kbyte
128kbyt
None
e
Rate
coder
Others
Default
BE
Rest(1)
QPSK
1Kbyte
½
128kbyt
e
Rate
coder
Table 3.5 :service class configuration for smart grid application
3.4 NETWORK MODEL
The WiMAX Network model adopted for this work is designedto provide
coverage to a target service area containing a number of ESI/Smart Meters(
Nk) and DA devices.The proposed model considers a single-cell,point to
multipoint IEEE 802.16 system using the OFDMA - TDD operating mode
with a cell radius of 2 Km.
46
None
An incremental sampling of 50, 100 and 200 randomly distributed ESI nodes
would be used for the purpose of analysis, as this would enable us know
how node density affects wiMax in meeting the SGLM requirements.Each
ESI would communicate six application flows which were earlier specified in
our traffic model with the DAP via its WiMAX interface. As seen in fig 3.1
application-based packet classification is used with three levels of priority
considered.
ESI/
DA
sac
p=3 rtPS
das
DAP
N1
dac
N2
Sas
N0
Snj
N3
802.16
802.16
Smd
p=4,UGS
dnj
dmd
dps
p=4,UGS
sps
Nk
p=2,nrtPS
p=3 rtPS
dfu
p=2,nrtPS
Sfu
ac-alarm commands
as-alarm signal
nj-networj joining
md-metering data
ps-price signals
fu-firmware updates
Fig 3.1 Architecture of proposed WiMax network
Fig 3.1 above depicts the architecture of the proposed model.Application
sources and destinations are denoted as si and di for i ∈[ac,as,nj,md,ps,fu]
and p denoting the value of the precedence field of IP packets for each type
of application.
47
Based on the adopted model for this work, alarm commands and signals
provokes a response which depicts a two way flow of information. This same
response can be said of the network joining function. Thus, for the mission
critical service class, two flows of information are studied for each of them. In
these flows, reliability and delay are the main parameters of interest and are
studied due to their periodic and fixed packet size characteristic.
Soft real-time notifications of metering data and price signal, require just
single flow of packets per application hence do not trigger a response. In this
case, reliability, delay, and throughput are the interests of evaluation and
would be the main focus of the investigation.
Finally, the analysis of the non-real-time component of the application set,
which corresponds to firmwareupdates, focuses on throughput, since latency
and delivery are irrelevant for a non-real-time reliable application.
Data Characteristic
Classification of Application
Mission
Soft real-time
critical
Non real-
Default
time
Information Quality
Critical
Important
Important
Low
Reliability
Availability
High
Medium
Medium
Low
Impact
Critical
Severe
Limited
Llimited
Table 3.6: mapped reliability parameters to service classes.
Reliability requirements for these applications service classes are variable.
Relaxed reliability requirements was allocated to non real time traffic,
medium requirements for soft real time application and very high for mission
critical applications. The availability for the non real time traffic is considered
medium and high for the mission critical class. In addition, the impact of the
information is limited for default, andnon real time, serious in the case of the
soft real time and catastrophic for the mission critical application.
48
3.5 SUMMARY
A traffic model has been developed in this chapter with the use of service
classes as a means of providing QoS differentiation in a point to multipoint
WiMax network. A classification based on the function of the SG applications
was made which paved the way for their assignment to service classes. The
misssion critical service class made up of alarm commands and signals are
not periodic and could not be assigned the UGS class as such they were
assigned to rtPS scheduling class, with the periodic soft real time class
assigned UGS. Firmware updates and other traffic classes assigned nrtPS
and BE scheduling respectively.
A network model was also specified for our model with defined MAC layer
parameters.
Quantitative results are expected from the simulation which would form the
basis of comparatively analyzing the performance of the WiMax SGLM
network in meeting the SGLM requirements established in chapter 2.
The next chapter would chiefly investigate via simulation if the SS scheduler
which designates bandwidth per connection, can effectively reduce the QoS
violation rate of rtPS service flow since UGS is expected to have negligible
delay and constant throughput due to its fixed size and periodic nature.
49
CHAPTER 4: SIMULATIONS AND RESULTS
4.1 INTRODUCTION
This chapter provides the performance of the smart grid last mile WiMax
networkusing OPNET Modeler,an object oriented discrete event simulator.
Here thetraffic and network models developed in chapter 3 is augmented
with transmission and base station OPNET configuration parameters to fully
arrive at the simulation model employed for the cause of evaluating the
WiMax network’s performance in the SG last mile.
Results from the simulation is presented in section 4.2 with performance
metrics of delay, throughput and reliability recorded via the simulation used
comparatively with the requirement’s recommendations from chapter 2 to
evaluate WiMax’s performance in the smart grid’s last mile.
4.2 SIMULATION MODEL
The WiMax model for simulation was built with one base station (BS), and
randomly distributed substation nodes of 50, 100 and 200 were simulated
within a 60 minute time duration in evaluating via node density, the effect on
delay, throughput and reliability requirements of WiMax in the SGLM.
The WiMax bandwidth employedin the simulation model is 20MHz in the
band of 2.5GHz.It is divided into 2048 subcarriers based in a 2048-fast
fourier transform (FFT) operation, which are further grouped into 24 subchannels for the DL and 70 sub-channels for the UL.
OPNET Modeler’s WiMax configuration mode was used to store profiles of
PHY and service class, which was referenced by all WiMax nodes in the
network. Service class parameters earlier decribed in chapter 3 is captured
in table 4.0(b) while transmission and base station configuration parameters
captured in the simulator and described above are shown in table 4.0a
below.
50
Base Station
Antenna gain= 15dBi
Parameters
Antenna type= omni
PHY Profile= wireless OFDMA 20MHz
PHY type= OFDM
Transmission Duplex Mode= TDD
Parameters
CH Bandwidth = 20MHz
CH Frequency = 5GHz
Max transmission power(w)= 0.5
Frame Duration=5ms
Table 4.0a Simulation Parameters
UGS
rtPS
nrtPS
10000
5000
5000
Minimum Reserved Traffic Rate 10000
5000
5000
Maximum Sustained traffic rate
(bps)
(bps)
Maximum Latency(msec)
1000
1000
Table 4.0b QoS Service Classes Configuration Parameters
The following traffic generation schemes were employed via FTP:
Smd = metering data transmission at periodic rate was used. 1024 B of data
from ESI to DAP. as stressed within the literature Smd =1/60 s-1.
Spc = is the same as metering data, but in the opposite direction.
Sac = this traffic corresponds to a single packet of 1024 B from ESI to the
DAP. Gomez posits its rate is 20% of metering traffic. = 0.2(1/60) = 1/300 s 1.
Sas = is the same as alarm commands.
The simulation was done using the simple scheduling algorithm adopted in
chapter 3 with it’s priority queing mechanism. The performance metrics used
51
in the evaluation of the smart grid last mile (SGLM) WiMax network are the
delay, throughput and reliability.
4.3 PRESENTATION OF RESULTS
This section provides a presentation of the results obtained for the
performance evaluation of the WiMax SGLM through the metrics of latency,
throughput and reliability.
For the purpose of simplicity, the performance of WiMax in the SGLM when
simulated with the SG traffic in the UL and DL are presented based on their
corresponding service classification. Table 4.1 below provides a
representation of the mappings to be used in the presentation of the
simulation results.
Service Classes
Mission Critical
Soft real time
Non-real time
UL Traffic
Alarm Signals,
Metering Data
Nil
Price Signals
Firmware update
Network Joining
DL Traffic
Alarm
Commands
Fig 4.1: UL and DL traffic service classification
Best effort (BE) traffic has not been included in this result presentation and
analysis as the adopted model is based on SGLM WiMax network that is
owned by the customer, hence only SG traffic is conveyed across the WiMax
SGLM network.
4.3.1 Latency Results Presentation
Uplink (UL) and downlink(DL) packet delay results for three simulation
scenarios with node densities of 50, 100 and 200 smart meters was
conducted and their results presented here. The simulation is conducted in a
mixed traffic scenario as established in the traffic generation profiles in
section 4.1.
52
4.3.1.1 UL Latency Result Presentation
UL delay in the soft real-time service class increases marginally with
increasing number of nodes. In comparison withthe delay of the mission
critical service class, results show thatthe soft real-time service class delay
rise is far less and of a lower percentage as captured in table 4.2 below.
Sevice
Affected
class
Application
Mission
Alarm
Critical
signals,
(msec)
Network
50 nodes
100 nodes
200 nodes
16.53
18.68
19.03
6.05
6.12
6.15
5.60
5.63
5.66
Joining
Soft real
Metering
time
data
(msec)
non real
Firmware
time
updates
(msec)
Table 4.2: UL average delay results for SGLM traffic service classes
Interestingly, the non real time service class from the results exhibits a fixed
and low latency in the UL. As observed from the results also graphically
represented in fig 4.2 below, the soft real time and non-real time service
classes of less criticality possessessignificantly lower latencies than alarm
signals and network joining traffic of the mission critical service class.
53
mSec
20
18
16
14
12
50 nodes
10
100 nodes
8
200 nodes
6
4
2
0
soft real time
mission critical
non real time
Service classes
Fig 4.2: UL average delay results for SGLM traffic service classes
4.3.1.2 DL Latency Result Presentation
The DL delay across service classes is lower than the UL delay except for
applications in the non-real time service class. The increment in delay with
increasing number of nodes for applications in the DL is minimal for both the
mission critical and soft real-time service classes.
Sevice
50 nodes
100 nodes
200 nodes
9.53
9.68
10.03
3.05
3.12
3.15
305.60
425.60
500.55
class
Mission
Critical
(msec)
Soft real
time
(msec)
non real
time
(msec)
Table 4.3: Average delay for DL SGLM traffic service classes
The non real time service class has the highest latency values of all service
classes in the DL and also exhibits the highest percentage of latency
54
increase with increasing number of nodes which is in wide contrast to its
behavior in the UL.
Kbps
600
500
400
50 nodes
300
100 nodes
200
200 nodes
100
0
soft real time mission critical non real time
Fig 4.3: Average delay for DL SGLM traffic service classes
Service classes
4.3.2 Throughput Simulation Results
In this section, UL and DL throughput of the varied applications as per their
service classes , with simulations of 50, 100 and 200 SG nodes is
presented.
4.3.2.1 DL Throughput Result Presentation
DL throughput results from the simulations indicate that applications of the
soft real time service class and the non real time service class experience a
near constant throughput with increasing number of nodes. Mission critical
service class experiences minimal degradation of throughput with increasing
nodes.The non real time service class, recorded the lowest throughput
amongst the traffic classes while the soft real time service class experienced
the highest values of troughput in the DL as seen captured in the table and
fig 4.4 below.
55
Service
50 nodes
100
200
Classes
(kbps)
nodes(kbps) nodes(kbps)
Critical
185
164
143
Soft real
252
250
249
60
58
55
Mission
time
non real
time
Table4.4: Throughput in the DL for SGLM service classes
Kbps
300
250
200
50 nodes
150
100 nodes
200 nodes
100
50
0
soft real time mission critical non real time
Service classes
Fig 4.4:Throughput in the DL of SGLM service classes
4.3.2.2 UL Throughput result presentation
In the UL direction, the throughput degradation with increasing number of
nodes remained almost constant with a little decrease for the soft real time
and non real time service classes with the mission critical class experiencing
the highest degradation. Mission critical and soft real time service classes
throughput in the DL hadfar lower throughput values than recorded in the UL
direction. Non real time service class traffic exhibited a rather surprising
high throughput recorded in the UL direction.
56
UL
50 nodes
100
200
Throughput (kbps)
nodes(kbps) nodes(kbps)
Mission
78
62
58
125
123
120
350
345
341
Critical
Soft real
time
Non real
time
Table4.5:UL Throughput of SGLM application
The results illustrated aboveand below for the DL and UL throughput
respectively, do not suggest that changes in throughput for different
numbers of nodes are related to the direction of traffic flow.
Service classes
Kbps
400
350
300
250
50 nodes
200
100 nodes
150
200 nodes
100
50
0
soft real time
mission critical
non real time
Fig 4.5:Throughput in the UL of SGLM service classes
The throughput in the polling of the rtPS service flows for increasing number
of nodes, for 50, 100 and 200 changes negligbly.
4.3.3 Reliability Results Presentation
The packet delivery ratio is used as the reliability metric and is computed as
the total number of packets sent out to the number of packets received.
OPNET simulator provides for the computation from within its WiMax
57
efficiency mode parameter, with the framing mode enabled. Soft real time
service class traffic petition delivery ratio exhibited the least percentage
packet loss.
In all service classes, results show a significant decrease in the packet
delivery ratio with increasing number of nodes, enough to assert that node
density has a considerable effect on the packet delivery success. Soft real
time service class exhibited the least packet loss with increasing nodes while
firmware update of the non real time service class has the lowest packet
delivery ratio with the highest number of nodes in the simulation.
Service
50 nodes
100
200
classes
(%)
nodes(%)
nodes(%)
99.999
99.998
99.987
99.999
99.989
99.899
99.899
99.799
99.599
soft real
time
Mission
Critical
Non realtime
Table 4.6: Packet delivery ratio (PDR) %
Service classes
%
100,1
100
99,9
99,8
99,7
99,6
99,5
99,4
99,3
50 nodes
100 nodes
200 nodes
soft real time
mission critical
non real time
Fig 4.6: Packet delivery ratio (PDR) for SGLM service classes
It is worth observing that node density has a perceivable effect on both the
data latency, throughput and communication reliability.
58
4.4 SUMMARY OF FINDINGS
This chapter presents the discrete event simulator results of the
performance of WiMaxin the SGLM. Delay, throughput and reliability are the
metrics used in analyzing its performance but have only been presented in
this chapter with their analysis to be made in the chapter 5. Delay results in
the UL and DL directions show far lower delays experienced across all
service classes in the DL compared to the UL. Throughput was shown by
the simulation to have a direct correlation to the direction of traffic flow with
non real-time service class exhibiting the greatest throughput in the DL.
Packet delivery success was used in computing the reliability metric and the
simulation results proved very little degradation occured with increasing
nodes across all service classes.
59
CHAPTER 5: DISCUSSION OF RESULTS
5.1 INTRODUCTION
This chapter provides insight via critical analysis to the results derived from
the simulation in the preceeding chapter. Each performance metric of delay,
throughput and reliability are used in evaluating WiMax performance in the
SGLM based on the minimum performance requirements established earlier
in the literature review of this work.
5.2 PACKET DELAY ANALYSIS
The latency valuesshownto increase in the UL and DL with the number of
nodes in the preceeding chapter isdue to the increased scheduled
packetsand polling time for the increasing number of nodes. A 42% increase
is recorded in the UL of the mission critical class, 33% in the soft real time
and a 98% rise in the non real time service class in comparison to their DL
transmissions. These figures reveal that the polling component creates an
overhead in the UL transmission which has a very significant bearing on
packet latency. DL transmissions use no polling, hence high values in the
non real time service class is due to its low priority in the scheduling
algorithm used as DL packet delays are a function of both transmission and
scheduling delays.
Though the mission critical service class is characteristic of high polling
(Gomez-cuba 2010), the polling overhead was addressed via the use of
deadline time stamps specified in OPNET for the rtPS service flows in the
UL to surpass the theoretically established requirement values in chapter 2.
The increment in delay with increasing number of nodes for applications in
the DL as compared to the UL is also less because as the numbers of
devices increase, the mapping information allocation is made closer to the
left side of the WiMAX DL subframe thus increasing the delay.
The soft real-time service class delay slightly increases by 1.6% with the
number of users at 200 as UGS flows is expected to experience little latency
60
with increase in nodes due to its fixed periodic bandwidth allocation.In
comparison with the delay rise at 63% of the non real-time service class in
the DL, it is observed that the mission critical service class delay rise is
much lower at 5.2%. This is due the fact the simple priority scheme used by
the rtPS scheduler in addition to specifying the deadline time stamps of rtPS
service flows assures a minimal delay because of a higher priority over the
nrtPS scheduler used for the non real time service class.
The less critical soft real time service class possess lower latencies due to
their mapping to the UGS service flows. For both the mission critical and soft
real-time service classes, difference between theirtraffic’s minimal latency
requirement established in the literature review and simulation is quite
considerable with both amply satisfying the required latency requirement as
shownin table 5.1 below.
Though no polling is used in the DL, the non real time service class
experienced unusually higher delays than recorded in the UL, which is a
function of the scheduling and transmission delays as established earlier in
this section. Despite that, its performance is still tolerable as it has latency
requirements is in the order of minutes as captured in the table below.
Traffic classes
Literature review
Simulation Result (UL
recommendation
and DL highest delay)
Mission critical
100ms
19ms
Soft real time
500ms
6.15ms
Non real time
2s
500ms
Table 5.1 Comparison of simulation and literature review packet
latency values
In light of the above, it is evident the adopted model and priority scheduling
used can differentiate between traffic classes guaranteeing lower queue
waiting times for higher priority messages, as it isdefined in the 802.16
61
standard and the WiMax network meets the latency requirement for SGLM
communications.
5.3 THROUGHPUT ANALYSIS
Soft real time service class exhibited a near constant throughput in both UL
and DL due to the fixed and guaranteed allocation of resources by the BS
and SS schedulers respectively during UGS service flows transmissions.
With the 100kbps minimum bandwidth requirement earlier established in the
literature review, the simulation results which recorded throughput of
250kbps and 120kbps for the application traffic in the DL and UL direction
respectively shows WiMax effectively meets the bandwidth performance
requirement for this class of SG traffic.
Throughput values for the non real time service class peaked in the UL as
firmware update flows seemingly consumed most of the available resource.
This can be attributed to the fact that no application traffic within the model
exists for the non real time service class in the uplink direction as such the
results recorded in the UL by OPNET represents polling requests from the
BS to SS and not actual data transmission. (Castellanos 2010) argues that
even with low delay values, it leaves a big space(78.8%) which increases
even further with the number of users. In the DL though, throughput values
attained via simulation meet up with the SGLM requirement of its service
class established in chapter 2 by over 40%.
Results proved the efficacy of this work’s model in two ways. Firstly, it shows
the correlation between throughput and packet loss, as flows with higher
throughput suffered less packet loss. Secondly, using non-real time service
class as UDP based transmission proved correct, as transmissions of nrtPS
flows with FTP packets would make it fall short of the SGLM requirements.
Mission critical flows throughput is degraded by the highest percentage with
addition of more SG nodes to the WiMax network but is maintained at a
62
higher value above the non-real time service class and lower throughput
than the soft real time traffic due to simple priority scheme maintained by the
BS and SS schedulers.
5.4 RELIABILITY RESULTS ANALYSIS
In WiMAX networks, packet loss occurs where nodes lose their connection
to the BS due to low link SNR. Packet loss with 200 nodes simulation
recorded a 0.013% loss for the soft real time, 0.101% for the mission critical
class and 0.401% for the non-real time service classes. Castellanos (2010)
refers to this occurrence as shadows fading where any packet arriving at the
node buffer is dropped. The non-real time service class experienced the
highest packet loss due to its not engaging any retransmission technique like
ARQ used in other service classes.
Additionally, traffic which are less delayed by the priority scheduling used
have a higher delivery success ratio and can be inferred hereon that packet
delivery success depends to a large extent on the scheduling algorithm
adopted. This should be an area for further research.
5.5 SUMMARY
This chapter shows that the traffic and network model developed in the
course of this work and used in the simulation meets up with their latency,
bandwidth and reliability requirements as established in the literature review
of this work.
The simple priority scheduling employed ensured latency values for alarm
signals and alarm commands of the mission critical class are improved
considerably amidst polling overheads in the DL transmission via
configuration of deadline timestamps for the rtPS service flows. The model
granularly differentiates between traffic classes ensuring lower packet
latencies for higher priority messages, as it isrecommended in the 802.16
standard.
63
A relationship between packet loss and throughput is uncovered as lower
priority packets sent with UDP having no retransmission mechanism
recorded lower throughput. The scheduler ensured higher throughput for
higher priority packets staying in line with the 802.16 recommendations and
meeting the SGLM communication requirements established in the literature
review of this work.
Reliability requirements were seen to also have a relation to packet delay
because traffic with lower packet latencies had higher delivery success
ratios. This observation was also advanced by this work as an area for
further research.
The SGLM communication requirements of reliability, latency and
bandwidth, though significantly affected by node densities were still met by
the WiMAx network.
64
CHAPTER 6: CONCLUSION AND FUTURE WORK
6.1 THESIS SUMMARY
In this chapter a summarization of the entire project is made and a
recommendation of future research efforts is also advanced.
The chapter 2 of this work defined a smart grid as the traditional electricity
grid embedded with ICT and telecommunications technologies to address
some needs such as the integration of cleaner forms of elctricity generation,
automated metering etc. A model of the smart grid as envisioned by the
NIST was presented with its seven domains and their peculiar
communication needs.The same chapter outlined the applications and
technologies of the smart grids and their quantitative communication
requirements as presented by the DOE. A host of wireline and wireless
telecommunication technologies were reviewed to ascertain their suitability
for the SGLM with particular focus on WiMax.
Chapter 3 began with an insight on the use of simulation as a research
methodology. Thereafter a traffic model was created based on service
classification of the SG traffic, with a simple priority queing mechanism
defined. This was used in conjunction with a network model’s defined
parameters in simulating the behavior of the WiMax network in the last mile
of the SG.
The simulation results for delay throughput and reliability were presented in
chapter 4. The results showed that though the modelling assigned UGS
service flows to the less critical traffic due to their fixed and periodic nature,
the simple priority algorithm in use with deadline time stamps ensured rtPS
service flows did meet up with their SGLM requirements established in
chapter 2.
The results from the simulation reveal indeed that WIMax meets the latency,
bandwidth and reliability requirements of SG applications in the SGLM
network.
65
6.2 RECOMMENDATIONS
More research effort is required in determining if enhanced scheduling
algorithms or their combination can further improve throughput, packet
latency and packet delivery success values for applications of the mission
critical class. As stated earlier also research efforts should also be geared
towards determining if packet delivery success depends on the scheduling
algorithm adopted and to what extent.
Packet delivery reliability requirements were observed in this thesis to have
a relation to packet delay, further research along this line would help in
determining how to meet thesee requirements with a single solution.
A point to multipoint architecture was employed for the WiMAx network.
Research efforts should be carried out in determining if WiMax can still meet
these SGLM requirements when used in a mesh topology.
66
REFERENCES
Adamiak, M., Communications for Smart Grid. Communications.
Adhikari, R., 2013. Strategy For Adopting Communication Technologies in
Smart Grid. Manager.
Aguirre, J.F. & Magnago, F., 2013. Viability of WiMax for Smart Grid
Distribution etwork. Journal of Science and Technology, 2(April),
pp.181-196.
Callaway, D.S. & Hiskens, I. a, 2011. Achieving Controllability of Electric
Loads. Proceedings of the IEEE, 99(1), pp.184-199. Available at:
http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=564308
8. NETL, 2008. ADVANCED METERING INFRASTRUCTURE. Energy,
(February).
Camacho, E.F. et al., 2011. Control for Renewable Energy and Smart Grids.
Control.
Castellanos, G., 2012. Wireless Communication Network Architecture for the
Smart Grid applications. Network.
Castellanos, G.D. & Khan, J.Y., 2012. Performance Analysis of WiMAX
polling service for Smart Grid Meter Reading Applications. Electrical
Engineering, (1), pp.1-6.
Cisco, 2012. Cisco Connected Grid Security for Field Area Network.
Transport, pp.1-6.
Clark, A. & Pavlovski, C.J., 2010. Wireless Networks for the Smart Energy
Grid : Application Aware Networks. Computer, II.
Claudio, L., 2010. Enabling a Smarter Grid. Agenda, (September).
67
Curran, K. & Condell, J., 2010. TOWARD MORE EFFICIENT WIRELESS “
LAST MILE ” SMART GRID COMMUNICATION SYSTEMS 1 . FAST
CARRIER DETECTING AND COLISSIONS PROBLEM. Science.
Energy, V.B. & Solutions, U., 2009. There is No SMART in Smart Grid
Without Secure and Reliable Communications. Security, (April), pp.1-8.
Ernst & Young, 2011. Attacking the smart grid. , (December).
Erol-kantarci, M. & Mouftah, H.T., 2012. Wireless Sensor Networks for
Smart Grid Applications. Engineering.
Forum, W., 2013. WiMAX Forum ® System Profile Requirements for Smart
Grid Applications Requirements for WiGRID WMF Approved WiMAX
Forum Proprietary. Forum American Bar Association.
Gao, J., Wang, J. & Wang, B., 2012. Research on Communication Network
Architecture of Smart Grid. Notes, 9, pp.496-501.
Gavin, S., 2013. Evolution of grid design From traditional to future grids (
SMART ).
Gharavi, H. & Xu, C., Analysis of Traffic Scheduling Technique for Smart
Grid Mesh Networks. Technology.
Gulich, O., 2010. TECHNOLOGICAL AND BUSINESS CHALLENGES OF
SMART GRIDS. Challenges.
Guzelgoz, S., 2011. Characterizing Wireless and Powerline Communication
Channels with Applications to Smart Grid Networks. Computer
Engineering.
Hammoudeh, M., 2012. Comparative Analysis of Communication
Architectures and Technologies for Smart Grid Distribution Network.
Architecture.
68
INTERNATIONAL TELECOMMUNICATION UNION, 2011. Deliverable on
Smart Grid Architecture.
Iyer, G., 2011. Wireless Mesh Routing in Smart Utility Networks.
Environment.
Jeon, Y.-hee, 2011. QoS Requirements for the Smart Grid Communications
System. Journal of Computer Science, 11(3), pp.86-94.
Jewell, W. & Leader, P., 2012. Communication Requirements and
Integration Options for Smart Grid Deployment.
Kaur, H. & Singh, G., 2011. Implementation and Evaluation of Scheduling
Algorithms in Point-to-Multipoint Mode in Wimax Networks. , 4333,
pp.540-546.
Kirkham, H., 2010. A Survey of Wireless Communications for the Electric
Power System. Contract, (January).
Knight, E., 2010. Rural, Mobile Wireless Mesh Networking for Community
Learning. , pp.1-4.
Laverty, D.M. et al., 2010. Telecommunications for Smart Grid : Backhaul
solutions for the Distribution Network. , pp.1-6.
Leclerc, A., Ph, D. & Crosby, M., 2010. Test and Evaluation of WiMAX
Performance Using Open- Source Modeling and Simulation Software
Tools. Security, pp.518-524.
Manuel, B.J., 2010. EUROPE 2020: A strategy for smart, sustainable and
inclusive growth.
Miller, R.R., 2011. 4G Neighborhood Area Networks.
Nampuraja, E.K., 2011. Smart Grid Management System. , 9(5).
Nampuraja, E.K., 2011. Smart Grid Management System. , 9(5).
69
Neptune Technology Group, (NTG) I., 2010. F O R B A T T E R Y - P O W E
R E D A M R / A M I S Y S T E M S. Technology, pp.1-2.
NETL, 2008. ADVANCED METERING INFRASTRUCTURE. Energy,
(February).
NIST, 2010. NIST Special Publication 1108 NIST Framework and Roadmap
for Smart Grid Interoperability Standards ,. Nist Special Publication.
Page, E.H., Nance, R.E. & Arthur, J.D., 1994. : PRINCIPLES AND
SIMULATION MODELING Techniques, (September).
Paolini, M., 2010. Empowering the smart grid with WiMAX TM. Security.
Patil, R. & Shinde, A., 2013. Suitability Analysis of Last Mile Connectivity
Protocols for Smart Metering_TCS.
Quang-Dung, H. & Le-Ngoc, T., 2012. Smart Grid Communications
Networks: WirelessTechnologies, Protocols, Issues and Standards.
Secrest, T. & Bloyd, C., 2011. Using Smart Grids to Enhance Use of EnergyEfficiency and Renewable-Energy Technologies. Energy, (May).
Sensus USA, I., 2013. Next Generation AMI : Utilities Benefit from
Optimizing AMI Network Operation Using Multi-Channel Allocation.
Shah, G.A., Gungor, V.C. & Akan, O.B., A Cross-layer Design for QoS
support in Cognitive Radio Sensor Networks for Smart Grid
Applications. Time, pp.1398-1402.
Sood, V.K. et al., 2008. Developing a Communication Infrastructure for the
Smart Grid. Sensors (Peterborough, NH), 4, pp.1-7.
Trilliant Inc., 2010. Application Domain Partitioning for the Smart Grid
Application Domain Partitioning for the Smart Grid. Security.
70
Tropos Networks, 2009. Tropos GridCom
TM :
A Wireless Distribution Area
Network for Smart Grids Evolution of Electric Utility Communications.
Communications, (June).
Tuna, G., Gungor, V.C. & Gulez, K., 2013. Wireless Sensor Networks for
Smart Grid Applications : A Case Study on Link Reliability and Node
Lifetime Evaluations in Power Distribution Systems. , 2013.
UCAIug, 2013. SG Network SRS Harmonized Version V3.
Ullo, S.L., Vaccaro, A. & Velotto, G., 2010. Performance Analysis of IEEE
802 . 15 . 4 based Sensor Networks for Smart Grids Communications.
Networks, 1, pp.129-134.
van der Drift, B.J.., 2011. Smart Grid: Combining RF mesh grid and public
carrier networks for last-mile communications. Science.
Wang, W., Xu, Y. & Khanna, M., 2011. A survey on the communication
architectures in smart grid. Computer Networks, 55, pp.3604-3629.
Wang, W., Xu, Y. & Khanna, M., 2011. A survey on the communication
architectures in smart grid. Computer Networks, 55, pp.3604-3629.
Yan, Y. et al., 2012. A Survey on Smart Grid Communication Infrastructures :
Motivations , Requirements and Challenges. Communication, pp.1-16.
Zaker, N., 2013. A CCESS N ETWORK I NTEGRATION ( F I -WSN ) F OR
THE S MART G RID.
Zamzow, K., 2011. New Wireless Solutions to Meet Smart Grid
Communications Requirements. Power System Engineering.
71
BIBLIOGRAPHY
Periklis, C., Dimitrios, G. Stratogiannis, Georgios, I.T. & Giwrgos, S., 2011.
HANDBOOK ON GREEN INFORMATION AND COMMUNICATION
SYSTEMS.
Swire, P., 2009. Smart Grid , Smart Broadband , Smart Infrastructure Bang
for the Buck. , (April).
Gohn, B. & Wheelock, C., 2010. EXECUTIVE SUMMARY : Smart Grid
Networking and Communications Substation Automation , Distribution
Automation , Smart Meters , and the Smart Energy Home. Architecture.
Puthal, D., Sahoo, B. & Sahoo, B.P.S., 2012. Effective Machine to Machine
Communications in Smart Grid Networks. Architecture, 2(1), pp.18-22.
Kurose, J., 2013. Networking (related) Challenges for the Smart Grid.
Godfrey, T., 2012. Standards Updates for Field Area Networks. Power.
Bui, N. et al., 2011. Implementation and performance evaluation of wireless
sensor networks for smart grids. , pp.1-26.
Arnold, G., Pes, I. & Technical, I., 2013. Interoperability and Standards.
Europe.
Al-omar, B., Ahmed, R. & Landolsi, T., 2012. Role of Information and
Communication Technologies in the Smart Grid. Renewable Energy,
3(5), pp.707-716.
72
Download