1

Spectrum Needs for Wireless Smart Meter Communications

Purva Adke, Jesse Bumanlag, Bryan Edelman, Uller Doetsch
[email protected], [email protected]. [email protected],
[email protected]

A capstone paper submitted as partial fulfillment of the requirements for the degree of
Masters in Interdisciplinary Telecommunications at the University of Colorado, Boulder,
2 May 2011. Project directed by Professor Frank Barnes.
1. Introduction
It has been known for at least ten years that the United States electric grid is operating beyond its
initial design specifications instantiating a decrease in reliability [1]. This has resulted in
scheduled rolling black outs, unexpected black outs, and brown outs, all of which can potentially
cost financial loss to businesses [2], not to mention emergency services can become overtasked
and unavailable. In some instances not much can be doneunpredictable events cause failures in
the system. In others, a network that is self-healing, incorporates resilience to intentional attacks,
provides enhanced power quality, integrates on site power storage and generation, and operates
more efficiently can reduce variations in power distribution and increase reliability. Research
shows that a Smart Grid network would save between $46 and $117 billion over 20 years, the
Smart Grid can bring knowledge to power causing an evolution in the electric grid industry
[3].
The transition to a Smart Grid means adding intelligence to the existing utility grid. This
new information needs to be transferred from remote sensors to a central point. As utilities
deploy smart meters, it must be determined if their networks could be optimized by
implementing a wireless infrastructure. In many cases this may be the most inexpensive route to
deployment when compared to laying additional cables or upgrading power distribution
hardware to accommodate data communications across power lines. A determination of potential
frequency bands, both licensed and unlicensed, was conducted and a determination of suitable
bands for Advanced Metering Infrastructure is made.
2. Background
The term Smart Grid is fairly generic and encompasses many aspects of the power
communications networkparticularly the Advanced Metering Infrastructure (AMI) between
utilities and customers. A full Smart Grid implementation generally consists of three separate
communication networks, two of which are considered components of the AMI: home area
network and the neighborhood area network.
A home area network (HAN) where devices that consume power communicate with the
power supplier through a gateway, usually integrated into a power meter. This enables devices to
be central controlled and aware of the current distribution and demand environment. The HAN
can communicate via low power wireless transceivers or in home power line communications.
A neighborhood area network (NAN) connects energy meters to data aggregating sub-
stations. This link can be established wirelessly, adding an additional wire line to the smart
meter, or communicating over the power lines via power line communications (PLC) technology.
In other applications this network is also known as the last mile. Finally, aggregation
2

substations must communicate back to a utilitys central servers. This connection is most often
made with leased access lines, wireless microwave links or via PLC.
In the NAN the cost of non-wireless implementations is often prohibitively high for many
areas as in many cases streets will have to be dug up to lay cables or power distribution hardware
will have to be upgraded to allow data to travel across them. Proof of this is provided in
Californias state-wide smart meter deployments. A number of utility operators in California are
currently installing 11.7 million electric smart meters with an intended completion date of 2012
[4]. PG&E, SDG&E, and Southern California Edison are all using wireless technologies to
communicate with smart meters in the NAN [5] [6] [7]. The city of Houston is deploying a Smart
Grid network on top of 4G technology to provide public safety communications, Smart Grid
communications, and free Internet with excess bandwidth [8]. This is projected to save millions
of dollars each year compared with leased T1 lines [9].
The issue of deploying of wired metering infrastructure is amplified in rural areas where
telecommunications service providers and broadband operators lack infrastructure. As of now
wireless solutions are the most widespread and easily available since hardware and software are
available commercially off-the-shelf. Deploying a wireless system will also take less time as
fewer rights of way are needed and existing transformers and distribution hardware does not
need to be upgraded. Many utilities are choosing wireless backhaul to support the AMI, and
some are using wireless for their entire Smart Grid networks [10]. In many cases there exists a
wireless network that a utility can piggy back on, such as a cellular network [10]. The economics
of each system will be key in deploying and operating a viable, robust and cost effective
technology that can sustain a scalable neighborhood area network.
2.1. Spectrum
The Federal Communications Commission (FCC) has two dominant mechanisms to regulate
spectrum, the command and control model and the commons model, otherwise known as
licensed and unlicensed spectrum, respectively. In the command and control model the FCC
gives licenses to transmit on a specified spectrum chunk in a certain location. If others transmit
or otherwise interfere within that spectrum, a licensee has legal recourse to collect damages and
stop the interference. In the commons model spectrum is available for anyone to transmit at any
location. However, maximum transmission power is limited to prevent the signal from covering
too wide an area and overpowering less powerful signals. In the United States the common
spectrum bands exist at several frequencies, with the most popular being the ISM
1
bands: 902 to
920 MHz, 2.4 to 2.5 GHz, and 5.725 to 5.875MHz. These bands are becoming increasingly
crowded and in the future may become unsuitable for mission critical communications [11].
Not all frequencies propagate equally. Different frequencies transmitted at the same power
will travel different distances, penetrate physical objects with varied effectiveness, and require
different sized antennas to be received optimally. In general the higher the frequency the less
distance it will travel [12]. As a result some spectrum is more valuable than others. Spectrum
between 300 MHz and 3GHz is known as beach front spectrum, and is scarce and expensive to
obtain as it has ideal propagation characteristics for many terrestrial applications [13].
2.2. Current State
Wireless NANs are currently deployed in various spectrum ranges, using various bandwidths and
technologies. There is disagreement about what the requirements of a fully implemented Smart

1
Defined by ITU-R 5.138, 5.150, and 5.280
3

Grid system are. The Utilities Telecom Council (UTC), a registered lobbying group, suggests
several key requirements for implementation [14]. These include at least four nines (99.99%)
reliability in the network, increased bandwidth over current networks, ubiquitous coverage, tight
security measures, and uninterrupted power back-up [14]. Current NANs make use of cellular
data networks
2
, ISM band mesh networks
3
, and various wire line technologies
4
to transport data
from the smart meter to aggregation nodes on the utilities network. Each of these has unique pros
and cons with regard to the reliability, security, coverage, available bandwidth, capital expenses,
and operating expenses. It should be noted that all have been deployed and are thus feasible.
3. Research Topic and Question
The paper deals with the broad question of: Which bands of wireless spectrum (if any) are
suitable for a nationwide residential AMI deployment? The goal is to examine how various
available bands of wireless spectrum would be utilized in areas of high, mid, and low population
density. These metrics are inputs to a custom model that determines the smallest nationwide
infrastructure deployment, and thus determine the lowest cost to deploy a wireless NAN. The
different population densities are defined via the US Census Bureau as urban, suburban, and
rural. It is expected that different wireless bands will work best with different population
densities as capacity and coverage requirements are different for each area.
4. Importance
There is no consensus on using licensed or unlicensed spectrum, what frequency to use, or how
much bandwidth is required to operate a Smart Grid, or even if wireless technology is the
appropriate solution [15]. The US National Broadband Plan makes a number of
recommendations regarding communications networks for the Smart Grid, including considering
enabling utilities to share a section of the 700MHz public safety spectrum, considering use of
other federal spectrum for the Smart Grid, and studying the communications requirements of the
Smart Grid.
It has also been suggested by the National Institute of Standards and Technology that a
study of whether or not dedicated spectrum is necessary for utilities be conducted [16]. In fact,
there has been much debate over whether or not utilities should be allocated licensed spectrum
for their work [16]. The advantages and disadvantages of licensed versus unlicensed spectrum
for the Smart Grid have been established [17]. Given these advantages and disadvantages, it has
both been asserted that utilities should or should not be given a section of wireless spectrum [18]
[19].
Utilities and other companies are spending billions of dollars on developing and deploying
Smart Grid technology. The American Reinvestment and Recovery Act of 2009 (ARRA)
provided approximately $4 billion in stimulus funds to Smart Grid deployment efforts. There is a
guaranteed $4.5 billion in stimulus funds allocated and an additional $34.6 billion [20] has been
earmarked and could be allocated to Smart Grids. Also, individual public utility commissions are
independently mandating the updating of infrastructure meaning utilities are deploying smart

2
General packet radio service on GSM networks
3
Silver Spring Networks
4
Copper lines, Fiber lines, and PLC
4

metering solutions with no guidelines or regulations which will lead to a patchwork of wireless
networks operating in different fashions [21].
5. Intended Audience
This research is intended for three main audiences: utility and telecommunications executives,
policymakers, and industry lobbyists. Business executives will be able to make use of the
information to better understand the spectral environment and general implications of the
residential Smart, making inferences about cost, efficiency, and deployments. Communications
policymakers can utilize the research to aid in designing a regulatory system that protects core
communications policies and opens the marketplace to interoperability and innovation. Finally,
industry lobbyists will find the research useful when working with technology clients and
politicians on spectrum reform and other Smart Grid spectrum issues.
6. Scope of Research
To best understand the business implications of flexible spectrum allocation, the research has
been confined to the following:
Design a quantitative study of propagation characteristics and capacity limitations on
selected bands, and align this information with United States Census Bureau
population data
Analyze qualitative metrics and make a determination of suitable frequency bands
based on these metrics
Model the respective scenarios, developing finite conclusions on the appropriateness of
each identified band
All analysis is limited to the United States
All data used consists of residential information only, commercial data is not included.
The following have been deemed out of the scope of this study:
Study firms outside of the utility industry
Smart Grid technologies outside of the wireless realm
Develop new business models or evaluate current business models that pertain to
residential Smart Grid deployments
Determine appropriate regulations and policies for the implementation of a wireless
spectrum allocation for Smart Grid
Smart Grid recommendations outside of the United States
Elements of the Smart Grid beyond the advanced metering infrastructure (AMI) (i.e.
the last mile" segment of the network).
7. Assumptions
The following are assumed:
A significant portion of the last mile section of the Smart Grid network will use
wireless technology.
Individual meters will be connected to utilities via wireless access points or cells.
5

The technology implemented will use a single channel to communicate that does not
implement spread spectrum technology
The wireless technology utilized is interference tolerant
8. Methodology
The limiting factors in the size of a wireless coverage area can either be capacity or the distance
the wireless signal can propagate. In areas with dense populations, coverage may be limited by
capacity and not propagation distance. If data transmitted by all smart meters in a coverage area
exceeds the capacity of the network, that cell is considered capacity limited, affecting the cell
size in those areas.
8.1. Capacity Limited Cells
For some cases, it is necessary to determine the number of meter transactions that could occur in
one day. Hence, is required to determine the required bandwidth of a single meter. The
bandwidth requirement of a single meter is dependent on the amount of data sent, how often data
is sent, and the transmission parameters used. Meaning, the first step is to determine the amount
of data sent across the network.
When data is sent across a network, it is encapsulated with header and trailer information
that allow it to successfully traverse the network [22]. So, when calculating the total amount of
data transmitted across a network it is important to include both the encapsulated data, or
payload, and the encapsulating data, the header and/or trailer.
The main standards collection that tends to be used in commercial smart meter
implementations today is the American National Standards Institute (ANSI) C12.X standards
collection [23] [24]. The ANSI C12 standards encompass all elements of electrical meter design,
from electrical characteristics to the physical dimensions of the meters [25]. Since this research
is focused in the way that smart meters communicate, the model primarily used the ANSI C12.22
Protocol Specification for Interfacing with Data Communication Networks, to determine
encapsulating data sizes, and ANSI C12.19 Utility Industry End Device Data Tables, to
determine payload data sizes [26].
In determining the payload data size, two scenarios were considered: a non-real-time data
transmission scenario and a real-time data transmission scenario. For the non-real-time data
transmission scenario, it was assumed that the meter would be taking and storing usage
measurements locally once every fifteen minutes. It was also assumed that, at the request of the
controlling utility, every 24 hours the meter would send all of its daily usage information to the
utility. In essence, every day the meter would send the stored data for the previous day to the
utility, repeating the process every day. . For the real-time data transmission scenario, it was
assumed the meter would be taking and transmitting usage measurements every two minutes,
indicating 720 transactions per meter per day. Each of these scenarios have unique data payloads
that affect the overall analysis of infrastructure deployment. These assumptions were made due
to the fact that the XCEL deployment of smart meters in Boulder, CO uses this method of
obtaining information from the meters [27].
For determining the size of the encapsulating data we utilized the ANSI C12.22 standard,
which defines the protocol layer for communication between a smart meter and the utility [25].
All underlying network layers are undefined, allowing the use of whatever protocols best suit the
implementation [25]. For these calculations it was assumed that the connection uses TCP, since
6

reliability is more important than performance in this application. Due to the depleting source of
IPv4 addresses, we assumed that any Smart Grid implementation would use IPv6 [28]. For the
data-link and physical layers we did calculations for 802.15.4 (Low Rate WPANs), 802.16
(WiMax), Motorola Canopy, and GSM to model transmitted bits that are related to enabling
communications.
The amount of data transmitted by a single meter was then used in conjunction with the
characteristics of the wireless transmission, such as the modulation type, frequency, band-width,
and maximum transmit power to determine the amount of time a single transmission would take
for both real-time and non-real-time scenarios.
The result of the corresponding modeling analysis yields, the meters per cell, which
highlights the cluster size of the meters in a region that send data to a particular regional
collector (depends on the meter capacities and the utilities). This is a very important aspect that
will need to be incorporated this into any Smart Grid network design [29]. The channel capacity
and the symbol rate are based on the modulation type selected for the wireless transmission [12].
The average data per transaction time is the fraction of the data per transaction and the bits per
second obtained in earlier calculations.
8.2. Propagation Limited Cells
The max cell size based on propagation was then calculated from the link budget analysis,
limited by the distance of propagation. Using link budgets the model can account for all the gains
and losses of the power of a radio signal in a wireless link. This includes the transmitter and
receiver antenna gains, losses in the transmission media, cables, path loss, fading margins, body
loss, etc., to gauge the coverage area based on the receiver sensitivity and transmitted power
[12].
The link budget used takes the following elements into account:
Fading margin
Free space loss
Gains of the antennas
Cables losses between antennas and transceiver
Receiver sensitivity

Together, these elements define path loss, which is the attenuation in power of the radio
signal between the transmitter and the receiver [12]. In this case, the path loss being considered
is the Free Space Path Loss (FSPL). FSPL gives the estimate of signal strength and the distance
covered assuming no obstacles in the path. FSPL is proportional to the square of the distance
between the transmitter and receiver, and proportional to the square of the frequency of the radio
signal [30]. The FSPL equation is generally used as a simple rule of thumb to determine the
power loss in open air of a radio signal. It does not take into account reflections, scattering, or
propagation through objects. More advanced models exist to do this that can better account for
specific different environments but the FSPL equation is best suited to generically model the
entire range of radio environments. The equation for FSPL is as follows:

(1)
7

Where:
is the signal wavelength (in meters),
is the signal frequency (in hertz),
is the distance from the transmitter (in meters),
is the speed of light in a vacuum, 3 10^8 meters per second [12]
It is also expressed in dB as
(2)
From the equation it is clear that higher frequencies exhibit more FSPL per distance [12].
Maximum allowable path loss is the maximum amount of attenuation the system can
tolerate between the transmitter and receiver and maintain reliable communication across the
link. This is calculated by a link budget analysis [12]. The maximum path loss determines a
cells coverage area and can estimate the subscriber count (number of households) in a cells
maximum coverage.

Table 1 - Link Budget Utilized for Analysis
5

Abbreviation Link Budget Item Magnitude
PTX transmitter output power (dBm) 30
GTX transmitter antenna gain (dBi) 0
LTX transmitter losses (coax, connectors...) (dB) 2
LM miscellaneous losses and margins (dB) 6
GRX receiver antenna gain (dBi) 0
LRX receiver losses (coax, connectors...) (dB) 2
PRX receiver sensitivity(dBm) -94
LFS free space loss or path loss (dB) 54

The simple calculations in table 1 indicate a free space loss of 54 dB. Using this value an
estimation of the maximum coverage distance that can be obtained from a single transceiver site
can be calculated. This distance is then used to obtain an estimate of the infrastructure that would
be necessary to connect the AMI meters in the United States using the specified wireless
parameters in various regions as defined by the Census Bureau. The values obtained for the link
budget were determined after reviewing specification sheets for smart meters using multiple
technologies, investigating transmitter output power allowed at each respective frequency [31]
[32] [33] [34], and adding a 6 dB loss margin. All this can be easily changed in the model that
was developed.
8.3. Population Data

5
The values are based on FCC regulations, standard practice in the wireless industry, free space loss, and
manufacture data sheets
8

Different areas of the North America were identified based on a census data. According to a
report issued by the Congressional Research Service, the Federal Government classifies three
major population types by geographic regionurban, suburban, and rural [35]. Urban
environments are those with population densities greater than 1000 people per square mile (621
people per kilometer). Suburban is classified as those areas with population densities of 500
999 people per square mile (311 620 people per square kilometer), and rural is any region with
a population density greater than 1 and less than 499 people per square mile (1 619 people per
square kilometer) [35]. Utilizing Census data granulated by county and by population density per
square mile, a determination was made to select the maximum housing unit density of each of
the respective categories to estimate a maximum state. Additionally, the data was aggregated to
sum the total number of housing units in each category based on the above criteria.
Since the households per kilometer varies with the geographic locations and the density of
population, there is a lower density of meters in lower household density areas (rural areas) and a
higher density of meters in higher population density areas (urban). Thus, the coverage distance
for each cell site is not uniform across every geographic area. Hence, it is incorrect to assume
that any one cell coverage distance is valid coverage area for all cases. Additionally, different
modulation techniques require different transmission distances to maintain efficient
communications. In our research, we used these considerations (equation 2) to arrive at the
transmission distances suitable for different frequencies.
Thus the categorization into urban, suburban, and rural was used to determine region specific
requirements of Smart Grid deployments due to changes in population density. This was done to
check that a cell would not be limited by radio propagation distance and not capacity.
8.4. Frequency Band Selection
As an unlimited number of frequencies exist, only those that are feasible were selected as input
into our model. Feasibility was determined primarily by spectrum availability. This includes
unlicensed bands and bands currently unallocated. While more suitable bands may exist, the
reallocation and allotment of spectrum is an extremely time consuming and costly process and
thus were not included in this analysis. However, spectrum explicitly suggested by some vendors
and industry groups were also considered, regardless of their availability. Because the U.S.
wireless industry as a whole has only deployed 250,000 cell sites nationwide [36], it has been
determined any spectrum band requiring more than 3 million cell sites will be deemed infeasible
and will not be included for recommendation.

9. Results
After running a nationwide scenario based on the aforementioned network design details for each
of the respective potential frequencies, metrics on the number of meters per cell site and total
number of cells for urban, metropolitan, and suburban was collected. These metrics were
9

examined as an indication of the overall cost of a wireless Smart Grid metering system for
residential use in the United States. Based on the Xcel Energy deployment, two test cases were
developed to test both a non-real time and a real-time network scenario. The details of how those
scenarios were evaluated in are shown in Table 2 and Table 3 below.
Table 2 Data Transmission Scenarios 6.78 MHz Test Case
Transmission Non-Real-Time Real- Time

Data Standard ANSI ANSI

IPv4 / IPv6 IPv6 IPv6

Frequency 6.78 6.78 MHz
Power Level 36 36 dBm
Modulation Type QPSK QPSK

Bits / Sec / Hz 2 2

Symbol Rate 30 30 Kbps
Data Rate 60 60 Kbps
Meter Data

Data Interval (mins) 2 15

Transactions Per Day 720 1

Data Samples Per Transaction 1 96

Data Per Transaction 3300 2200 bytes
Average Transaction Time 430 300 msec

The metrics collected from each of these scenarios include:
Number of meters in cell, based on population densities
Number of cells deployed, based on population densities
Total number of cells deployed nationwide

Additionally, increasing the modulation techniques and data standards yielded little
changes in the overall infrastructure deployment because a majority of cells were propagation
limited, not affecting overall capacity in those cells. Modulation encodes several bits into one
symbol (symbol rates) and thus the capacity for propagation of data or essentially the bandwidth
of the system. In our simulations, we have examined that using a higher order modulation only
does not affect propagation limited cells because these cells do not use their full bandwidth as is.
However, the number of meters per cell increased significantly when the modulation technique
was changed to accommodate higher symbols per second (higher data rates) in capacity limited
situations. For example, using QPSK over BPSK would essentially double the number of meters
per cell in a capacity limited cell. This result only occurs in urban environments, with low
bandwidth and low frequency, such as the unlicensed 6.78 MHz band with a bandwidth 30 kHz,
under a real-time data transmission scenario.
After a careful analysis of the results, it was determined only 4 of the 17 modeled bands
were feasible in a nationwide deployment of AMI infrastructure, based on the requirement a
deployment would consist of fewer than 3 million cell sites. Appendix 1 displays the results of
the feasible frequencies generated by the model, using a non-real-time transmission scenario and
real-time data transmission scenario, respectively. The results of this chart indicate the expected
10

meters per cell for each geographic category, as well as the total number of cells expected for a
nationwide deployment for each band. These values can be extrapolated to make estimations of
total infrastructure cost for rural, suburban, and urban geographies for each band evaluated. The
results of these scenarios proves that although there are multiple options for deploying a wireless
smart metering system for residences in the United States, there could be a better option in terms
of spectral access, overall cost, and deployment and development. An analysis of these results
and a recommendation for the optimal bands and bandwidth can be found in the
recommendations section of this paper.
10. Conclusion and Recommendations
While it is difficult to predict the legal, regulatory, or financial situations involved with acquiring
or using each of these bands on a nationwide basis, there are obvious indicators about the
efficiency and usability of each band described based on the ease of deployment, the
infrastructure required, the costs associated with deployment and upkeep, and the overall
implications of managing an already complex network.
Based on the selected scenarios, it is obvious that the main limitation for smart grid
deployments will be propagation distance associated with various bands and their respective
transmitting powers. Even under a real-time pricing scenario, only certain environments, namely
urban environments, are affected by channel capacity limitations. This means that the overall
bandwidth associated with each of these bands in most cases is overly sufficient and many
estimates of bandwidth for Smart Grid deployments are overstated.
The results gleaned from evaluating the available and recommended bands for the Smart
Grid yield results which indicate deploying a nationwide wireless Smart Grid technology for
residences in the United States would be done best on unlicensed bands. The band the model
demonstrates is the best is the unlicensed and low frequency 6.78 MHz band, due to the fact
most of the cell sites in the United States are propagation limited. The propagation characteristics
of this band would yield the smallest infrastructure deployment, particularly in suburban and
rural areas. In fact, based on the various scenarios, rural and suburban geographies were always
propagation limited, even under the most demanding of data loads. Of the seventeen bands
evaluated, the only four that are feasible in terms of least infrastructure deployed are in
unlicensed spectrum, indicating the overall suitability of unlicensed bands. Only in dense urban
environments is each cell site limited by capacity, making this the only scenario in which larger
bandwidths or higher order modulation would be useful.
In response to both trade groups and government regulators, two bands were evaluated
especially for inefficiencies. First, the UTC advocates for a 20MHz band of spectrum at
1800MHz for AMI applications [37]. Our calculations show that a frequency allocation at this
band would be a greatly inefficient use of spectrum and require over 15 million cell sites. Even
in high-population density urban areas the cell size would be limited by the propagation distance
of the signal. Each cell would likely service only 32 smart meters. There would be such an
excess of capacity that even collecting real-time data from meters would not utilize all available
spectrum. Additionally, the National Broadband Plan has suggested possible use of the 700MHz
spectrum band for Smart Grid applications as well [34]. Although the cell size would be larger
with this band, allowing for up to 76 meters per cell, there would still be a high degree of
inefficiency with regard to spectrum use when compared to unlicensed spectrum options in
11

particular and a nationwide deployment would require over 25 million cell sites based on the
modeled data.
Ultimately, it is hard to make a convincing case for the deployment of residential smart
meters in the United States using licensed spectrum. In fact, it may be in various utilities best
interests to more heavily explore unlicensed bandsparticularly the lower frequency bandsas
the overall costs associated with deploying the infrastructure network are many times higher
when considering licensed bands, which does not even include the costs of acquiring such
frequency licenses. Finally, with a political push to deploy smart meters to residences, utilizing
unlicensed spectrum may provide the only feasible timetable for a rapid deployment.
12

Appendix A: Infrastructure Deployment Results By Frequency







Center
Frequency
(Mhz)
Symbol
Rate
(Mbps)
Morphology
Non-Real-Time Data Transmission Real-Time Data Transmission
Meters
Per Cell
Number of
Cells
Total
Number of
Cells
Meters
Per Cell
Number
of Cells
Total
Number of
Cells
6.78 0.03
Rural
280 220,000
270,000
280 220,000
360,000 Suburban
330 45,000
330 45,120
Urban
27,000 1,500
410 100,000
13.56 0.014
Rural
140 440,000
530,000
140 440,000
740,000 Suburban
170 90,000 170 90,000
Urban
13,000 3000 190 210,000
27.12 0.326
Rural
69 880,000
1,000,000
69 880,000
1,000,000 Suburban
83 180,000 83 180,000
Urban
6,700 6,000 4,400 9,400
40.68 0.04
Rural
46 1,300,000
1,500,000
46 1,300,000
1,600,000 Suburban
55 270,000 55 270,000
Urban
4,400 9,000 540 76,000
13

References

[1] John German, "Sandia's Critical Infrastructure Protection initiative picks up steam," Sandia Lab
News, vol. 52, no. 3, p. 1, February 2000.
[2] AUS Consultants, "Impact of a Continuing Electricity Crisis on the California Economy,"
Moorestown, New Jersey, 2001.
[3] Pacific Northwest National Laboratory, "GridWiseTM: The Benefits of a Transformed Energy
System," U.S. Department of Energy, 2003.
[4] California Public Utilities Commision, "Report to the Governor & the Legislature on Smart Grid
Plans and Recommendations," December 2010.
[5] San Diego Gas and Electric. (2011) Smart Meter: Frequently Asked Questions. [Online].
http://www.sdge.com/smartmeter/faq.shtml
[6] Pacific Gas and Electric. (2011) SmartMeter System - How It Works. [Online].
http://www.pge.com/myhome/customerservice/smartmeter/howitworks/
[7] Southern California Edison. (2011) About Edison SmartConnect. [Online].
http://www.sce.com/CustomerService/smartconnect/about_smartconnect.htm
[8] Alvarion. (2011, March) City of Houston Deploys Alvarion 4G Wireless Broadband Network for
Advanced Municipal Services.
[9] Jessica Scarpati. (2010, February) Rural wireless operator ditches T1s for microwave backhaul plan.
[10] S.Reedy. (2009, august) Connected planet. [Online].
http://connectedplanetonline.com/business_services/news/utilities-advanced-meter-infrastructure-
0820/
[11] Jennic, "Co-existence of IEEE 802.15.4 at 2.4 GHz," 2008.
[12] Beasley, Jeffrey S., and Gary M. Miller, Modern Electronic Communication. Upper Saddle River:
N.J.: Pearson/Prentice Hall, 2008.
[13] Peter Kaplan. (2008, March) Reuters. [Online]. http://www.reuters.com/article/2008/03/20/us-
wireless-auction-idUSN2042023420080320?feedType=RSS&feedName=technologyNews
[14] Utilities Telecom Council, "A study of Utility Communications Needs: Key Factors That Impact
Netowk Utility communications Networks," Washington, DC, 2010.
[15] Federal Communications Commission. (2010, March) National Broadband Plan. [Online].
http://www.broadband.gov/plan/12-energy-and-the-environment/
[16] Electric Power Research Institute, "Report to NIST on the Smart Grid Interoperability Standards
Roadmap," 2009.
14

[17] Sensus, "Licensed versus Unlicensed Spectrum for Utility Communications," Raleigh, NC, 2010.
[18] B. Bender, K. Kilbourne, "Spectrum for Smart Grid: Policy Reccomendatins Enabling Current and
Future Applications," in 2010 First IEEE International Conference on Smart Grid Communications,
Gaithersburg, MD, 2010, pp. 578-582.
[19] Silver Spring Network, "Why Unlicensed Spectrum Dominates the Smart Grid," Silver Spring
Networks, Redwood City, Whitepaper 2010.
[20] Department of Energy. (2009, october) smartgridnews. [Online].
http://www.smartgridnews.com/artman/publish/Stimulus_Awards_Projects_Resources/Smart-Grid-
Investment-Grant-Awards-by-Category-1328.html
[21] California Public Utilities Commission. (2010, June) CPUC Launches Plan To Modernice Electric
Grid.
[22] Vinton et. al. Cerf, "Specification of Internet Transmission Control Program," 1974.
[23] SilverSpring Networks, "Communications Module for Electricity Meters," Redwood City, CA, 2011.
[24] SmartSynch Inc., "kV2c SmartMeter Data Sheet," Jackson, MS, 2010.
[25] American National Standards Institute, "C12.22: Protocol Specification for Interfacing to Data
Communication Networks," 2008.
[26] American National Standards Institute, "C12.19: Utility Industry End Device Data Tables," 2008.
[27] XCEL Energy Corp., Boulder SmartGridCity Presentation, February 25, 2011.
[28] Internet Corporation for Assigned Names and Numbers. (2011, February) IANA IPv4 Address Space
Registry.
[29] Michael J Martin. Generating insights. [Online].
http://www.generatinginsights.com/whitepaper/backhaul-networks-customizing-solutions.html
[30] D.McGinnis, Exelon Smart Grid Communications Strategy & Architecture, 2011.
[31] Silverspring Networks. (2011, February) Communications Module for Electricity Meters. [Online].
http://www.silverspringnet.com/pdfs/SilverSpring-Datasheet-Communications-Modules.pdf
[32] Smart Synch. (2010) GE I-210+c Meter. [Online]. http://www.smartsynch.com/pdf/I-
210+c_SmartMeter_e.pdf
[33] Smart Synch. (2010) Centron GPRS Meter. [Online].
http://www.smartsynch.com/pdf/CENTRON_SmartMeter_e.pdf
[34] Federal Communications Commission of the United States. (2009, October) Wireless
Telecommunications Bereau. [Online]. http://www.gpo.gov/fdsys/pkg/CFR-2009-title47-
15

vol1/pdf/CFR-2009-title47-vol1-part15.pdf
[35] Congressional Research Service - United States Congress. (2005, June) CRS Report for Congress.
[Online]. http://ncseonline.org/nle/crsreports/05jun/97-905.pdf
[36] CTIA. (2010, December) CTIA. [Online]. http://ctia.org/media/industry_info/index.cfm/AID/10323
[37] Utilities Telcom Council, "The Utility Spectrum Crisis: A Critical Need to Enable Smart Grids,"
2009.
[38] et. al. K. Budka, "Communication Network Architecture and Design Principles for Smart Grids,"
2010.
[39] Alcatel-Lucent, "Smart Choices for the Smart Grid: Using Wireless Broadband for Power Grid
Network Transformation," 2010.
[40] Tylor Seaman, Charles Upshaw, Ehab Haron Melissa Lott, "The Smart Grid in Texas, A Primer,"
University of Texas, 2011.
[41] Department of Energy, "A Vision for Smart Grid," Office of Electricity Delivery and Reliability,
Washington, DC, 2009.
[42] Utilities Telecom Council, "The Utility Spectrum Crisis: A Critical Need to Enable Smart Grids,"
Washington, DC, 2009.
[43] Verizon Wireless, 4G LTE, April 2010.
[44] E.Gubbins. (2009, June) Connected planet. [Online].
http://connectedplanetonline.com/residential_services/news/qwest-current-smart-grid-network-0616/
[45] Congress, Communications Act of 1934, Ammended Telecommunications Act of 1996, June 19,
1934, 47 U.S.C. 151.