Study of Future Demand for Radio Spectrum in Canada 2011-2015

6. Key Findings and Spectrum Demand Forecast

6.1 Cellular Services

6.1.1 Overview

The Canadian cellular industry has significantly grown since the mid-1990s with the introduction of personal communications, advanced digital networks and consumer devices with a wide range of smartphones and data-rich services and features. As of early 2011, the Canadian cellular market served close to 25 million users with a penetration of 74%. It is estimated that one or more mobile operators serve 99% of the population, with the coverage concentrated mainly along the southern land area of Canada.

To date, the three large incumbents support more than 95% of the subscriber base. Some regional operators and several new entrants have won spectrum at the 2008 Advanced Wireless Service (AWS) auction. They are currently gaining some traction by adding new subscribers.

Since 1985, Industry Canada has licenced a total of 270 MHz for cellular services (50 MHz in 800 MHz cellular band; 130 MHz in the 1900 MHz PCS band and 90 MHz in the 1700/2100 MHz AWS band). In general, Canada follows the lead of the U.S. in releasing new frequency bands within 18-24 months. In addition to the 270 MHz of spectrum outlined, the U.S. has released 60 MHz of commercial mobile in the 700 MHz band and 200 MHz in the 2500 MHz (BRS) band.

The three largest cellular operators with the regional operators retain approximately 85% of the licenced spectrum (240 MHz of the 270 MHz available). The new entrants in the AWS band have 40 MHz of the 270 MHz of spectrum licenced across Canada, and some of the new entrants have 10 MHz to 20 MHz in the four largest cities.

Industry Canada has held consultations to release new spectrum for commercial broadband mobile service in the 700 MHz (up to 84 MHz) and has re-farmed the band 2500-2696 MHz (196 MHz) to Broadband Radio Service (BRS), including high-mobility broadband applications. The addition of up to 270 MHz would double the amount of spectrum available to commercial cellular operators for a total of 540 MHz.

The Canadian cellular industry has been undergoing a rather rapid transformation with the rollout of HSPA/HSPA+ technology supporting 21/42 Mbps throughput and significant capability to handle IP data services, multimedia, video streaming and other high-speed services. The phenomenal adoption of social networks, the capability of advanced smartphones, computing power of tablets, use of aircards for laptops and netbooks, and the tens of thousands of mobile applications (apps) is driving mobile data traffic at exponential growth rates since 2009. Recent HSPA+ networks and the new 4G (LTE and WiMAX) networks just being introduced are likely to experience an exceedingly high data traffic growth rate and relative spectrum demand. This section will study the cellular service growth and spectrum demand for the current time on up to 2015.

6.1.2 Spectrum Inventory and Spectrum Utilization

Industry Canada's Inventory Report, in Section 1.0, provides important information on the distribution of spectrum amongst the various operators, the number of cell sites, frequencies and other information. For example, the spectrum holdings for the operators in large cities are presented in Figure 6.1.1, below (Inventory Report, Figure 1.3).

Figure 6.1.1 — Spectrum holdings by operators in large cities

Spectrum holdings by operators in large cities (the long description is located below the image)

Source: Inventory Report, Figure 1.3

Description of Figure 6.1.1

This chart provides data on the bandwidths of service providers for Canada's largest markets. Service providers shown are Rogers, Bell, TELUS, Globalive, Videotron, Shaw, Mobilicity, Public Mobile, and 'Other'. Markets represented are Quebec City, Montreal, Ottawa, Toronto, Winnipeg, Edmonton, Calgary and Vancouver.


The 700 MHz Consultation process has summarized the overall spectrum-population holdings of each wireless carrier and Industry Canada (spectrum for auction), including the 700 MHz and 2500 MHz spectrum, as can be seen in Figure 6.1.2, below. The figure assumes that 84 MHz of commercial spectrum in the 700 MHz band is available.

Figure 6.1.2 — Cellular, PCS, AWS, BRS and 700 MHz spectrum
(Total: 544 MHz weighted by population)

Cellular, PCS, AWS, BRS and 700 MHz spectrum (the long description is located below the image)

Note: The 700 MHz band is based on a maximum of 84 MHz of spectrum available. The ESMR spectrum, at 800 MHz, is not studied as part of this section and not addressed in Figure 6.1.2.
Source: Industry Canada

Description of Figure 6.1.2

This pie chart provides a breakdown of spectrum-population holdings of each wireless carrier and Industry Canada, including the 700 MHz and 2500 MHz spectrum. Cellular, PCS, AWS, BRS and 700 MHz spectrum are all considered, assuming that 84 MHz of commercial spectrum in the 700 MHz band is available. Rogers has 29%, Bell 19%, TELUS 10%, MTS 1%, Sasktel 1%, Videotron 2%, Globalive 2%, Shaw 1%, Mobilicity 1%, Bragg 1%, and ‘others' 1%. Industry Canada is shown to have 32%, with 16% in 700 MHz and 16% in BRS.


Spectrum Utilization

Preliminary analysis of spectrum utilization in urban core areas is developed from a number of sources, including the 2009 CRTC CMR Report, Spectrum Holding record (Figure 6.1.1, above) and demographic data. This is illustrated in Table 6.1.1, below.

The Table raises some interesting information about the cellular spectrum usage in the three largest markets: Toronto-Hamilton, Greater Vancouver Area and Montréal.

First, the population density and potential cellular market in the Toronto-Hamilton area is relatively higher than Vancouver and Montreal. According to the market share estimate by the three largest cellular operators and their spectrum holdings, it is seen in the last column that in Toronto, each operator has relatively the same density of subscribers per MHz (23.2 K to 28.7 K per MHz), but in Montreal and Vancouver, some operators have twice the subscribers per MHz than others.

The Table, below, shows that spectrum holdings per each operator in large cities are not necessarily proportional to their market shares (density of subscribers per MHz), and different spectrum pressures may exist among cellular operators.

Table 6.1.1 — Projection of a typical profile of Canada's three largest carriers' operations in key urban centres
Province
(Percent of Market share,
from 2009 CRTC-CMR)
Bell Group TELUS Rogers Others Cell penetration,
by Province
City Pop., areas sq. km., and spectrum holdings (Fig. 6.1.1) Calc. of subs/MHz Van.,Toronto and Montreal [100% pen. In peak period]
British Columbia 16% 41% 42% 0% 75% penetration Vancouver
2.0 M pop.
1176 sq. km.
1748 pop./sq. km.
Bell: 50 MHz
TELUS: 65 MHz
Rogers: 105 MHz
Bell = 6.4 K subs/MHz
TELUS = 12.6K subs/MHz
Rogers = 8.0 K subs/MHz
Ontario 32% 20% 47% 1% 70% penetration Toronto-Hamilton
5.8 M pop.
2279 sq. km.
2540 pop./sq. km.
Bell: 70 MHz
TELUS: 50 MHz
Rogers: 95 MHz
Bell = 26.5 K subs/MHz
TELUS = 23.2 K subs/MHz
Rogers = 28.7 K subs/MHz
Quebec 39% 26% 33% 2% 55% penetration Montreal
3.4 M pop.
1676 sq. km.
2005 pop./sq. km.
Bell: 50 MHz
TELUS: 60 MHz
Rogers: 105 MHz
Bell = 26.5 K subs/MHz
TELUS = 14.7 K subs/MHz
Rogers = 10.7 K subs/MHz
Total Subscribers

Canada (1Q/2010)

6.9 M 6.5 M 8.5 M 1.0 M 22.971 M

Source: CRTC CMR Report and Red Mobile Analysis

6.1.3 Stakeholder Input and Research Analysis

As part of the research, several industry participants, including operators and vendors, were surveyed for their input and perspectives. As well, the submissions made by several parties to the 700 MHz Consultation were reviewed.

The input received is not being shared in this document to protect contributor information. However, the input provided by various sources has been taken into consideration, as part of the overall approach to evaluating the growth of subscribers, services and traffic, and impact on spectrum. Where numbers are shown, these numbers are not the exact figures and projections shared by various parties, but rather what was determined based on analysis. The analysis was based on in-house expertise and assessment of the factors that influence growth, alongside information provided by all parties, as well as secondary research.

6.1.4 Service and Spectrum Demand

In this section, the projections for subscribers and traffic, the assumptions used to convert them into a demand for spectrum, and the results for alternative scenarios and the sensitivity analysis using alternative definitions of spectrum demand are presented.

Service Demand: Market Analysis

In developing the projections for service demand, Red Mobile used a combination of in-house expertise and primary research, and then reviewed this against several sources of reputable secondary research. While the initial analysis took place at YE 2010, the projections were revised several times over the course of the year, as new sources of current information became available. This included data from operators on actual consumption, as well as from other sources that monitor traffic. While the revisions meant several iterations of the models for service demand, which results in further iterations of spectrum demand, the exercise proved to be both necessary and highly valuable.

When the initial projections were made for YE 2010 and reviewed against industry input and secondary research, the projections aligned well with several reputable secondary sources, as well as with long-term market trends. However, over the course of 2011, a marked increase in data consumption was noted that was greater than what many had projected in the industry and constituted unusually high-percentage growth rates, when compared with the preceding two decades. As a result, projections were revised to be more reflective of the Canadian consumers' appetite for data services.

Several factors have led to this:

  1. The rapid growth in adoption of high-end smartphones has continued to grow, despite any presumed economic pressures;
  2. There has been a greater uptake of tablets and netbooks as they become more affordable;
  3. The enhanced network throughput and improved overall user experience for data services through better devices, engaging applications, etc. has propelled data usage;
  4. Canadian tariff rates for data have become significantly more attractive over the last two years, and consumers have responded to this.

A top-down analysis was conducted to determine consumer-driven demand over the next five years. The analysis included determining growth of subscriptions, changing the mix of the types of devices (i.e. feature phones to different types of smartphones, connected broadband devices/dongles, etc.), traffic-growth based on device type and mix of traffic (i.e. voice, SMS, email, web browsing, etc.), among other factors.

Overall, Red Mobile predicts that, with the growth in the variety and capability of device types, the insatiable appetite of users, and the evolving ability of networks to support higher data rates, the annual growth rates of traffic will remain extremely high in the short term.

Another factor, while not explicitly shown in this Study, but considered as part of the overall data growth, is the growth of Machine-to-Machine (M2M) communications. While M2M communications involves many devices, the aggregate traffic per connection is nominal, compared to that of the high-end broadband devices. While M2M devices are not counted as part of the overall subscriptions shown in this Study, the traffic generated by such devices as part of the overall offered traffic calculations has been accommodated.

The charts below summarize the projections for subscribers, data traffic and total traffic.

Figure 6.1.3 shows the number of subscriptions growing from 25 M at the start of 2011 to more than 34.6 M by 2015, translating to a penetration rate of a little more than 96%.

Figure 6.1.3 — Growth of cellular subscriptions

Growth of cellular subscriptions (the long description is located below the image)

Source: Red Mobile Analysis and Projections

Description of Figure 6.1.3

This chart provides the numbers of subscriptions per year from 2005 to 2015. The trend is a growth in the number of subscriptions, starting at approximately 17,000,000 in 2005, and increasing steadily to almost 35,000,000 by 2015.


Canadians typically replace their devices every couple of years, and a growing number have been upgrading to smartphones. It is estimated that, by the end of 2015, 60% of devices will be mobile-broadband devices, consisting of smartphones, dongles and mobile-capable tablets and netbooks. This represents a growth in mobile-broadband devices of 2.7 times over the period from the end of 2010 to 2015.

As can be noted in Figure 6.1.4, according to projections, the number of high-end smartphones will surpass the number of entry-level smartphones by the end of 2015. There will also be a marked increase in the number of mobile-enabled tablets over this period, as they become more affordable, with a large variety of them available in the sub-$300 range. By year end 2015, it's estimated that close to 57% of all non-smartphone mobile-broadband devices will be tablets or an equivalent type of device. In the figure below, these devices are differentiated from the remainder of non-smartphone mobile-broadband devices, which include netbooks, dongles, etc.

Figure 6.1.4 — Distribution of device types

Distribution of device types (the long description is located below the image)

Source: Red Mobile Analysis and Projections

Description of Figure 6.1.4

This chart shows the numbers of mobile subscriptions per year, from 2005 to 2015, as in the previous figure (6.1.3). In this chart, the numbers of subscriptions are broken down in to 5 categories of devices used:

  1. No. of Mobile HC BB Laptops, Dongles, etc.,
  2. No. of Mobile HC BB Tablets and similar devices,
  3. No. of High-end Smartphones,
  4. No. of Entry-level Smartphones, and
  5. Non-Broadband Devices.

The main trend is that the use of non-broadband devices is at almost 100% of in 2005, slowly decreasing over the years to about 40% by 2015. In turn, the percentage of use of the other devices increases over the years.


In determining the projections for data traffic, Red Mobile analyzed the growth in usage of each type of application, taking into account evolution towards convergence of multiple services in a given application; the evolving mix of the types of devices; and the average consumption of data per device type. In addition, the inherent improvement in capability within a device class was also considered. That is to say that, 2015's entry-level smartphones will likely perform at the levels close to those of 2010's high-end smartphones. Likewise, the high-end smartphones will continue to grow in capability with dual-core processing power. Also noted in the figure is that 2015's feature phones will be far more sophisticated than those of today and may, in fact, see the data usage that entry-level smartphones had experienced in 2010.

Four main sources of traffic were treated separately:

Voice: While voice usage has been declining slightly over time, incoming and outgoing calls combined were estimated to remain, on average, between 300 and 400 MoU (Minutes of Use) monthly per subscription. For the purpose of the modelling, it was estimated that this would remain flat at 400 MoU per subscriber.

Messaging and IM: Different forms of messaging — including SMS, IM, and MMS — were assessed for traffic. Converged applications, using messaging, social media, and forms of community-based communications, including RIM's BBM and Apple's iMessage services, are all seen as drivers for this medium of communication. The data usage for long messages, such as MMS, was arrived at separately from that of short messages, and the overall impact on traffic was calculated. As a result of the ongoing evolution of this space, messaging traffic has been projected to grow to 1 MB/month per subscription by YE2015.

Video, Web browsing, downloading, emails, content sharing: There has been rapid growth in the popularity of high-end smartphones and connected devices, including cost-effective tablets and netbooks. This has been a major catalyst in the uptake of data services in Canada. Extensive cellular operator investments in networks to support 3.5G (HSPA/HSPA+) and the recent start of 4G (LTE) deployments means that the majority of Canadians have access to networks that support broadband mobile communications, which, when combined with the high-end devices, have fuelled the insatiable appetite for data. As a result of the significantly lower cost to deliver data over 3.5G+ networks versus their predecessors, and growing competition in the market, consumers now enjoy lower data tariffs than in the past.

All of these factors have resulted in Canada experiencing one of the fastest rates of data growth in the developed world. Our projections show a growth rate of 140% year over year (YoY) at YE2011.

Figure 6.1.5 — Average monthly traffic generated by each device type

Average monthly
 traffic generated by each device type (the long description is located below the image)

Source: Red Mobile Analysis and Projections

Description of Figure 6.1.5

This figure provides the average monthly data traffic per year generated by each of the four device types:

  1. Entry-level smartphones
  2. High-end smartphones
  3. Highest capacity mobile BB (dongles, tablets, etc.) and
  4. Fixed substitution subscribers – traffic per line 0.50 x average HH.

The main trend is that from 2005 and on, the majority of data traffic is generated by fixed substitution subscribers, high-end smartphone and highest capacity mobile BB traffic increase slowly through the years, while entry level smarphones remain at 0 over the 10  year period.


Fixed Mobile Substitution (FMS) for broadband: In some cases, subscribers may use cellular for access to broadband Internet services. While broadband data usage via FMS represents a very small percentage of overall subscriptions, the usage per subscription is very high.

Figure 6.1.6 — Total monthly traffic per device

Total monthly traffic per device (the long description is located below the image)

Source: Red Mobile Analysis and Projections

Description of Figure 6.1.6

This chart provides the total voice and traffic data per device per year, from 2005 to 2015. Voice accounts for the majority of traffic from 2005 to 2009. In 2010, the split between voice and data traffic moves closer to 50%. From 2011 and on, data increasingly accounts for more traffic than voice, reaching about 95% by 2015.


In historical terms, the projections shown in Figure 6.1.6 represent a relatively strong growth scenario over the next five years, with an approximate doubling of data traffic per year, as Canada's cellular traffic grows to take advantage of improving technologies, devices and tariffs.

The total data traffic by 2015 is 30 times the 2010 level, as can be noted in Figure 6.1.7, below.

Figure 6.1.7 — Total monthly traffic growth for data, only

Total monthly traffic growth for data, only (the long description is located below the image)

Source: Red Mobile Analysis and Projections

Description of Figure 6.1.7

This chart provides the data traffic volume in GB/mo per year for data only, broken down by cellular service. The majority of traffic volume from 2010 to 2015 comes from 3.5 G HSPA. From 2012 and on, 4G – LTE traffic becomes an increasing but small factor, and by 2015, traffic volume is made up of about 66% from 3.5 G HSPA and 33% from 4G LTE. The values for traffic volume are summarized as follows:

Traffic volume
3.5G HSPA 4G - LTE
2007 0 0
2008 0 0
2009 576,396 0
2010 2,059,836 0
2011 5,657,313 0
2012 13,025,833 290,217
2013 24,927,501 1,909,259
2014 38,418,985 5,990,653
2015 53,658,455 17,566,543

It can be noted that Canadian operators have invested heavily in evolving their networks to embrace the latest technologies. In addition, Canadian consumers were quick to adopt newer smartphones and broadband devices using the latest HSPA networks. In fact, the Canadian smartphone uptake has been one of the highest in the world. Figure 6.1.8, below, shows the total traffic across all services (voice and data).

Figure 6.1.8 — Cellular traffic: Total traffic, all applications (GB/mo equivalent)

Cellular traffic: Total traffic, all applications (GB/mo equivalent) (the long description is located below the image)

Source: Red Mobile Analysis and Projections

Description of Figure 6.1.8

This chart provides the total traffic volume in GB/mo per year for all applications, broken down by cellular service. The traffic volume is summarized as follows:

Traffic Volume
2G - GSM2.5G - GPRS / EDGE 3G - WCDMA (UMTS) 3.5G - HSPA 4G - LTE 2G / 2.5G - CDMA 1xRTT 3G-3.5G - EV-DO (incl Rev A)
2007 479,570 0 0 0 1,090,199 26,036
2008 503,263 0 0 0 1,211,060 63,807
2009 463,341 0 957,910 0 1,082,327 116,388
2010 358,275 0 2,906,530 0 812,816 285,373
2011 277,853 0 6,911,858 0 631,358 322,707
2012 182,581 0 14,634,100 314,105 401,691 311,656
2013 118,016 0 26,669,222 1,998,195 243,836 279,900
2014 76,502 0 40,260,349 6,182,068 146,216 209,396
2015 57,721 0 55,626,347 17,996,038 101,661 122,196

As can be noted in Figure 6.1.9, the majority of the traffic (most of which is data traffic) is carried over HSPA networks today, and this trend is expected to continue to extend to LTE.

Figure 6.1.9 – Distribution of traffic across network technologies

Distribution of traffic across network technologies (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling
Note: in log scale

Description of Figure 6.1.9

This chart provides the data traffic volume in GB/mo by technology from 2007 to 2015 in the logarithmic scale. Traffic volume for 3.5 G HSPA increases significantly from 2007 to 2015 as does traffic volume for 4G LTE from 2012 to 2015.


Notes and Other Assumptions Regarding Traffic Projections

  1. Each year, approximately one-third of the subscriber base (existing subscribers and new addition) acquires new devices.
  2. Voice MoU (Minutes of Use – monthly per subscriber) will stay broadly constant over the next five years at 400 MoU.
  3. Smartphone projections: based on survey and validated by Q series paper for Canada and U.S. in 2014 - 5 yr = CAGR of 18.6%. Estimates indicate a 16% CAGR over six years, which seems well aligned with the Study's estimates.
  4. Based on input from stakeholders, uptake of HSPA devices has been significantly more than initially expected. Also, a significantly larger percentage of data traffic is carried over the HSPA networks than initially anticipated.
  5. Smartphone-to-other-mobile-device breakdown: According to CWTAFootnote 10, mobile broadband subscriptions in Canada comprised 5.68 M as of June 2010, which represents 24% of total wireless subscriptions. Of the total mobile broadband subscribers, 86% were subscribers with a smartphone voice/data plan, and 14% were subscribers of data plans exclusively with a mobile Internet stick.
  6. Smartphone data usage: Based on secondary sources and various papers, six-year projections (2010-2015) of smartphone data usage show an average CAGR of approximately 47%. Red Mobile's projections were conducted separately for entry level versus high-end smartphones and indicate a six-year (2010-2015) CAGR for data usage of 34% for entry-level smartphones and a 75% CAGR for high-end smartphones. These align with an average CAGR of 47% across all types of smartphones.

As a sense-check of these growth rates, a brief comparison was conducted of the Study's forecasts for growth of cellular traffic against those published by other parties in the industry, including Cisco (VNI), the GSMA, FCC, stakeholder input and submissions made to Industry Canada for the 700 MHz Consultation. From this review, it can be concluded that, while, the growth rate used here could be viewed as being on the high side, there are goo d arguments for assuming a high rate of traffic growth over the period through to 2015, and the projections are greatly out of line with those used in other sets of forecastsFootnote 11.

Key Assumptions and Relationship between Service and Spectrum Demand

In this section, the main assumptions used in analyzing how subscribers and traffic changes translate into changes in the demand for cellular spectrum are summarized below.

  • Spectral Efficiency of Cellular Technologies. There are major gains in spectral efficiency as networks move from older technologies, such as GSM, to newer ones, such as HSPA and LTE. There are also some gains over time as technologies make incremental improvements in their spectral efficiency of the deployed networks.

The table below shows the assumptions made regarding the spectral efficiency and frequency reuse of the various technologies over the time period of the Study. The spectral efficiencies are per sector.

Table 6.1.2 — Spectral efficiency and frequency reuse of the various technologies over the time period
Technology Frequency Reuse Factor Combined effect of (i) Frequency Reuse and
(ii) Link Spectral Efficiency in Bits / Sec / Hz
2007 2008 2009 2010 2011 2012 2013 2014 2015
2G - GSM 9 0.06 0.06 0.06 0.06 0.06 0.06 0.06 0.06 0.06
2.5G - GPRS / EDGE 9 0.11 0.11 0.11 0.11 0.11 0.11 0.11 0.11 0.11
3G - WCDMA (UMTS) 1 0.17 0.17 0.17 0.3 0.42 0.55 0.67 0.8 0.92
3.5G - HSPA 1 0.5 0.5 0.58 0.68 0.76 0.82 0.86 0.93 1.01
4G - LTE 1 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.35 1.4
2G - CDMA 1 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17
2.5G - CDMA 1xRTT 1 0.29 0.29 0.31 0.34 0.36 0.37 0.38 0.4 0.41
3G-3.5G -
EV-DO
(incl. Rev. A)
1 0.5 0.5 0.58 0.68 0.76 0.82 0.86 0.93 1.01

Source: PA Consulting Group


  • Numbers of sites and growth. Site and sector counts assumptions have been obtained by comparing a wide range of data sources, including online sources (Industry Canada's TAFL data, the Loxcell and Ertyu Web sites), as well as highly confidential discussions with the operators themselves. By the end of 2010, it is assumed that, in general, the operators were approximately midway through their rollout of HSPA networks. For the projections, the Study assumes that this rollout will be completed in all areas of medium and high demand by 2015. And that the headline site counts of each operator will continue to grow at 5%-10% per annum, as has occurred over the last 5-10 years, and that LTE achieves a level of rollout similar to that of HSPA in 2010 cell/sector counts are not disclosed, as doing so by technology would identify confidential operator information.
  • Offload of traffic to Wi-Fi. Much recent interest has been shown in offloading cellular traffic to Wi-Fi, with some operators internationally reporting 20% to 40% of the traffic in certain areas being offloaded. It can be assumed that, by 2015, approximately 8% of the projected cellular data traffic is offloaded by the operators (not offloaded directly by the users) to Wi-Fi and other short-range wireless technologies. This does not include any "offload instigated by the end user." For example, if a subscriber has a 1GB cap on their mobile data allowance, and they are close to that cap, they may elect to instead use the same device to generate traffic over Wi-Fi and other short-range technologies. This is counted as end-user demand for the Wi-Fi service, but it is not counted toward the projections of traffic or subscribers for cellular services.
  • Dimensioning of Networks, and, hence, the required capacity and spectrum, for the busy hour. It is assumed that spectrum is required to be able to carry the traffic offered in the busy hour, and that this traffic is 3.5 times that of the average hour (24x365).
  • Dimensioning of networks and spectrum requirements to provide the required quality of service. Simply dimensioning the spectrum for the busy hour is not sufficient. Additional spectrum is required, in order to handle variation in the traffic, for example:
    • Subscribers and traffic are not evenly spread across sites and sectors.
    • A certain amount of overhead may be required in order to provide high burst rates for individual subscribers when required, e.g. downloading a file or browsing a Web page.
    Therefore, the required capacity of the networks and, hence, the demand for spectrum is scaled-up by a further factor, typically 1.75, to allow for this variation in demand over space and over short-term (i.e. seconds, rather than over the course of a day or week) timescales.
  • Spectrum pairing. It is assumed that all cellular traffic is carried over paired spectrum across the Study period, rather than assuming that, for example, LTE will be making significant use of unpaired spectrum by 2015 to reduce the spectrum requirements of the networks.
  • Current approximate mix/routing of cellular traffic to technologies. General evidence available for Canada, alongside that from comparable markets, has been used to calibrate the current routing of cellular traffic to GSM/HSPA/CDMA. It has been assumed that, by mid 2011, the vast majority of data traffic was being routed over HSPA.
  • Downlink/Uplink mix of traffic. For Cellular data traffic, changes in the ratio of Uplink to Downlink traffic are assumed.
    • For 2010, the assumption is that Downlink traffic is 87% of the total, i.e. a UL:DL ratio of approximately 1:7.
    • For 2015, as user-generated content and file sharing become more prevalent over time, the assumption is that downlink traffic is 80% of the total, i.e. a UL:DL ratio of 1:4.
  • Allocation of cellular demand to neighbourhood types, and allocation of sites and sectors to neighbourhood types. The modelling includes an assessment of the differences between seven different types of local areas, according to the density of demand and population (daytime and nighttime), and makes some assumptions regarding where the cell sites are relative to where the traffic is. Broadly, the analysis assumes that deployment of sites and sectors follow traffic needs in urban areas and in dense rural areas. Additional sites, often with just one sector each, are deployed in rural areas in order to provide coverage.

The modelling has been informed by the evidence available, regarding the distribution of cellular sites and sectors across Canada, based, in particular, on the central (most dense) areas of major cities, such as the Greater Toronto Area (GTA) and Greater Vancouver Area (GVA).

This evidence confirms that, in Canada, areas where there is a high density of population and cellular traffic, operators install a higher density of sites and sectors to provide additional capacity. This relationship appears to be close to pro-rata across the areas where 95% of Canadians live and work, i.e. downtown areas of major cities, urban areas, including suburbs, and the most densely populated rural areas.

In the lower density rural areas, cell sites are mainly required to provide coverage, rather than capacity, and the demand for spectrum in these areas is somewhat less than it is in the urban areas.

Finally, in the sparse rural areas of the lowest population density, cellular coverage may be patchy, and the level of traffic is lower. Demand for spectrum is reduced still further.

The model uses assumptions that reflect these findings. It makes allowances for orders-of-magnitude differences in population density across the neighbourhood types, and, for urban areas, it uses a distribution of sites and sectors that is largely (but not wholly) driven by population density.

The model also makes allowance for the movement of subscribers between the neighbourhood types during the day and week, which further inflates the demand for spectrum, as downtown cells have heavier loads during the working day, and suburban cell sites have heavier loads in the evenings. The assumption is that this tidal flow scales up the spectrum requirement by a further factor of 1.2.

The foregoing assumptions are broadly consistent with those used for the UK study for Ofcom (looking at the demand for spectrum 2010-25) and with similar studies undertaken for other clients in Europe, Africa and Asia.

Demand for Spectrum

The main impact of the calculations is to convert the growth in traffic and subscribers into the growth in demand for spectrum as follows:

  • Over the period between 2010 and 2015, there is a 30-fold growth in data traffic per device.
  • The continuing shift of traffic from GSM and CDMA to HSPA and LTE gives a large improvement in spectral efficiency. GSM and CDMA require 3-16 MHz/Mbps/sector; whereas HSPA and LTE require only require 0.7-1.0 MHz/Mbps/sector.
  • Some other factors have a modest mitigating effect on the demand for spectrum; for example, increases in number of sites and sectors, and some offload of traffic to Wi-Fi, and a slight reduction in the ratio of downlink-to-uplink data traffic.

The chart below shows the growth in the demand for spectrumFootnote 12 for cellular, over the 2007-15 period.

The substantial growth in traffic that is forecast for the period of 2010-15 translates into considerable growth in the demand for spectrum — from 55 MHz in 2010, to 190 MHz in 2015.

Figure 6.1.10 — Cellular demand for spectrum, by technology

Cellular demand for spectrum, by technology (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling

Description of Figure 6.1.10

This chart provides the annual cellular demand for spectrum in MHz required to serve the offered traffic. The chart spans from 2007 to 2015 and is broken down by technology. In 2007, the spectrum demand comes from 2-2.5G CDMA/1xRTT and 2-2.5G GSM/GPRS/EDGE, each making up about 50% of the total demand of just over 40 MHz. This continues to be the case, with 3.5G HSPA slowly introduced in 2009. By 2011, 3.5G HSPA accounts for ore than half the demand and continues to increase beyond. 4G is slowly introduced in 2012, and by 2015, the demand is composed of a small percentage from 2-2.5 G technologies (about 2%), about 24% from 4G LTE, and 74% from 3.5G HSPA.


The demand for spectrum to carry cellular traffic over the period of 2010-15 is projected to be quite volatile as networks and customer devices migrate from the widely deployed 2G-2.5G technologies to 3.5G (HSPA), and, by 2013-2015, they begin to migrate to 4G (LTE).

The high growth in cellular traffic, with data traffic doubling every 12-15 months, does not simply translate into a pro-rata increase in the demand for spectrum.

A main reason for this occurrence is that the newer technologies offer greater spectral efficiencies than the legacy ones.

The secondary reasons include:

  • Rollout of HSPA is completed;
  • There are some modest increases in the number of base stations and sectors, in areas of high-spectrum demand. Micro-, femto- and picocell deployments pick up momentum.
  • There is some offload of cellular device traffic to Wi-Fi hot spots/short-range wireless.

This result — i.e. a significant growth in data traffic, that causes some growth in spectrum demand, albeit slower than the growth in data traffic — warrants some consideration of the true drivers at work.

One important part of the picture is that, over the last five years, network operators have invested in the capability to handle this growth. Taken together with a migration to LTE, and some continuing trends, such as offload to Wi-Fi, and a reasonable growth in the number of sites, the networks have invested to ensure that they are well placed to serve the growing demands placed on them by end users.

Assessment of Alternative Scenarios

The projections given above are for Scenario 1 (Business as Usual (BAU)), which constitutes the central scenario considered for the Study. However, two other scenarios have also been modelled to give an indication of some of the possible alternatives:

  • Scenario 2 – Wire-Free World (WFW)
  • Scenario 3 – Low Investment (LI)

The two figures below compare the three scenarios. The first chart shows the assumptions regarding traffic growth, and the second chart shows the projections for spectrum demand.

Figure 6.1.11 — Cellular demand (traffic), by Scenario

Cellular demand (traffic), by Scenario (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling

Description of Figure 6.1.11

This chart presents the annual total offered traffic for cellular in GB/mo for each of the three scenarios (BAU, WFW and LowInv). By 2015, the offered traffic totals are as follows: for WFW – over 140,000 GB/mo; for BAU, about 75,000 GB/mo; for LowInv, about 70,000 GB/mo.


Figure 6.1.12 — Cellular demand for spectrum by Scenario

Cellular demand for spectrum, by Scenario (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling

Description of Figure 6.1.12

This chart provides the total cellular demand for spectrum in MHz by scenario. In 2010 demand for each scenario is at about 50 MHz. Demand for spectrum for each scenario increases steadily each year and by 2015, demand for spectrum is as follows: for WFW – over 240 MHz; for BAU, over 180 MHz; for LowInv, over 275 MHz.


In Scenario 2, Wire-Free World, traffic in 2015 is projected to be roughly twice what it is in Scenario 1.

This translates into a further growth in spectrum demand, some 25% to 30% higher than in Scenario 1.

The effect of the traffic growth is partly offset by faster technology enhancements: greater improvements in spectral efficiency, replacement of devices, increased offloading to Wi-Fi, and further growth in base stations and sectors.

Scenario 3, Low Investment, uses the same set of traffic projections as in Scenario 1, but with somewhat lower levels of investment in new networks and devices.

This translates into much higher spectrum demand, with a figure for spectrum demand in 2015 that is slightly higher than that in Scenario 2 and some 50% higher than the figure from Scenario 1.

The cumulative effects of the lower degree of investment are significant even over the 2010-2015 period. The main components that impact demand are:

  • Slower rollout of LTE;
  • Slower growth in the number of base stations and sectors;
  • Slower churning of consumer devices and technologies.

Sensitivity Analysis: Assessing Spectrum Demand Using Alternative Metrics

Throughout this Report, the primary analysis of the projections for spectrum demand uses a "Traffic" view of the demand for spectrum. In other words, it focuses on what spectrum is required, in order to carry the offered traffic over the existing (or likely future) network. It considers:

  • What spectrum is required to serve the Offered Traffic;
  • Dimensioning for the Busy Hour (BH), with a typical Quality of Service (QoS);
  • A location of high-spectrum demand (highest of six or seven neighbourhood types, including city downtown, dense urban, suburban, dense rural, sparse rural) rather than a simple average across the country.

For cellular, two additional views of the demand for spectrum were analyzed, from the perspective of practical demand based on network operations, rather than just being based on user demand:

The two alternative definitions of "demand for spectrum" that are used in this Report are:

  • Allowance for channels: Demand for spectrum after allowing for minimum channel widths for each network operator in each band. This is calculated using appropriate detailed assumptions, regarding the minimum channel width for each technology, and the number of operators using each technology in each band in the same local area. Assumptions regarding minimum channel widths take into account, where appropriate, the needs for larger channels if advances in spectral efficiency are to be achieved.
  • Allowance for other practical issues / timing rules: This same demand, after allowing further headroom for some of the practical issues of network management, on the grounds that operators cannot precisely predict the timing of changes in demand, and associated shifts. Operators need sufficient spectrum to cope with fluctuations in market share or changes in demand occurring faster or slower than expected. The analysis allows a timing margin of +/- two years for the demand of spectrum for each individual technology.

These alternative views of the demand for spectrum are shown in the following two charts.

Figure 6.1.13 — Cellular demand for spectrum: Alternative metrics for "Demand"

Cellular demand for spectrum: Alternative metrics for Demand (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling

Description of Figure 6.1.13

This chart shows the cellular demand for spectrum under the BAU scenario. As in the previous chart, demand under this scenario increases steadily from about 50 MHz in 2007, to about 180 MHz by 2015. The chart also provides lines to indicate the increased demand as a result of adding operational constraints. In 2015, after applying minimum channel widths, the demand rises from 180 MHz to 300 MHz, and after applying busy hour traffic, the demand rises to about 370 MHz.


The solid bars represent the primary (traffic-oriented) projections of demand, as shown previously, allowing for busy-hour traffic and for an acceptable Quality of Service.

The lines show the two alternate measures of demand: the solid line shows the demand after allowing for minimum channel widths, and the dotted line shows the demand after making an allowance for "timing rules". The alternate measures reflect a higher need for spectrum to allow for these parameters.

The driver of the large increase in 2012 is the additional deployment of LTE, with sufficient channel widths to get proper benefit from the gains in spectral efficiency that it offers.

The subsequent changes over the period 2013-15 are due to a mix of falls in the spectrum, required for older technologies, and rises in the demand for spectrum for HSPA-enabled devices use newer technologies as traffic shifts away from GSM/CDMA. There is little or no corresponding growth in the demand for spectrum for LTE, because most of, or all the, traffic can still be carried by a single 2x 10 MHz channel per operator.

The next two charts show how the demand with operational constraints looks in each of the other two scenarios.

Both scenarios have higher future projections for the primary measure of demand or spectrum.

In each scenario, the extra demand for spectrum translates additively, not multiplicatively, into higher figures for the alternative measures of spectrum demand.

Figure 6.1.14 — Cellular demand for spectrum: Alternative metrics for "Demand": Scenario 2 (Wire-Free World)

Cellular demand for spectrum: Alternative metrics for Demand: Scenario 2 (Wire-Free World) (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling

Description of Figure 6.1.14

This chart shows the cellular demand for spectrum under the WFW scenario. Demand under this scenario increases steadily from about 50 MHz in 2007, to almost 250 MHz by 2015. The chart also provides lines to indicate the increased demand as a result of adding operational constraints. In 2015, after applying minimum channel widths, the demand rises from about 250 MHz to 350 MHz, and after applying busy hour traffic, the demand rises to over 450 MHz.


Figure 6.1.15 — Cellular demand for spectrum: Alternative metrics for "Demand": Scenario 3 (Low Investment)

Cellular demand for spectrum: Alternative metrics for Demand: Scenario 3 (Low Investment) (the long description is located below the image)

Source: based on Red Mobile and PA Analysis, and PA PRISM Modelling

Description of Figure 6.1.15

This chart shows the cellular demand for spectrum under the LowInv scenario. Demand under this scenario increases steadily from about 50 MHz in 2007, to over 275 MHz by 2015. The chart also provides lines to indicate the increased demand as a result of adding operational constraints. In 2015, after applying minimum channel widths, the demand rises from about 275 MHz to 375 MHz, and after applying busy hour traffic, the demand rises to over 500 MHz.


6.1.5 Considerations beyond 2015

Although the time period beyond 2015 is outside the scope of this Study and demand has not been modelled, some commentary and insight is possible from extrapolating the trends for 2010-15, and considering whether the reasons for the observed growth in the demand for spectrum are likely to remain in force.

Considering the four factors that have increased the efficiency of cellular networks, namely:

  • Gain in spectral efficiency from newer technology;
  • Operators completing the rollout of networks based on newer technology;
  • Increases in the numbers of base stations/sectors;
  • Offload of some traffic to fixed Wi-Fi/short-range wireless.

It seems likely that most of these forces will continue beyond 2015; the first of these will be weaker, the others about the same or slightly stronger than they were in 2010-15. It, therefore, seems plausible that, over the time periods of 2015-2020 and 2015-2025, a considerable part of the demand growth in those time frames will translate into continued growth in the demand for cellular spectrum, as there appears to be less scope for improvements in cellular spectral efficiency moving forward, as systems are more constrained by the Shannon Limit.

It is certainly possible to construct a scenario where the demand for spectrum does not continue to grow rapidly beyond 2015 — for example, if cellular traffic growth slows down to merely doubling every two or three years, and/or if the cost of adding new base stations in congested neighbourhoods declines substantially, and/or if offload to Wi-Fi becomes routine for stationary or nomadic users, so there is less pressure for operators to exploit each site to the maximum.

However, these outcomes seem far from certain, and it seems possible that demand for spectrum beyond 2015 will continue to grow, fuelled by traffic growth, and will be only partly offset by the four factors that are improving spectral efficiency of the cellular networks.

6.1.6 Conclusion

The Study has modelled three spectrum demand scenarios: Business as Usual, Wire-Free World and Low Investment, as shown in Figures 6.1.13, 6.1.14 and 6.1.15. The projections for the main traffic-oriented measure of spectrum demand climb from a current level of 50-60 MHz, reaching 190 MHz by 2015, or approximately 250 MHz in the alternative scenarios.

The analysis has also quantified the additional spectrum overhead required to handle minimum channel widths, and also to allow for the challenge of forecasting exactly when and how quickly demand will grow and shift to new technologies.

This typically adds a requirement for a further 100-150 MHz of spectrum to the current demand, increasing by 150-200 MHz once LTE is launched in the 2012-2013 timeframe.

This additional spectrum overhead is an additive factor (e.g. +150 MHz) rather than a multiplicative factor (e.g. x2). It is not particularly sensitive to the traffic projections; instead, it is driven by the degree of proliferation of bands, technologies and licencees.

The Study projects that by 2015 Canada will need a total of 300 MHz to 500 MHz of cellular spectrum, including the overhead for channel widths and variance in the timing of growth in traffic and migration to new technologies to accommodate the service demand. The exact figure depends on

  • which scenario unfolds;
  • how much provision is made for new entrants and for infrastructure competition;
  • and, of course, depending on the many other factors covered by the modelling and the assumptions, such as the rollout of additional sites and sectors and the degree to which operators make efficient use of their spectrum resources.

In reality, there is likely to be somewhat of a balancing feedback loop in the demand for spectrum;

  • Large amounts of underutilized cellular spectrum tend to lead to operators adding fewer sites and sectors, and/or pricing to attract customers.
  • Highly utilized spectrum tends to lead to operators adding more micro and picosites and sectors to maximize frequency reuse, more offloading of traffic, less liberal "fair use" policies, and pricing to discourage very heavy use of the networks.

So, there are good grounds to expect something analogous to Parkinson's LawFootnote 13: i.e. that spectrum demand will expand — or contract — to fill what is available. And, this is likely to somewhat reduce any imbalance between supply and demand for spectrum, which is, in turn, likely to trade these off against other cost and benefits, such as changes to fair use limits or to data tariffs.

Date modified: