Research In Motion — Digital Economy Strategy Consultation Submission: Mobile Broadband Capacity Constraints And the Need for Optimization

Une partie des informations de ce site Web à été fournie par des sources externes. Le gouvernement du Canada n'assume aucune responsabilité concernant la précision, l'actualité ou la fiabilité des informations fournies par les sources externes. Les utilisateurs qui désirent employer cette information devraient consulter directement la source des informations. Le contenu fournit par les sources externes n'est pas assujetti aux exigences sur les langues officielles, la protection des renseignements personnels et l'accessibilité.

Soumis par RIM 2010-07-14 08:04:27 HAE

Thème(s) : L’acquisition des compétences numériques

Sommaire

During an announcement on July 6, 2010 at the world renowned Perimeter Institute, Prime Minister Stephen Harper made the insightful comment that "science drives commerce." We could not agree more. The viability of Canada's economy relies on advances in scientific knowledge, research, and development.

Research In Motion Limited (RIM) welcomes the opportunity to contribute to Industry Canada's public consultation on Canada's Digital Economy Strategy. The Government's efforts to create an effective and sustainable strategy are paramount to the country's success in the global marketplace.


Soumission

About RIM

RIM is a leading designer, manufacturer, and marketer of innovative wireless solutions for the worldwide mobile communications market. Through the development of integrated hardware, software, and services that support multiple wireless network standards, RIM provides platforms and solutions for seamless access to time-sensitive information including email, phone, SMS messaging, Internet and intranet-based applications. RIM technology also enables a broad array of third-party developers and manufacturers to enhance their products and services with wireless connectivity to data. RIM has the world's largest wireless design group dedicated to wireless data innovation. RIM's portfolio of award-winning products, services, and embedded technologies are used by thousands of organizations around the world and include the BlackBerry® wireless platform, the RIM Wireless Handheld™ product line, software development tools, radio-modems, and software/hardware licensing agreements. Founded in 1984 and based in Waterloo, Ontario, Canada, RIM operates offices in North America, Europe, Asia Pacific and Australia. RIM has extensive R&D and manufacturing sites in Europe, which provides substantial direct and indirect employment. For more information, visit www.rim.com or www.blackberry.com

Overview

During an announcement on July 6, 2010 at the world renowned Perimeter Institute, Prime Minister Stephen Harper made the insightful comment that "science drives commerce." We could not agree more. The viability of Canada's economy relies on advances in scientific knowledge, research, and development.

Research In Motion Limited (RIM) welcomes the opportunity to contribute to Industry Canada's public consultation on Canada's Digital Economy Strategy. The Government's efforts to create an effective and sustainable strategy are paramount to the country's success in the global marketplace.

Spectrum

RIM believes the issue of spectrum scarcity to be the single greatest challenge facing the wireless communications industry today. The world is facing an impending spectrum ‘capacity crunch' arising from the explosive growth in mobile communications and the increasing use of smartphones and attendant bandwidth intensive applications. This situation must be addressed urgently by both business and policy makers in order to enable the continued development of mobile and other forms of wireless communications. It must be addressed at two key levels:

  • Wireless hardware and software equipment must operate more efficiently so as to consume less spectrum. RIM and its BlackBerry technology is a world leader in this area;
  • More spectrum must be made available for mobile communications and fixed wireless solutions.

In addition to our comments, we have included a recent independent report entitled Mobile Broadband Capacity Constraints by wireless technology consultants Rysavy Research. The research gives detailed, impartial and opportune insight into key global trends in the mobile broadband industry; we hope you will find this useful.

Export Controls

It has become necessary to modernize the export controls regime for cryptography products and technology.

Cryptography is the critical underpinning of the modern digital economy. It permits the safe and reliable transmission of financial and other critical data over the Internet, protects the integrity of information and communications systems, and has become an indispensable tool for protecting privacy and intellectual property.

RIM urges the Canadian government to review its Cryptography Policy and ensure the legal and procedural frameworks are in place to make certain that Canadian exporters of cryptography are not disadvantaged as compared to their competitors in the United States, European Union and other Wassenaar Member countries.

People Talent

Across Canada, we need to encourage youth to pursue career paths that are aligned with emerging labour market requirements and will lead to greater economic prosperity for all of us. Instilling the excitement of discovery, invention, and entrepreneurship requires comprehensive programs and policies that provide a sufficient supply of researchers, developers, and technology-literate citizens to keep Canada's economy globally competitive. At the same time, the solution to the talent shortage in Canada needs to include the full participation of new immigrants.

Spectrum Bands Threatened by Impending Capacity Constraints

The Issue

Whilst it is difficult to predict future communications landscapes—especially where some of the most popular applications, such as user-driven video uploading and social networking were virtually unknown a decade ago—one trend is certain: the overwhelming, ubiquitous and inexorable move towards mobility and wireless access. On March 23, 2010 Ericsson announced that, for the first time, mobile data traffic surpassed voice on a global basis during December of 2009Footnote a. This historic milestone powerfully illustrates the overwhelming global shift towards online mobile communications.

Providing sufficient capacity to enable this quickly unfolding landscape has now become an overriding imperative that spectrum planning regimes must urgently address. This is because spectrum is a finite resource and capacity demands are beginning to constrain available supply, as the growing popularity of smartphones and consequent use of bandwidth and data intensive applications place extreme demands on existing networks. Rysavy Research is a respected US-based wireless analyst company, whose latest researchFootnote b reveals trends indicating that the surge in wireless data usage is leading to a ‘capacity crunch'.

Based on mobile broadband market trends, Rysavy Research estimates that typical operators are likely to find their available spectrum completely consumed in the next three to five years. These constraints will become apparent through dropped calls, web pages that load slowly and applications that are unable to extract requisite data. Industry Canada must therefore swiftly implement effective policies to deal with this impending crisis.

Action required

RIM believes that the challenge of impending capacity constraints needs to be addressed on two levels:

  1. Smarter hardware and software; and
  2. Focused spectrum planning and allocation policies.
1) Smarter hardware and software

Thoughtful engineering can conserve network bandwidth and defer the capacity crunch. From a commercial perspective, it needs to be recognized that carrying data over a network has a fixed cost and that, unless industry starts factoring in the cost of carrying data-rich applications, sufficient incentives will not be in place to make data applications and services more efficient.

The use of efficient, energy-conserving hardware and software technologies is crucial. Since the inception of the BlackBerry solution, RIM has made efficient use of spectrum resources through continuous innovation in the compression, rendering and transmission of data.. BlackBerry devices were originally designed to run on what were little more than glorified paging networks, and operates on the concept of "just-in-time data"", meaning that BlackBerry smartphones only download as many bits as are required to provide users with a rich experience. BlackBerry devices are designed to scale better, allowing for fast and efficient mobile e-mail, document attachment, instant messaging, browsing, and faster and more reliable applications. Research has proven that the BlackBerry service can send 11 times as many email messages per 500 Mbytes of data capacity than an iPhoneFootnote c. By using state of the art technology such as efficient file viewing, a BlackBerry smartphone sends only the most relevant email data without unnecessarily impacting a network operator's bandwidth constraints.

In terms of web efficiency, Rysavy Research reveals that network operators are able to support three BlackBerry browsing sessions for every one session on other platformsFootnote d. This efficiency also provides for greater reliability in times of congestion, such as peak demand periods or moments of crisis, when other forms of communication cease to function. An additional advantage of this efficiency is that BlackBerry handsets have superlative battery life: the more efficiently a device handles data, the less it needs to activate its radio, and the longer its battery lasts.

The larger picture is that these efficiencies are not only better for the environment and spectrum resource management, they also enable a very exciting roadmap for applications that can be built on top of this sophisticated, proven and efficient platform.

2) Focused spectrum planning and allocation policies

The fundamental problem remains the paucity of new spectrum being made available to support the unrestricted growth of mobile services. To date, industry has relied on workarounds such as Wi-Fi offloading and femtocells to ease this load. However, even with greater engineering efficiencies, the industry will still remain dependent on governments to make more spectrum available in order to satisfy the growing consumer demand for rich mobile products and services. In relation to spectrum planning and allocation, we believe that this situation gives rise to the following policy imperatives:

  1. More spectrum must be made available for mass-market wireless telecommunications. Authorities should closely assess market demands and trends and, where necessary, act pro-actively.
  2. Spectrum planning policies must promote an efficient use of spectrum. Interference management should strive for optimum precision, spectrum in popular bands must be utilised fully and, where necessary, re-allocation processes must allow for the expedited prior clearance of bands.

Outdated & Inefficient Export Controls Hurting Canadian Industry

The Issue

The need exists to modernize Canada's export controls regime for cryptography products and technology.

Cryptography is the critical underpinning of the modern digital economy. It permits the safe and reliable transmission of financial and other critical data over the Internet, protects the integrity of information and communications systems, and has become an indispensable tool for protecting privacy and intellectual property.

The importance of cryptography was recognized by the Government of Canada in its 1998 Cryptography Policy. Since then, encryption technology has become much more ubiquitous and can be found in many commercial ICT products, including BlackBerry smartphones and other BlackBerry products.

Canada's cryptography sector has been a tremendous success story in recent decades, in part due to Canada's cryptography policy, which includes a commitment that "Canada will take into consideration the export practices of other countries and the availability of comparable products when rendering export permit decisions."

To maintain that success in the face of global competition, it is imperative that Canadian exporters be on a level playing field with their competitors from other countries, facing a domestic export-control regime that is no more onerous than those faced by companies elsewhere, in accordance with Canada's stated policy. RIM encourages the Canadian government to adhere to this stated policy and to review and update the current Canadian framework and licensing practices as necessary to keep Canada competitive. This is particularly urgent given the recent reforms announced by the US government to their export-control regime for cryptography products and technology.

We have identified some specific examples in which export controls requirements on encryption products/technology are more stringent in Canada than in the US, for example, and have included them below to illustrate our concerns.

  • Intra-company transfers of commercial encryption products or technology across borders do not require an export license in the US, while in Canada an export permit is required.
  • The US has special licensing provisions for beta-software or pre-commercial exports to most countries.
  • When Canada issues an export permit for non-mass-market cryptographic software, the permit has a two-year validity period (or up to five years upon request for Open Policy Countries when there is a valid long-term contract).
  • The US has a set timeframe for review of commercial encryption export filings. Until recently, the US has had a set 30-day timeframe for review of commercial encryption export filings. The review period in the US is set by statute and typically the clock does not stop on this review period even if additional information was requested from a company filing for an encryption review. Recent changes in the US eliminate this review period for certain categories of products, including non-mass market software. In contrast, the review period in Canada if a company does not self-classify a product as not controlled is often longer than 30 days (particularly for non-Open Policy Countries) and can take up to several months for new products, causing delays for Canadian exporters of cryptographic products and technology. Current RIM products would not be subject to a 30-day review period in the US and could be shipped without delay. In Canada, while Canadian companies have the option to self-classify, if they choose to approach the Department of Foreign Affairs and International Trade (DFAIT) for advice, they may be subject to lengthy and unpredictable delays in obtaining advice or a permit.
  • In Canada, DFAIT requires that a signed end-use statement be provided along with export-permit applications for controlled cryptography products, while this requirement does not exist in the US and a number of other countries.
  • Other jurisdictions may not require reporting of exports as a condition of license or permit, while this is currently a condition that Canada attaches to broad-based export permits issued for commercial cryptography products.

Action required

RIM urges the Canadian government to review its Cryptography Policy and ensure the legal and procedural frameworks in place ensure that Canadian exporters of cryptography are not disadvantaged as compared to their competitors in the US, EU and in other Wassenaar Member countries. Below are some specific recommendations as to how to improve Canada's export controls system, and we urge the Canadian government to:

  1. Use all of the tools available to them to ensure Canadian developers and exporters of cryptography products and technology are not at a competitive disadvantage. This should include the introduction of general export permits (GEPs) for commercial cryptography products.
  2. Set longer validity periods for export permits, so that companies do not need to re-apply every two years for a new permit where there have been no changes to the cryptographic functionality of a given controlled products.
  3. Ensure no additional requirements in Canada vis-a-vis other Wassenaar Member countries (e.g. requirement for end use statements) for cryptography exports.
  4. Set straightforward service standards and realistic fixed timeframes for review of export permit requests.
  5. Implement changes to the Wassenaar Arrangement Export Control List so that Canadian companies can quickly take advantage of any new decontrols agreed at the international level.

This is a critical issue to address as Canada moves to implement a national digital economy strategy and seeks to attract to and retain in Canada research and development for ICT products.

On a related note, we understand that the Government of Canada is working on a Cyber Security Strategy, and we look forward to learning more about this initiative and contributing input as it is developed.

People are at the Centre of a Successful Digital Economic Strategy

The Issue

A truly digital economy requires both visionary producers and skilled end users. In short, people must be at the center of Canada's Digital Economy Strategy. Canada requires comprehensive support programs and policies that will provide a sufficient supply of technologically-literate workers and citizens to keep Canada's economy globally competitive. Canada's ongoing commitment to far-sighted public investment in universities, hospitals, and research institutions helps to attract world-class faculty, students, and research projects as well as create, attract and retain skilled workers. Federal and provincial governments must cooperate on these issues if the Digital Economy Strategy is to move forward expeditiously.

Enrolment in science, technology, engineering, and mathematics programs are declining while employment opportunities in these fields continue to grow. The choice to pursue a program of post-secondary education related to ICT would seem like a practical one, but it is a choice that Canadian students are making less frequently.

Action required

Across Canada, we should encourage youth to pursue career paths that are aligned with emerging labour market requirements and will lead to greater economic prosperity for all of us. We need to instill the excitement of discovery and invention in today's youth, by celebrating entrepreneurship in all its forms, and sharing the real-life success stories of Canada's technology innovators. Some initiatives to help engage K-12 student in technology include:

  • Reviewing the traditional learning system to make better use of ICT including a mix of face-to-face and online learning, and increasing the opportunities to use digital media in the classroom
  • Business visitation, innovator lectures and tours for teachers
  • Hands-on exploration of ICT
  • Integration of information on ICT industry trends
  • Connect curriculum to required ICT skill sets
  • Encouragement to explore ICT career opportunities
  • Connecting classroom learning to ICT-related programs and career opportunities

The issue of funding for post-secondary education and training is obvious: In order to create the workforce for a vibrant digital economy in Canada, students and teachers in our country's colleges and universities must have access to cutting edge facilities and equipment in order to reach their greatest potential.

Research councils and granting agencies must be funded beyond the modest growth they have seen of late. At Research In Motion, we enjoy the benefits of past investments in education, but we also face ongoing shortages of the sort of talent required to generate tomorrow's innovations. Canada must continue to invest in a culture of excellence. Investment must increase at all levels.

Research In Motion is also a long-time supporter of co-operative education, and is the largest private sector employer of co-op students in Canada. Over the years, this synergistic relationship has seen thousands of students gain valuable work experience and knowledge as our company was propelled forward by the energy and creativity of young minds. Co-operative education is a great example of an educational model that works. For this reason, Canada should take a more deliberate, systematic approach to co-operative education through the creation of a national system for ICT co-operative education programs and placements.

Recognizing Skills

The solution to the talent shortage in Canada will necessarily include the full participation of new immigrants trained outside of the country. Both immigration policy and professional licensing requirements should be modernized to recognize the shortage of highly skilled ICT professionals in Canada because cutting edge research and development will only occur where there are people with the skills to do it. Protectionist policies that inhibit the free movement of labour will only weaken Canada as capital moves where the talent is, and immigrants are prevented from participating to their full potential.


Footnotes

  1. a http://hugin.info/1061/R/1396928/353017.pdf (retour à la référence de note en bas de page a)
  2. b http://rysavy.com/Articles/2010_02_Rysavy_Mobile_Broadband_Capacity_Constraints.pdf (retour à la référence de note en bas de page b)
  3. c http://rysavy.com/Articles/2010_02_Rysavy_Mobile_Broadband_Capacity_Constraints.pdf (p.29) (retour à la référence de note en bas de page c)
  4. d http://rysavy.com/Articles/2010_02_Rysavy_Mobile_Broadband_Capacity_Constraints.pdf (p.27-8) (retour à la référence de note en bas de page d)

Table of Contents

Rysavy Research provides this document and the information contained herein to you for informational purposes only. Rysavy Research provides this information solely on the basis that you will take responsibility for making your own assessments of the information.

Although Rysavy Research has exercised reasonable care in providing this information to you, Rysavy Research does not warrant that the information is error-free. Rysavy Research disclaims and, in no event, shall be liable for any losses or damages of any kind, whether direct, indirect, incidental, consequential, or punitive arising out of or in any way related to the use of the information.

Executive Summary

Powerful smartphones, fast networks, compelling applications, and user awareness are causing a dramatic surge in the use of mobile-broadband technology. Previously relegated to business executives or vertical-market applications, wireless data is now experiencing mass-market adoption. The advantages are obvious – flexible lifestyles, greater productivity, and the addictive sensation of always being connected. This market growth comes at a good time for operators, who are seeing increasing data revenue compensating for declining voice revenue.

But there is a problem. There simply is not enough network capacity to address the emerging demand, and we are already witnessing the effects of network congestion, with many users complaining of slow network operation on some networks. Capacity is based on a number of factors, but foremost is the amount of spectrum available for broadband services. The FCC chairman himself recently stated that he saw the biggest threat to the future of mobile activity in America as the looing spectrum crisis.

What is primarily driving network usage currently is rapidly increasing smartphone penetration, at more than 25% now and ready to hit 50% within a year or two. In early days, people used mobile phones to access mobile-specific content, of which there was little. But today's phones can do so much more: browsing the Web at large, e-mail with attachment viewing, navigating with maps, video, social networking, banking, business information access, cloud computing, and entertainment. People love their smartphones, because a small handheld device gives them access to the same tools and information that previously required a desktop compute. And this is just the beginning. New platforms, such as netbooks, are also seeing strong initial adoption and are about to be followed by entirely new categories of devices such as mobile Intenet devices and smartbooks. Nobody can anticipate exactly how this world of new mobile computing devices will evolve, but the trends are clear: people desire powerful mobile computers with broadband connections.

If we were restricted to just mobile computing, application developers might design their apps with wireless capacity constraints in mind. But at the same time as mobile broadband is becoming ubiquitous, wireline broadband networks are becoming much faster, with technologies such as fiber to the home (FTTH) and next generation cable-modem technologies. These are providing throughput rates of tens of megabits per second, enabling applications not previously possible such as high definition video over the Internet. Users obviously would like to use the same applications on their mobile connections. Though that's not feasible, at least in large numbers of subscribers, users will certainly try.

To satisfy this quickly growing demand, especially since it will take five years or more to bring any new spectrum online, operators are using multiple strategies. One is building new cell sites. Spectrum reuse, which cellular technologies accomplish through the use of the same frequencies over and over in different cells is, in fact, the greatest determinant of overall network capacity. But building new sites is an expensive and time-consuming process. Offloading data onto other networks, such as Wi-Fi, is another option, and one that operators are pursuing aggressively. Femto cells could also eventually offload data in buildings, but the femto market has been slow to develop. New technologies, such as WiMAX and LTE, are spectrally more efficient than previous technologies, but not that much more, and wireless technology is approaching theoretical limits of spectral efficiency. Wireless network deployment in the 700 MHz band will provide a boost in network capacity, but it will be 2014 before these networks will be broadly deployed, and, even then, their capacity is quite finite.

All of these approaches, plus eventual new spectrum, will help address the demand. But even then, wireless capacity will remain constrained relative to demand. This is because augmenting capacity is only part of the answer. The other part is more efficient use of spectrum. It is imperative that mobile applications consume only the amount of bandwidth they really need. This is how solutions like BlackBerry provide a profound advantage, consuming significantly less data in applications such as e-mail and Web browsing. The benefit to operators is huge, as it means lower network costs and a greater number of users supported in the same amount of spectrum. And with pricing plans likely to move more to a usage-based model, users will benefit from lower monthly fees.

There is yet another consideration. Even a well dimensioned network will experience times of unexpected heavy usage, such as through dense user congregation or by a subset of users with high-bandwidth (e.g., video) applications. Operators may have sufficient spectrum, but in many markets have deployed only a limited number of radio channels for broadband. Or their backhaul may be constrained. Consequently, congestion is unavoidable. The effect on applications in these scenarios can be highly disruptive, resulting in timeouts and other failures, unless those applications are designed specifically for wireless connectivity. Again, BlackBerry has a significant advantage through its use of highly-optimized wireless-specific protocols. The result is greater reliability and availability, even under adverse conditions. Not only are the protocols more resilient, but with more efficient access, download times are faster, making it more likely to successfully complete data exchanges. Beyond this, BlackBerry offers multiple management options with which IT staff can control how much data BlackBerry devices consume. Tests have shown BlackBerry to be significantly more efficient for mail and web browsing. These efficiencies translate to significantly lower costs for users and operators.

The mobile broadband market is emerging as an extremely successful industry, but it is facing significant challenges that can only be met through a broad sweep of measures to augment demand. But, just as important, is efficient use of the network. This is the BlackBerry advantage.

Wireless networks inherently have far lower capacity than wireline networks. One fiber optic cable has greater data capacity than the entire RF spectrum. A shared, inherently unreliable medium like radio simply cannot match what wire can bring. And therein lies the problem. Just a smaller number of mobile users with bandwidth-intensive applications can consume the available wireless network capacity. We are not quite at the stage of capacity exhaustion, but we are seeing early instances of it, and analysis shows that the available capacity can be consumed by a relatively small percentage of high-bandwidth subscribers. Based on current trends in mobile broadband usage, a spectrum-demand model developed by Rysavy Research shows that many operators' spectrum could be consumed within three to five years.

Introduction

The mobile and wireless industries have succeeded beyond anybody's expectations. The smartphone has become the convergence point that brings together the capabilities of today's wireless networks, miniaturization of computers, innovative user interfaces, handheld operating systems, and vast numbers of applications. Other types of portable computers are also proving hugely successful, first with notebook computers, now with netbooks, and soon with new categories such as smartbooks and mobile Internet devices. Also coming online are electronic books, digital picture frames, cameras, and gaming consoles. The essential capability that makes all of these platforms so attractive to their owners is mobile broadband connectivity.

But there is a problem. Millions of new devices able to consume large amounts of data threaten to overwhelm the capacity of today's networks. How serious is this threat, and what are the consequences? Those are the topics of this report, sponsored by RIM, which begins with an overview of the growth in mobile broadband, the actual capacity of today's networks and how soon that capacity could be consumed, options available to operators to address demand, and the inevitability of network congestion and the effect on applications. The report demonstrates the advantages of BlackBerry in this context, including its e-mail efficiency, web browsing efficiency, resilient protocols, and management of data transmission. The report then quantifies the financial benefits of BlackBerry under different scenarios of adoption and usage.

Mobile Broadband Demand

To understand the impact of mobile broadband on wireless networks, one needs to quantitatively understand how much demand mobile broadband actually places on networks. There are several ways of doing this beginning with metrics on broadband growth in general, then mobile-specific metrics in particular, and by looking at the bandwidth requirements of different applications.

Broadband Growth

People are clearly drawn to broadband for the instant access to information, entertainment, web applications, and rich communications such as social networking. Cisco reports that the average broadband connection, on a global basis, already generates 11.4 Gbytes of Internet traffic per month, which is equivalent to 375 Mbytes per day.Footnote 1 This exceeds current average mobile usage, but there are a number of trends that indicate that mobile broadband usage will increasingly mirror usage in wireline networks:

  • Fixed Mobile Substitution. An increasing number of subscribers will use mobile connectivity as their only form of connectivity. For voice, this already is about one fifth of US households.Footnote 2
  • Netbooks. Usage behavior on netbooks and other emerging devices, such as smartbooks and Mobile Internet Devices, will be increasingly Web-centric, demanding constant connectivity.Footnote 3 Also, with their larger screens, data consumption from these devices will more closely mirror desktop or notebook computers rather than phones.
  • Smartphone Destinations Match Wireline Destinations. The most popular Web sites for wireless connections largely overlap those used with wireline connections. For example, the top ten US mobile sites for January to October 2009 were: Google, Yahoo! Mail, Gmail, Weather Channel, Facebook, MSN Hotmail, Google Maps, ESPN, AOL Email, and CNN News.Footnote 4

As for growth, Cisco predicts annual global IP traffic to double every two years through 2012.Footnote 5

Mobile Broadband Growth

Mobile broadband growth is even faster than wireline, because not only are individual users consuming ever more data, but the percentage of users using mobile broadband is increasing. The result is a huge projected increase in data consumption as shown in Figure 1, a Cisco projection of global mobile broadband traffic measured in petabytes (million gigabytes) per month. This growth is at a 108% compound annual rate over five years.Footnote 6

Figure 1: Cisco Global Mobile Broadband Data Projection
Figure 1: Cisco Global Mobile Broadband Data Projection

Consistent with this study, Coda Research Consultancy anticipates a 40-fold increase in traffic in 2017 over 2009.Footnote 7 Chetan Sharma shows even more accelerated growth with aggregate US mobile data reahing 20 exabytesFootnote 8 in 2013.Footnote 9 Would this much data overwhelm the carrying capacity of today's networks? The answer is yes.

Smartphones

Nowhere is mobile broadband visibly growing faster than with smartphones. From a historical perspective, smartphones weren't even possible before about 2002 when widespread availability of cellular-data became possible with General Packet Radio Service (GPRS). Before then, phones and networks were highly voice centric. Data capabilities since then have improved at a rapid rate with Enhanced Data Rates for GSM Evolution (EDGE), then 3G technologies like High Speed Packet Access (HSPA), Evolved Data Optimized (EV-DO), WiMAX, and 3GPP Long Term Evolution (LTE). Each wireless networking technology has enabled greater data consumption by phones. Coping with this rate of change, especially with data now outstripping voice in traffic volume, is extremely challenging for operators.

In looking at today's data usage by smartphones, Nielsen says that the average iPhone user consumes 400 Mbytes per month.Footnote 10 This is consistent with other industry data.Footnote 11 Smartphones already account for some 25% of phones today, on track to reach 50% within a year or two.Footnote 12

Bytemobile issued a report on the impact of smartphones on mobile networks showing how smartphone usage is beginning to approac laptop usage.Footnote 13 This includes touch smartphone browsing sessions of 38 minutes being approximately 63% of laptop sessions. On networks that Bytemobile tracks, on wireless networks not offering touchscreen smartphones, laptops account for nearly all wireless data usage. But for those networks offering touch smartphones, those phones account for 52% of usage. The report also shows that one video user consumes fifteen times more network bandwidth than a Web user. At this time, there are still ten to fifteen times more Web users than video users, but video usage is growing quickly as Web sites offer more and more video.

CTO Derek McManus of O2 in the UK stated "World-class smartphones have brought about an unprecedented demand on mobile data networks. Data on our network has increased 20-fold in the last year alone."Footnote 14 T-Mobile USA CTO Cole Brodman, at the Open Mobile Summit 2009, stated that Android users consume fifty times the data of other users.Footnote 15

Applications and Bandwidth Requirements

It is difficult to specify exactly how much bandwidth each application requires. Does a Web page need to download in one second, five seconds, or is ten seconds sufficient? Generally, the faster a network responds, the better. Nevertheless, one can recommend bandwidths that provide an experience that would satisfy the majority of people. Some applications, such as streaming audio or video, must have a certain amount of bandwidth; otherwise the stream can be interrupted.Footnote 16 The following table lists recommended bandwidths for different applications.

Table 1: Recommended Bandwidths for Different Applications
Application Recommended Bandwidth
Mobile voice call 6 kbps to 12 kbps
Text-based e-mail 10 to 20 kbps
Low-quality music stream 28 kbps
Medium-quality music stream 128 kbps
High-quality music stream 300 kbps
Video conferencing 384 kbps to 3 Mbps
Entry-level, high-speed Internet 1 Mbps
Minimum speed for responsive Web browsing 1 Mbps
Internet streaming video 1 to 2 Mbps
Telecommuting 1 to 5 Mbps
Gaming 1 to 10 Mbps
Enterprise applications 1 to 10 Mbps
Standard definition TV 2 Mbps
Distance learning 3 Mbps
Basic, high-speed Internet 5 Mbps
High-Definition TV 7.5 to 9 Mbps
Multimedia Web interaction 10 Mbps
Enhanced, high-speed Internet 10 to 50 Mbps, 100 Mbps emerging

Mobile voice and text-based e-mail communications require relatively little bandwidth, whereas high-definition video consumes more bandwidth than any other application. A high-definition YouTube video at 2 Mbps consumes as much bandwidth as 200 voice calls, and a normal-definition YouTube video at 1 Mbps consumes as much bandwidth as 100 voice calls.

Equating Capacity with Demand

This report now turns to the actual capacity available in today's networks.

Capacity Introduction

The question is how much capacity do today's mobile broadband networks really have? Answering this is a complex exercise for the following reasons:

  • Networks have a blend of technologies. At any moment in time, an operator has a mix of technologies, and even for the same technology there are ongoing improvements.
  • Users are mobile. It is hard to know how many actual users there are in any coverage area at any moment in time.
  • Cell site density varies. Operators attempt to deploy sites to achieve coverage objectives, but where they can actually place sites depends on many factors such as zoning restrictions, community acceptance, and availability of physical locations or structures to actually mount a tower or antennas. Operators also do not publicly disclose the location of their sites.
  • Efficiency/performance tradeoffs. The network configuration for highest efficiency (users per amount of spectrum) is not necessarily the same as the configuration for the best user experience.

One initial way of thinking about capacity is to look at how 3G networks have been deployed. Operators initially deployed High Speed Packet Access (HSPA) in 5 MHz + 5 MHz radio channels. HSPA uses a 5 MHz radio channel, so this means one channel was for the base-station to mobile-user (forward) direction and one channel was for the reverse direction. Based on a spectral efficiency of .5 bps/Hz, HSPA in initial deployments had a data capacity of 2.5 Mbps in each sectorFootnote 17 and EV-DO in a 1.25 MHz radio channel with the same spectral efficiency had a data capacity of 600 kbps. With improvements in radio technology, there will be a 50% increase in the capacity. But one can see how small a number of simultaneous YouTube viewers each at almost 1 Mbps can occupy the entire bandwidth of th data channel.

Looking forward to advanced technologies such as LTE, capacity will higher, but it will still be extremely limited compared to wireline capacity. Verizon Wireless' LTE network will operate in the 700 MHz band using 10 MHz radio channels. With a spectral efficiency of 1.5 bps/Hz, this delivers a sector throughput of 15 Mbps.

Meanwhile, there are about 1000 subscribers in the US for every cell site, which makes for an average of 333 subscribers per sector. If 10% of them were using the LTE data service, that would mean 33 users for the 15 Mbps data channel. Now, compare this with a subscriber of a wireline high-speed Internet service of 50 Mbps that is dedicated, and not shared, as shown in Figure 2.

Figure 2: Graphical Depiction of Wireline versus Wireless Capacity (Representative Scenario)
Figure 2: Graphical Depiction of Wireline versus Wireless Capacity (Representative Scenario)

The point is not that the wireless network cannot deliver extremely useful and valuable services, since it can, but rather that wireless capacity is inherently limited compared to wireline capacity.

One network in the US that has considerably more capacity is Clearwire's WiMAX network. This has been deployed in 30 MHz of spectrum, which is considerably more than what any 3G operator has deployed for data so far. As demand increases, Clearwire has indicated it can make up to 120 MHz of spectrum available.Footnote 18 Whereas many 3G networks place caps of 5 Gbytes on monthly data usage, the Clearwire network currently has no caps. Even the Clearwire network, however, cannot match the capacity of wireline access networks that are fiber oriented (e.g., fiber to the home).

Demand Projection

To know to what extent demand is likely to exceed capacity, Rysavy Research has developed a spectrum demand model. Though this is a first-order analysis that focuses on the most important variables, the model offers considerable insight. This section discusses how the model works, the assumptions used, and the predictions.

The model is based on an examination of how much data users consume in a month, which depends on the type of device. The model considers both smartphone platforms and other devices such as netbooks and notebooks. The model assumes that these "other" devices will consume considerably more data, since sessions are likely to be longer (e.g., watching longer videos) and because screen sizes are so much larger. A screen that is three times wider and three times higher has nine times the area and, assuming the same pixel density, can thus consume almost ten times the amount of data with graphical and video elements. On the other hand, there are many more users of smartphones than of these other devices.

The monthly usage amount with one value for smartphones and one value for other devices is a good starting point, because there are a considerable number of statistics available on monthly usage amounts, both for wireline networks and wireless networks. From that, the model calculates an average amount of data that a user consumes each day. The usage across a day, however, is not even. For example, Cisco reports for Internet traffic that 25% of the day's traffic is consumed in the busiest four hours.Footnote 19 With this information, the model calculates the bit-per-second load per broadband subscriber per device type during the busiest times of the day.

The model then multiplies the per-user traffic amount by the number of mobile broadband users in a typical cell sector to obtain a total data load in that sector Then, knowing the spectral efficiency of the technology being used, the model determines the amount of spectrum needed to support that load.

To reflect the growth in mobile broadband, the model then makes projections for the following items:

  • The increase in time of the amount of monthly data usage.
  • The increase in penetration of mobile broadband users for both smartphones and other device types such as notebooks, netbooks, and smartbooks.
  • Increasing spectral efficiency as operators deploy new technologies such as HSPA+ and LTE.

The results of the model are shown in the following charts. First, the model anticipates rapidly growing data consumption by smartphones and other devices, as shown.

Figure 3: Monthly Smartphone Data Consumption per Subscriber over Time
Figure 3: Monthly Smartphone Data Consumption per Subscriber over Time
Figure 4: Potential Monthly Data Consumption per Subscriber for Other Devices over Time
Figure 4: Potential Monthly Data Consumption per Subscriber for Other Devices over Time

One can reasonably extrapolate the following penetration rates of smartphone users and other device-type users as follows.

Figure 5: Penetration of Smartphones and Other Devices over Time
Figure 5: Penetration of Smartphones and Other Devices over Time

Based on these projections, the model calculates the amount of spectrum that a large operator would need to support this level of demand. The model assumes all services are supported in either 3G or enhanced 3G mode. In reality, any operators today have services deployed in less efficient 2G mode, meaning they would need more spectrum than shown. The model further assumes that the spectral efficiencies of the technology will improve as operators deploy technologies such s HSPA+, WiMAX and LTE. The model also accounts for spectrum required by voice services.

The following figure shows the amount of spectrum an operator would require in their busiest markets to meet the demand shown in the prior figures.

Figure 6: Projected Spectrum Requirements for a Large Operator
Projected Spectrum Requirements for a Large Operator

With many operators in the US holding 55 to 90 MHz of spectrum, one can see that total available capacity could soon be severely challenged. AT&T, in its most recent earnings presentation, stated that it is deploying third and fourth radio carriers to support some of its busiest markets, representing up to a 40 MHz spectrum commitment for mobile broadband.

It is important to note that the spectrum situation varies by operators. Some may experience shortages well before others depending on multiple factors such as the amount of spectrum they have, their cell site density relative to population, type of devices they offer, and their service plans.

Another important consideration is that the spectrum demand model looks at the amount of spectrum to carry the amount of projeced traffic. But a user's actual experienced throughput at this level of loading may be less than desired. To provide true broadband experiences with typical user throughputs in the 500 kbps to 1 Mbps range could require even more spctrum than anticipated by the model.

There is a range in the projected spectrum requirements depending on the different assumptions used. But even experimenting with a range of values for the key assumptions, one can come to conclude the following with a high degree of confidence:

  • Mobile broadband traffic has the potential to consume all available spectrum in the next three to five years, depending on the assumptions used.
  • Substituting wireless connections for wireline for large percentages of subscribers would require significant amounts of additinal spectrum.
  • As spectrum is consumed, it will result in congested situations.

Of particular significance is that usage from cell site to cell site is not uniform, and varies based on the location of users and their behavior. Even if a network has sufficient "average" capacity, these variations will result in some cells being overused and some cells being underused.

Operators acknowledge the problem. One leading operator recently stated to the FCC, "Wireless carriers continue to spend billions of dollars annually on infrastructure upgrades, but they will continue to face severe capacity constraints, particularly with demand growing far faster than anticipated."Footnote 20 Another large US operator stated "Wireless carriers face spectrum constraints, expanding yet highly unpredictable demand, interference hurdles, handset and device coordination requirements, and ongoing and fast-paced technological evolution."Footnote 21 This paper examines the overused congested scenario and its effects further below.

Another means of looking at spectrum versus demand is to consider the average data usage across all subscribers as per the prevous charts, and to compare that with the average capacity for each data user, assuming that an operator has 50 MHz of spectrum (25 MHz + 25 MHz) deployed for just broadband data services. This is shown in the following figure, which again assumes an operator's busy market. (Note, however, that most data services today are deployed in significantly less than 50 MHz, with 10 MHz or 20 MHz being much more common.)

Figure 7: Average Demand Per User Versus Average Capacity Per User
Average Demand Per User Versus Average Capacity Per User

Other Reports on Spectrum Demand

There is not that much public data available that quantifies spectrum requirements. One report is from the New Zealand Ministry of Economic Development.Footnote 22 This reports projects the spectrum requirements for different degrees of market penetration looking at different levels of mnthly data use per use. The technology is LTE. For 10 Gbytes per month per subscriber with 20% market share, the New Zealand data anticipates 61.7 MHz of spectrum. This is far greater than the amount of spectrum planned by various operators for their LTE deployments. Yet, 10 Gbytes per month is the amount of data being consumed today on wireline connections, and this amount of spectrum represents the amount needed for LTE to be able to compete with other broadband networks today. Over time, the spectrum requirement will only increase.

Table 2: Spectrum Requirement for LTE Deployment Relative to Monthly Usage and Market Share
Monthly Data Use Per User 20% Operator Share 30% Operator Share 40% Operator Share
5 Gbytes 30.8 MHz 46.2 MHz 62.0 MHz
10 Gbytes 61.7 MHz 92.5 MHz 123.0 MHz
15 Gbytes 92.0 MHz 139.0 MHz 185.0 MHz

Another report that has received a considerable amount of attention is in the International Telecommunications Union (ITU) report ITU-R M.2078, "Estimated spectrum bandwidth requirements for the future development of IMT-2000 and IMT-Advanced." This report uses a sophisticated model to predict spectrum requirements through the year 2020. It calls for total spectrum of 1300 MHz for the commercial mobile radio industry in 2015, as shown in the following table. This is about three times more spectrum than is currently available for commercial mobile radio service.

Table 3: ITU Projection on Total Spectrum Requirements
Type of Forecast 2010 2015 2020
Lower Adoption Forecast 840 MHz 1300 MHz 1720 MHz
Higher Adoption Forecast 760 MHz 1300 MHz 1280 MHz

Spectrum Deployment Considerations

The previous analysis is for the amount of spectrum required relative to network loading. For actual network deployment, however, there are additional considerations. One is that data and voice requirements are different, especially while voice is carried by circuit-switched means. Ideally, an HSPA operator can use one 5+5 MHz radio channel for voice-oriented traffic, and then an additional carrier, or additional multiple carriers, for data. This results in at least 20 MHz of desired spectrum for HSPA (2 X [5+5] MHz).

Another consideration is that while a particular coverage pattern may be effective for voice, it can be less than ideal for data. This is because the minimum signal quality that still works well for voice can be too low for best data performance, resulting in relatively low data throughputs. Today's data technologies employ advanced features such as high-order modulation, but these are only possible at higher signal to interference ratios. A common approach to address this disparity is to use a separate radio channel layer for addressing performance holes in what re called infill sites. Lower frequency bands might be used for coverage, but higher frequency bands for these infill sites. This means yet more radio channels.

A further consideration is how to deploy femto cells. While femto cells promise significant benefits in offloading data (as well as improving indoor coverage in some situations), they work best if operating on separate channels. In dense three-dimensional urban environments, two femtocell radio channels may be needed.

Figure 8 summarizes the idealized spectrum deployment for HSPA. With one data-oriented channel and one femto channel, this adds up to 25 MHz in each direction for a total of 50 MHz of spectrum.

Figure 8: Ideal Radio Channel Deployment for HSPA
Ideal Radio Channel Deployment for HSPA

For LTE, this same approach with 20 MHz radio channels results in a 200 MHz spectrum requirement for an ideal type of deployment, and this is for just one operator. Compare this with the 20 MHz that operators will be using for their actual LTE deployments.

If the operator had overlapping HSPA and LTE coverage, then they could use as much as 250 MHz. This is clearly far greater than what any operator has today. Consequently, congestion will be experienced even faster than predicted by a model that simply equates demand linearly against spectrum.

Another consideration in spectrum deployment is that new technologies, such as LTE, achieve highest efficiency with wider radio channels like 20 MHz in each direction, which represents a 40 MHz spectrum commitment. Furthermore, with contiguous spectrum, fewer guard bands are required. While future spectrum allocations will hopefully accommodate these requirements, deployments of new technologies in current bands will likely involve narrower channels in which these technologies will not acieve their full potential.

Managing Network Capacity

There are a number of ways that operators can manage capacity. These include using new spectrum, deploying new technologies, offloading data onto other networks, and through pricing plans.

More Cell Sites

Though wireless technologies have become more efficient with respect to the amount of bits they carry relative to amount of spetrum used, by far the greatest gain in overall network capacity has been from aggressive reuse of frequencies through smaller cell sites. This represents a million-fold gain since 1957.Footnote 23

Operators will continue to deploy more cell sites, but there are practical limits, including the difficulty of obtaining physical sites for towers and zoning restrictions. In addition, modern 3G and 3G+ sites can no longer be served with copper-based T1 or E1 circuits, but need fiber or broadband microwave backhaul connections. Obtaining more physical locations and connecting all these sites increases fixed network costs and complicates operation.

More cell sites will play a role in increased network capacity, but likely will not be a dominant factor.

Spectrum

New spectrum will be essential for the growth of the mobile broadband market. In the US, 354 MHz of spectrum has been allocated for commercial mobile radio service, including cellular, Personal Communications Service (PCS), Specialized Mobile Radio (SMR), Advanced Wireless Service (AWS) and 700 MHz bands.Footnote 24 But this is a small fraction of total spectrum required for the industry to address current market trends. Julius Genachowski, chairman of the Federal Communications Commission in the United States stated at the CTIA conference in San Diego in October 2009, "I believe the biggest threat to the future of mobile in America is the looming spectrum crisis."Footnote 25

The CTIA stated in its recent comments to the FCC on new spectrum allocation that another 800 MHz are needed in the next six yers.Footnote 26 AT&T indicated in its comments to the FCC that an additional 800 MHz to 1 GHz are required.Footnote 27

It is highly unlikely that this much new spectrum can be made available any time soon. The process of identifying, auctioning, licensing, moving incumbents and deploying new spectrum is a process that takes many years. Chetan Sharma, a wireless analyst, states that it takes seven to ten years to procure spectrum for wireless use.Footnote 28

Given this spectrum shortfall, other strategies are needed including making more efficient use of available bandwidth.

Backhaul and Core Network

As quickly as operators augment the amount of capacity in the radio link, they also have to be able to support the resulting traffic in the backhaul and core network. For the backhaul connection between cell sites and core network, traditional T1 circuits are insufficient, and operators have begun a massive upgrade to fiber and microwave radio connections. These upgrades are by no means complete, and may take years to complete. Similarly, core infrastructure elements such as Serving GPRS Support Nodes (SGSNs) are designed to handle certain levels of traffic (e.g., 1 Gbps), and operators must add infrastructure to handle higher traffic loads. There are no theoretical limitations on capacity in the backhaul and core network, but there are no guarantees either that these will be able to scale fast enough to match demand.

Using New Technologies

Technologies such as WiMAX and LTE increase spectral efficiency. Compared to typical HSPA and EV-DO deployments today, LTE will approximately double spectral efficiency. Further spectral-efficiency gains are also available for HSPA through approaches like MIMO. These gains in efficiency will be important, but it is imperative to realize that wireless technologies are reaching the theoretical limits of spectral efficiency, due to what is known as a Shannon bound, which dictates the maximum possible spectral efficiency for a specific signal-to-noise ratio.Footnote 29

Offloading onto Other Networks

Operators can also reduce the demand on their networks by off-loading onto other networks such as femtocells and Wi-Fi. This will definitely help, although there are limitations. First, both femto and Wi-Fi represent only a small percentage of the overall coverage area available to the user. Second, both presuppose an existing wireline broadband connection in home environments. Neither Wi-Fi nor femto are possible if a user is trying to make mobile broadband their primary source of connectivity.

To minimize interference, femtocells are easiest to deploy in separate radio channels, which mandate a certain level of spectrum commitment even before any spectrum benefits are actually realized. Once deployed, femtocells do provide a higher aggregate capacity for a certain amount of bandwidth than using those same frequencies in a maco cell. This is due to the higher level of frequency reuse, as shown in Figure 9. The question is whether a sufficient number of femto cells can be deployed fast enough to head off congestion issues. So far, deployment has been slow, and issues remain such as making the devices sufficiently easy to install and manage. Consequently, it is unlikely that femtocells will make a material difference in capacity consumption over the next three years.

Figure 9: Femto Cells Provide High Capacity Due to High Reuse
Femto Cells Provide High Capacity Due to High Reuse

Service Pricing Strategies

Service pricing and terms of agreement are another way that operators are controlling demand. Today's "flat-rate" plans usually have caps imposed, which for a laptop device is commonly 5 Gbytes per month. The projected monthly usage of broadband for non-phone devices shown above in Figure 4 is significantly higher than this. New lower-priced netbook plans have monthly limits of about 300 Mbytes, which represents a relatively low network load. Monthly laptop plans in the US are still priced at levels that discourage mass adoption. It is only with the current smartphone plans that are priced at relatively low levels and with which users are able to consumelarge amounts of data, that there is actually substantial growth in mobile-broadband usage.

If operators constrain usage through their service plans, there won't be necessarily be any so-called spectrum crisis. At the same time, however, if the limits are too restrictive, then people won't subscribe to the services in the first place. For platforms such as netbooks and smartbooks to become popular mobile broadband devices, they must be able to run the applications that users desire, including those that consume the most amounts of data such as video and social networking. This means that operators must deliver the amount of bandwidth these applications demand.

Pricing plans are the easiest way for operators to limit data consumption. Plans that are too restrictive, however, and that prevent users from doing much of what they do over wireline connections will significantly constrain market developmet and the data revenue opportunity.

Market Implications

Going forward, there are fundamentally two scenarios for the mobile-broadband market.

  1. Credible Broadband Alternative. Through additional spectrum and other means like data offloading, the industry provides sufficient capacity that mobile broadband networks support users bandwidth-intensive applications at attractive price points.
  2. Constrained Broadband Alternative. Due to delays in obtaining additional spectrum, operators are unable to deploy sufficient capacity to meet demand, and must rely on higher prices, limits on allowed applications, traffic shaping, and other means that result in mobile-broadband being a poor broadband alternative. Capabilities will be sufficient for phones and "light" Internet usage, but most subscribers will still need a fixed-broadband connection for data-intensive applications.

The Inevitability of Congestion and Impacts

The preceding discussion has demonstrated that with plausible increases in mobile broadband penetration and increases in data cnsumption per user that spectrum available for mobile broadband could be consumed within three to five years. But it will not take that long for the effects of congestion to manifest themselves, and, in fact, they already have on some networks. This section explains the reasons for congestion and the impacts on applications.

Reasons for Congestion

There are a number of reasons that congestion will occur on a localized, if not widespread, basis. Many of these are items have already been discussed and are summarized in Table 4.

Table 4: Reasons for Congestion Occurring
Source of Congestion
Explanation
Disproportionate network usage
Though there may be sufficient capacity on average, heavy usage (e.g., streaming video) by a subset of users in a specific coverage area can consume all available capacity.
Unpredictable user densities
Users are mobile and unpredicted large concentrations of users in a coverage area can quickly exceed local capacity.
Deployment on limited number of radio carriers
An operator may have sufficient spectrum relative to demand, but to minimize costs, deployment occurs on a radio carrier by radio carrier basis. The capacity of each radio carrier is relatively low and an insufficient number may be available for peak activity in some covrage areas.
Backhaul constraints
Even if the radio link has sufficient capacity, a large number of cell sites today have limited backhaul capability.
Cell site deployment restrictions
An operator may have spectrum and equipment available, but zoning or other restrictions may prevent installation of cell sites in needed areas.
Lag factor
Market growth can be faster in certain areas than the ability for an operator to upgrade their network to support demand.

As one operator states, "Accordingly, wireless providers face unique challenges in predicting how much capacity should be available or will be required at a particuar location, because the number of users at that location can change minute by minute. As many customers have experienced when shopping at a crowded mall or attending a popular sporting event, use of the network by other customers in a given location can dramatically impact the speed and availability of the network."Footnote 30

While congestion effects appear inevitable, it does not mean that networks will be unusable on a widespread basis. It is more likely that congestion effects will occur in certain locations at certain times, and not evenly across operators. The types of devices operators sell, the pricing plans they offer, and how many radio channels they have that support data, are all variables that will come into play.

An example is the most recent Consumer Electronics Show in Las Vegas this year where an operator experienced congested issues ad a spokesman for the operator stated, "In preparation for CES, we optimized our network in Las Vegas by significantly augmenting our network capacity. However, at an event such as CES, where large numbers of people in a dense area are using smartphones over finite spectrum, periods of network congestion can occur."Footnote 31

Application Effects

The effects that congestion has on applications are multifold. The core problem is that there are more packets to send or receive than opportunities to send them over the radio link. Hence, they get queued up within devices or within the network. In milder cases, the queuing simply results in eventual transmission with some delay. Everything works as desired, just a bit slower. In more severe cases, the delays increase until application performance is so slow that it becomes unusable. For example, a Web page will take a minute or more to load instead of a few seconds. In worst-case situations, the following can occur:

  • Packets are dropped. Most packet queues have a maximum size, and when that size is exceeded, the infrastructure simply discards packets. This can cause severe application malfunction.
  • Applications time out. Most applications that employ communications have maximum times that they will wait for communications to complete. When that time is exceeded, how applications respond depends on the application. Some will make another attempt at communications; others will report a failure. Some applications, developed for more stable wireline environments, will lock up, requiring users to terminate the application or even restart their computers.

TCP/IP Limitations

One of the challenges with wireless networking is that the most widely used communications protocol, TCP/IP, is not ideal for the wireless environment, particularly Transmission Control Protocol (TCP), which employs sophisticated timers and acknowledgment protocols to provide reliable communications across disparate networks. TCP was designed for wireline networks that behave differently with respect to packet delays, and hence, does not always handle communications optimally. It is for this reason that many wireless applications or middleware employ their own wireless-specific transport-layer protocols.

Vinten Cerf, one of the key engineers in the development of Internet technologies, states "Mobile operations are highly stressed. Mobiles are used where people congregate. In a sense, mobile is already a dense and hostile environment. We all know that when you drive around, coverage isn't very good. It's so hostile, it's clear that mobile could take advantage of these more-resilient protocols. TCP/IP is very brittle."Footnote 32

Figure 10 illustrates the problem. The fundamental issue is that a missing acknowledgment over wireline or fiber typically means that a router queue was full, and the router had no choice but to discard some packets, and therefore a retransmission is needed for the missing packet. In wireless, a missing acknowledgment is normal at times, and usually means that the user moved out of coverage or the network was temporarily congested. The wireless network stores data for the mobile until it returns to coverage at which time the network sends the delayed data o the mobile. This procedure confounds TCP's behavior and results in a flooding of the radio network with unnecessary re-transmissions.

Figure 10: TCP/IP Behavior Under Different Conditions
TCP/IP Behavior Under Different Conditions

Wireless-Optimized Applications

Most TCP/IP-based networking applications were never designed specifically for operation over wireless connections. While today's 3G and tomorrow's 4G networks can deliver IP packets reliably and efficiently, in a congested situation, or even with just a very weak radio signal, throughput rates can go down significantly, delays can increase, packets may be dropped, and connections can be lost entirely. Getting reconnected might be with a different IP address, which can confuse an application that is in mid-transaction. Moving rapidly such as in a train or car also stresses connections.

There are communications algorithms, however, that are designed to cope with such difficulties. Examples of systems that implement more robust communications include wireless e-mail systems; applications developed specifically for operation over wireless networks; and mobile middleware systems.

The methods used by wireless-optimized applications include:

  • Handling communications in the background so the user never notices any communications difficulties.
  • Longer timeouts so applications are more tolerant of delays.
  • Sending only the portions of files that mobile users need.
  • Compression to reduce the amount of data sent.
  • Caching so previously sent data can be reused.
  • Resuming from point of failure.
  • Ability to handle IP address change.

Benefits of optimized applications are multifold: they impose a lower network load; transactions complete more quickly and, hence, are more likely to succeed in congested situations; a lower amount of communications translates to better battery life; and users incur lower monthly service charges, especially as the industry moves in the direction of usage-based pricing models. RIM BlackBerry is a prominent example of an extremely efficient wireless application environment that benefits both operators nd its users.

The BlackBerry Advantage

There are multiple areas in which RIM BlackBerry provides advantages. One is in its efficient e-mail handling. Another is superior browsing efficiency. A third area is a policy management capability that allows managers to control bandwidth consumption of user devices. The last area is in the resilient communications protocols that BlackBerry uses.

E-Mail Efficiency

Rysavy Research has done a series of tests that compare BlackBerry e-mail efficiency with competing systems. Table 5 summarizes the test results.Footnote 33 The first column indicates the message size in bytes, the second column the type of attachment, if any, the third column the size of the attachment, and the fourth column the combined size of the message plus attachment. Subsequent columns show the results for Microsoft Direct Push and BlackBerry, listing the total number of bytes communicated over the radio interface as well as what percentage that number of bytes constiutes relative to the size of the message (plus attachment if any). Rysavy Research tested Direct Push using three devices: the Motorola Q9h, the HTC TyTN II (the AT&T version is called the "Tilt"), and the Apple iPhone 3G. Tests included Windows Mobile devices both with and without the Microsoft System Center Mobile Device Manager (SCMDM).

A percentage value greater than 100 percent means that the wireless e-mail system communicated more bytes than the original message size, whereas a percentage value lower than 100 percent means the wireless e-mail system communicated fewer bytes than the original message size. Lower percentage values represent better wireless efficiency.

Table 5: BlackBerry E–Mail Efficiency Comparison
        DP, Motorola DP SCMDM, Motorola DP, TyTN DP SCMDM, TyTN II DP, iPhone BlackBerry 9000
Msg Size Attach Type Attach Size Msg + Attach Sent OTA % Sent Sent OTA % Sent Sent OTA % Sent Sent OTA % Sent Sent OTA % Sent Sent OTA % Sent
5120 None 0 5120 12147 237% 15051 294% 11077 216% 13547 265% 18634 364% 3445 67%
10240 None 0 10240 14668 143% 17923 175% 13767 134% 16641 163% 22888 224% 6121 60%
20480 None 0 20480 20244 99% 24447 119% 19552 95% 23030 112% 32041 156% 11527 56%
136737 None 0 136737 134060 98% 152889 112% 133656 98% 154224 113% 180044 132% 68757 50%
5120 JPG 152148 157268 267867 170% 299132 190% 266817 170% 297764 189% 271454 173% 14613 9%
5120 PDF full 363139 368259 564186 153% 629022 171% 563110 153% 626489 170% 570480 155% 577645 157%
5120 PDF text 363139 368259 564186 153% 629022 171% 563110 153% 626489 170% 570480 155% 85856 23%
5120 Word Doc 511488 516608 594921 115% 667484 129% 593882 115% 665745 129% 601919 117% 41922 8%
5120 PPT file 966144 971264 1438081 148% 1599568 165% 1436744 148% 1598236 165% 1453720 150% 329103 34%
5120 Excel 51200 56320 34395 61% 41039 73% 33310 59% 39714 71% 40675 72% 10126 18%

In nearly all cases, BlackBerry was significantly more efficient than competing solutions. And in nearly all cases, BlackBerry sent less data over the air than the original file size. In some cases, the amount sent was only a small percentage of the original file. One method by which BlackBerry achieves gains in network efficiency is by having efficient file viewers, so that a user can view portions of a file without having to download a whole file. Another is by employing superior text compression algorithms that are twice as efficient as common approaches like GZIP.

Web Efficiency

BlackBerry users have the ability to choose the image quality settings for Web browsing including low, medium, and high. The default setting is medium. The lower the image quality setting, the less data is transferred from the BlackBerry Enterprise Server or BlackBerry Internet Server to the device and the faster the downloading time. The rationale for providing this quality setting is first that the original image quality is unnecessarily high in many cases, and second that BlackBerry users may prefer an acceptable degraded image quality for a faster browsing experience and lower data usage. RIM also uses advanced image compression algorithms that are more efficient than common JPEG and GIF image-compression approaches.

Rysavy Research also tested BlackBerry browsing efficiency using RIM's latest browser technology that is used in the 9700 Bold, with the results shown in Figure 11. Browsers compared included Windows Mobile 6.5 Internet Explorer, Windows Mobile 6.5 with Opera, Android, IPhone 3G, iPhone 3GS, Nokia N97, and Samsung Jet. These are randomly represented as browsers 2 to 8 in the figure.

Figure 11: BlackBerry Efficiency Relative to Other Mobile Browsers
BlackBerry Efficiency Relative to Other Mobile Browsers

The test measured the number of bytes communicated to download popular mobile Web sites. Because BlackBerry employs multiple mechanisms to reduce the amount of traffic including compression of objects, it is able, on average, to use only a third of the data consumed by other browsers. This represents a substantial savings in network traffic for operators, as well as lower costs for users on usage-based data plans.

BlackBerry Policy Management

Beyond efficient operation, the BlackBerry Enterprise Server provides a number of ways that IT staff can control data consumption on BlackBerry devices.Footnote 34 These include the following:

  • Restricting Web addresses that users can request when connecting to the Internet or an organization's intranet.
  • Specifying which Web address patterns users can and cannot use to access Web servers from the BlackBerry browser and other applications on their BlackBerry devices.
  • Controlling what media types can be accessed. For example, MP3 and video could be blocked.
  • Preventing users from accessing specific media file types that exceed a maximum value.
  • Specifying the maximum file size that can be downloaded.
  • Controlling the maximum file size for attachments that users can receive.
  • Preventing users from viewing certain attachment file formats.
  • Limiting the maximum file size for attachments that users can send.
  • Controlling which connections can be used for upgrading BlackBerry device software.

Protocol Resiliency

BlackBerry communications protocols were designed originally for slow networks, such as Mobitex, which ran at 0.1% of the throughput rate of today's fastest mobile-broadband networks with significantly higher communications delays. This heritage makes them highly bandwidth efficient and particularly resilient in congestion situations with today's modern networks.

RIM's network operations center (NOC) architecture means that there is only one connection for a device to maintain no matter how many services the device is communicating with, compared with competitors' approaches in which TCP connections need to be maintained for every service. This results in significant reduction in overhead for BlackBerry connection maintenance.

RIM's proprietary communication protocols further reduces the connection maintenance overhead compared with TCP.

Others agree with the BlackBerry advantage. A research note from Peter Misek at Canaccord Adams states that the BlackBerry service can send 11 times more emails per 500 Mbytes of data capacity than an iPhone. Alternatively, the Blackberry can deliver 7,000 Web pages versus 3000 for the iPhone."We believe that in a scarce spectral environment, RIM's NOC/BES architecture and compression technology will be worth tens of billions of dollars to global operators."Footnote 35

Financial Benefit of Efficiency

Given the cost of mobile bandwidth, greater efficiency translates directly to saving for both users and for operators.

User Savings

Users on "unlimited" plans may not see any costs savings from more efficient applications, but they still experience other benefits such as battery life, and more reliable operations. Meanwhile, those on usage-based plans could see significant savings, as shown in Table 6, which projects some typical usage scenarios. Common usage-based pricing plans today range from 50 cents to $2 per megabyte. BlackBerry e-mail traffic volume is typically less than one half of other solutions, and Web traffic is typically one third.

Table 6: User Email Cost Savings Based on Different Profiles
Email Usage Light User Medium User Heavy User
Emails per Month 100 300 1000
Size per e–mail body (Kbyte) 5 5 5
Percentage with attachments 10% 10% 10%
Average size attachment (Mbyte) 1 1 1
Message Body Consumption
Total body per month (Kbytes) 500 1500 5000
BlackBerry data volume (Mbytes) 0.335 1.005 3.35
Typical non–BlackBerry data volume (Mbytes) 1 3 10
Attachment Consumption
Total attachment volume (Mbytes) 10 30 100
BlackBerry data volume (Mbytes) 5 15 50
Typical non–BlackBerry data volume (Mbytes) 15 45 150
Total Data Volume (body + attachment)
BlackBerry (Mbytes) 5.335 16.005 53.35
Typical non–BlackBerry (Mbytes) 16 48 160
Monthly Savings with BlackBerry
Assuming 50 cents/Mbyte usage plan $5.33 $16.00 $53.33
Assuming $1/Mbyte usage plan $10.67 $32.00 $106.65
Assuming $2/Mbyte usage plan $21.33 $63.99 $213.30

One can do a similar savings analysis for Web browsing, as shown in Table 7. These are for relatively low-usage scenarios. User streaming video, for example, can generate far more traffic than what is calculated.

Table 7: User Web Browsing Costs Savings Based on Different Profiles
Web Usage Light User Medium User Heavy User
Pages Viewed Per Month 100 300 1000
Typical Mobile Web Page Sze (Kbytes)) 100 100 100
Total Volume Web Traffic (Mbytes) 10 30 100
Data Consumed on Mobile Device
BlackBerry (Mbytes) 4.2 12.6 42
Typical Non–BlackBerry (Mbytes) 12.5 37.5 125
Monthly Savings with BlackBerry
Assuming 50 cents / Mbyte usage plan $4.15 $12.45 $41.50
Assuming $1/Mbyte usage plan $8.30 $24.90 $83.00
Assuming $2/Mbyte usage plan $16.60 $49.80 $166.00

Actual savings clearly depends on actual volume and actual service plans. But across a wide range of assumptions, users will experience significant benefits from lower data consumption in usage-based pricing scenarios.

Mobile Broadband Network Costs

The savings are also dramatic for operators. Assuming a per megabyte cost of HSPA of $.03 EurosFootnote 36, the e-mail and Web browsing volume of a medium user from the above tables, 4 million BlackBerry users out of a subscriber base of 50 million, an operator could save more than 100 million dollars a year in operating costs.

Table 8: Savings to Operator from BlackBerry
Total subscribers 50,000,000
% Using Smartphones 20%
% Smartphones that are BlackBerry 40%
Total BlackBerry Devices 4,000,000
Montly Data Saved (MBytes, Medium User) 56.9
Total Monthly Data volume saved (Mbytes) 227,580,000
Operator cost per megabyte ($) HSPA 0.042
Montly savings $9,558,360
Annual savings $114,700,320

Conclusion

The mobile-broadband industry is experiencing tremendous success, yet its very success is undermining its ability to deliver a consistent, trouble-free experience. As the number of users increases with ever more demanding applications, it is inevitable that there will be ever more cases in which the volume of traffic in different coverage areas exceeds capacity, resulting in congested operation.

More efficient applications not only reduce the likelihood of congestion occurring in the first place, but they also are inherently more resilient, since they require less time and data to operate. They also reduce battery consumption, and most importantly for users, reduce costs, especially with usage-based pricing plans.

Beyond user benefits, greater application efficiency results in significant savings for operators including lower costs in the radio network, lower costs in backhaul, lower infrastructure costs and the need for less new spectrum.


Footnotes

  1. 1 Source: Cisco, "Cisco Visual Networking Index: Usage Study," October 21, 2009. (retour à la référence de note en bas de page 1)
  2. 2 Source: Nielsen, "Call My Cell: Wireless Substitution in the United States," September 2008. (retour à la référence de note en bas de page 2)
  3. 3 Source: PCCA meeting, "Emerging Mobile Platforms," November 5, 2009. (retour à la référence de note en bas de page 3)
  4. 4 Source: Nielsen, "Top Mobile Phones, Sites and Brands for 2009," December 21, 2009. (retour à la référence de note en bas de page 4)
  5. 5 Source: Cisco, "Approaching the Zettabyte Era", June 16, 2008. (retour à la référence de note en bas de page 5)
  6. 6 Source: Cisco, "Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update," February 10, 2010. (retour à la référence de note en bas de page 6)
  7. 7 Source: Coda Research Consultancy, "Mobile Broadband and Portable Computers: Revenue, User and Traffic Forecasts 2009-2017," July 19, 2009, http://www.fiercebroadbandwireless.com/story/report-laptops-netbooks-drive-exponential-mobile-broadband-growth/2009-07-19. (retour à la référence de note en bas de page 7)
  8. 8 A terabyte is 1000 billion bytes, a petabyte is a million billion bytes, an exabyte is one billion billion bytes, and a yottabyte is one thousand billion billion bytes. (retour à la référence de note en bas de page 8)
  9. 9 Source: Chetan Sharma, "Managing Growth and Profits in the Yottabyte Era," 2009. (retour à la référence de note en bas de page 9)
  10. 10 Source: Edible Apple, "Average iPhone user consumes 400MB of data every month," June 17, 2009. (retour à la référence de note en bas de page 10)
  11. 11 For example, 500 Mbytes/smartphone/month was quoted by an operator at a 3G Americas analyst meeting attended by Peter Rysavy, October 2009. (retour à la référence de note en bas de page 11)
  12. 12 Nielsen, "The Droid: Is this the Smartphone Consumers are Looking For?" November 11, 2009, http://blog.nielsen.com/nielsenwire/consumer/the-droid-is-this-the-smartphone-consumers-are-looking-for/. (retour à la référence de note en bas de page 12)
  13. 13 Source: Bytemobile, "Mobile Minute Metrics: November 2009." (retour à la référence de note en bas de page 13)
  14. 14 Source: GigaOm, "Like AT&T, O2 Pays the Price for Heavy iPhone Usage," November 19, 2009, http://gigaom.com/2009/11/19/like-att-o2-pays-the-price-for-heavy-iphone-usage/. (retour à la référence de note en bas de page 14)
  15. 15 Source: Fierce Wireless, "T-Mobile CTO: 40% of Q4 sales will be smartphones," November 4, 2009, http://www.fiercewireless.com/story/t-mobile-cto-40-4q-sales-will-be-smartphones/2009-11-04?utm_medium=nl&utm_source=internal. (retour à la référence de note en bas de page 15)
  16. 16 Some streaming applications, e.g., Netflix and Skype, can dynamically adjust to the amount of bandwidth available. Nevertheless, the effect can be quite disruptive as the system recalibrates itself. Most also maintain a buffer. (retour à la référence de note en bas de page 16)
  17. 17 Cell sites are typically divided into three sectors, with each sector operating as a separate radio-coverage area. (retour à la référence de note en bas de page 17)
  18. 18 Source: PCCA Meeting, October 2008. (retour à la référence de note en bas de page 18)
  19. 19 Source: Cisco, "Cisco Visual Networking Index: Usage Study," October 21, 2009. (retour à la référence de note en bas de page 19)
  20. 20 Comments of AT&T to the FCC, Exhibit 2, Jeffrey H. Reed & Nishith D. Tripathi, "The Application of Network Neutrality Regulations to Wireless Systems: A Mission Infeasible," January 14, 2010, http://fjallfoss.fcc.gov/ecfs/document/view?id=7020377220. (retour à la référence de note en bas de page 20)
  21. 21 Comments of T-Mobile USA to the FCC, January 14, 2010. http://fjallfoss.fcc.gov/ecfs/document/view?id=7020377832. (retour à la référence de note en bas de page 21)
  22. 22 Source: New Zealand Ministry of Economic Development, "Future Demand for 900 MHz Spectrum." Based on Vodafone's projected 3G site count, Vodafone network traffic distribution, and long term evolution of 3G. http://www.rsm.govt.nz/cms/policy-and-planning/current-projects/radiocommunications/rights-at-expiry/cellular-rights/past-consultation-and-documents/submissions/cross-submissions/cross-submission-no-04/3-future-demand-for-900-mhz-spectrum/pdf/3-future-demand-for-900-mhz-spectrum.pdf (retour à la référence de note en bas de page 22)
  23. 23 Source: Mark Pecen, Vice President of Advanced Technology, RIM, 2010. (retour à la référence de note en bas de page 23)
  24. 24 Source: Rysavy Research, "Mobile Broadband Spectrum Demand," December 2008. (retour à la référence de note en bas de page 24)
  25. 25 Source: MSNBC, "FCC warns of mobile's looming spectrum crisis," October 7, 2009, http://www.msnbc.msn.com/id/33216878/ns/technology_and_science-wireless/. (retour à la référence de note en bas de page 25)
  26. 26 Source: CTIA, "Comments of CTIA – The Wireless Association, NBP Public Notice #6, October 23, 2009, http://fjallfoss.fcc.gov/ecfs2/document/view?id=7020143313. (retour à la référence de note en bas de page 26)
  27. 27 Source: AT&T, "Comments of AT&T Inc. on NBP Public Notice #6, Spectrum for Broadband," October 23, 2009, http://fjallfoss.fcc.gov/ecfs2/document/view?id=7020143280. (retour à la référence de note en bas de page 27)
  28. 28 Source: RCR Wireless, "Analyst Angle: Solutions for the Broadband World, November 4, 2009," http://www.rcrwireless.com/apps/pbcs.dll/article?AID=/20091104/OPINION/910309995/analyst-angle-solutions-for-the-broadband-world. (retour à la référence de note en bas de page 28)
  29. 29 Source: Rysavy Research, "HSPA to LTE-Advanced, 3GPP Broadband Evolution to IMT-Advanced (4G)," September 2009. (retour à la référence de note en bas de page 29)
  30. 30 Comments of T-Mobile USA to the FCC, January 14, 2010. http://fjallfoss.fcc.gov/ecfs/document/view?id=7020377832. (retour à la référence de note en bas de page 30)
  31. 31 Source: Washington Post, http://voices.washingtonpost.com/posttech/2010/01/at_the_worlds_largest_high-tec.html. (retour à la référence de note en bas de page 31)
  32. 32 Source: GigaOm, "Vint Cerf Plugs His Plucky Space Web Protocol Into Android ," November 6, 2009, http://gigaom.com/2009/11/06/vint-cerf-plugs-his-plucky-space-web-protocol-into-android/ (retour à la référence de note en bas de page 32)
  33. 33 Source: Rysavy Research, "Wireless E-Mail Efficiency Assessment – RIM BlackBerry and Microsoft Direct Push (Including iPhone)," January 27, 2009. (retour à la référence de note en bas de page 33)
  34. 34 Source: RIM, "BlackBerry Enterprise Server for Microsoft Exchange, Version: 5.0, Administration Guide." (retour à la référence de note en bas de page 34)
  35. 35 Source: Rethink Wireless, "Data caps could give RIM a new day in the smartphone sun," November 5, 2009, http://www.rethink-wireless.com/article.asp?article_id=2108. (retour à la référence de note en bas de page 35)
  36. 36 Source: UMTS Forum, "A White Paper from the UMTS Forum Mobile Broadband Evolution: the roadmap from HSPA to LTE," February 2009. (retour à la référence de note en bas de page 36)

Si vous ne pouvez pas accéder au document qui suit, veuillez communiquer avec la personne ci-dessous afin de l’obtenir dans le format approprié.

Guylaine Verner
Industrie Canada | Industry Canada
300, rue Slater, Ottawa ON K1A 0C8 | 300 Slater Street, Ottawa ON K1A 0C8
Guylaine.Verner@ic.gc.ca
Téléphone | Telephone 613-990-6456
Télécopieur | Facsimile 613-952-2718
Téléimprimeur | Teletypewriter 1-866-694-8389


Visualisez et imprimez ce document en format PDF pour une meilleure apparence :
Research In Motion — Digital Economy Strategy Consultation Submission: Mobile Broadband Capacity Constraints And the Need for Optimization (PDF, 666 Ko)

URL suggéré :
RIM website

Avis

La consultation publique a pris fin le 13 juillet 2010. Il n'est plus possible de faire des commentaires ou de présenter des mémoires par l'entremise de ce site Web.

Du 10 mai au 13 juillet 2010, plus de 2 000 personnes et organismes canadiens se sont inscrits sur le site Web de la consultation publique pour faire connaître leurs idées et présenter des mémoires. Vous pouvez en prendre connaissance — et lire les commentaires des autres visiteurs — sur la page Soumission de mémoires et dans le Forum d'idées.

Partagez cette page

Pour faire connaître cette page, cliquez sur le réseau social de votre choix :

Aucun appui n'est accordé, soit de façon expresse ou tacite, à aucun produit ou service.