Managing Growth and Profits in the Yottabyte Era Second Edition

2010

MANAGING GROWTH AND PROFITS IN THE YOTTABYTE ERA

SECOND EDITION CHETAN SHARMA

MANAGING GROWTH AND PROFITS IN THE YOTTABYTE ERA 2ND EDITION June, 2010

2 Table of Contents | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Table of Contents

Table of Contents ......................................................................................................................... 2

Executive Summary ...................................................................................................................... 3

The Arrival of the Yottabyte Era .................................................................................................. 4

Mobile Media Evolution ............................................................................................................... 4

Mobile Technology Evolution ...................................................................................................... 6

Correlation of data traffic with data revenues ........................................................................ 7

The Mobile Data Tsunami ............................................................................................................ 9

The March Towards the Yottabyte Era .................................................................................... 12

Methodology ........................................................................................................................... 13

The Distribution of the Mobile Data Traffic .............................................................................. 14

Comparing Mobile Data Growth to Wireline ......................................................................... 15

Managing Growth and Profits in the Yottabyte Era .............................................................. 18

The need for a holistic approach to managing data ....................................................... 18

Role of Regulators ................................................................................................................... 21

The Case for Policy Management........................................................................................ 21

Faster HSPA+/LTE deployment .............................................................................................. 22

Data Offload - Femtocell/Wi-Fi/Core .................................................................................. 25

Congestion Management..................................................................................................... 25

Caching ................................................................................................................................ 26

Incentives to fill the troughs ............................................................................................... 26

Network Optimization ............................................................................................................. 26

Adopt Broadcast Mobile Video ............................................................................................ 27

Ecosystem needs to step up as well ........................................................................................ 28

New Business Models and Services .......................................................................................... 28

Conclusions ................................................................................................................................. 30

Acknowledgements ................................................................................................................... 31

About Chetan Sharma Consulting .......................................................................................... 31

About the Author ........................................................................................................................ 31


3 Executive Summary | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Executive Summary The year 2010 will be remembered for many milestones. One of them clearly will be the significant migration from voice to data services and revenues. In Q1 2010, the number three operator in Japan - Softbank Mobile reported 55% of its service revenues coming from data thus becoming the first major operator1 to have more revenues from data services than from voice. Over the course of the rest of the year, other operators like NTT DoCoMo will take this data leap as well. US, the nation with the most mobile data service revenues went past $14 Billion in quarterly mobile data revenues and is expected to go past the $50 Billion mark for the year in 2010.2 The subscription penetration in the US is well over 94%3 and the mobile data usage is on the rise. While the rate of new subscription addition has slowed down, the pace of innovation is going very strong. Just like Japan, other major economies will slowly transition from a voice-centric universe to the one where voice is just another application on the all-IP network. Operators will make significant transition from voice to data, from making calls to getting lost in applications and from voice communications to multimedia communications. Helped by the ever expanding wireless broadband networks, and release of hit devices every week, and the consumer’s insatiable appetite for information and content has brought us to the surge of a data tsunami that will shake the industry to its core. With everything moving to digital, information repositories across the web are almost doubling every day moving rapidly to the yottabyte (YB) era.4 The information, the desire and the capability to consume oodles of data is increasing exponentially. As a result the traffic – both wireline and wireless is also increasing at a predictably fast rate. In 2009, the global yearly mobile data traffic reached a new milestone – 1 Exabyte (EB) or 1 Million Terabytes (TB). In the US, the data traffic is growing so fast that we are likely to exceed the 1 EB barrier in 2010. By 2016-17, the global yearly mobile data traffic is likely to exceed 1 Zettabyte (ZB) or 1000 Exabytes. How does the industry go about managing such growth in a profitable manner when the cost of supporting such traffic will increase exponentially? Will the move to LTE offer some respite? 5 This paper is the second edition of the “Managing Profits and Growth in the Yottabyte Era”6 research paper. It discusses the research and analysis done by Chetan Sharma Consulting on the growth of mobile data traffic in over 45 countries (with a detailed look at the US market) and how the ecosystem can apply some strategies to manage growth and profits. We have built detailed models to estimate the rise of mobile data network traffic and to understand as to how the margin per bit can be maintained. Over the course of the last year, we have worked with several global players in the ecosystem to deploy effective strategies and solutions. This paper also draws from this experience on the ground.

1 The two operators in Philippines - SMART and Globe have been garnering more revenues from data than voice

for a number of years but the relative revenue size is much smaller and most of the data revenues are from SMS. Incidentally, for both the operators, the data % dropped below 50% in Q1 2010. 2 Source: US Wireless Data Market update Q1 2010, Chetan Sharma Consulting. It was also the first time any nation

exceeded the $10B mark in a quarter for mobile data revenues. 3 In fact, if we consider the demographics ages 5 and higher, we are at almost 100% penetration as of Q1 2010

4 For reference, 1 TB = 10

12 bytes, 1 PB = 10

15 bytes, 1 EB = 10

18 bytes, 1 ZB = 10

21 bytes, 1 YB = 10

24 bytes

5 For the purposes of this paper, we consider LTE/WiMAX as 4G technologies though they hasn’t been officially

designated as such. 6 The first edition of the paper is available at http://www.chetansharma.com/yottabyteera.com

http://www.chetansharma.com/yottabyteera.com


4 The Arrival of the Yottabyte Era | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

The Arrival of the Yottabyte Era Mathematically, Yottabyte refers to a septillion or 1,000 billion terabytes. In 2006, IBM predicted that by 2010 the size of world’s codified information base will be doubling every 11 hours.7 That’s beyond astronomical - paving the way to the yottabyte era.8 When we were writing our Mobile Advertising book9 in 2007, we started thinking about yottabytes and what it means to have all this data, how will this be accessed and understood. What does that mean for information consumption and the resulting traffic? How do we extract intelligence out of this data without jeopardizing the network or the storage? How do we do that on a per user basis? How do we manage peaks and surges? How do we put the intelligence from the data back into the system to empower the feedback loop in real-time? There are significant opportunities in how yottabytes are going to be managed and understood. Rather than being overwhelmed by the data tsunami, we should learn to tame the beast. As we wrote in the book, for analytics and intelligence, data should be revered. But for the network, it needs to be managed. It is inevitable that the business models and financial metrics will also undergo a significant transition as a result of the growth in the mobile data segment. The pricing models, the customer acquisition strategies, the metrics that Wall Street pays attention to will change in due course. Instead of focusing exclusively on Average Revenue Per User (ARPU), we will start looking into Average Margin Per Subscription (AMPS), Average Connections Per User (ACPU), Customer and Family Lifetime Value (FLTV) that maximizes profits across all connections that a family or the user has. Instead of promoting family minutes, carriers will promote family terabyte plans where one can bundle usage across multiple devices used by a family. Operators around the world will have to focus on how to both profit from the deluge of data as well as manage the surge in a cost-effective way that optimizes spectrum, financial, and the network resources. Companies who plan early and navigate this complex maze will benefit while the unprepared will be left to play catch-up. This also opens up the opportunity for the ecosystem to develop tools and technologies to help carriers manage growth as well as use the data generated in an intelligent manner to enhance the user experience.

Mobile Media Evolution The era of instant on, always available broadband connectivity is altering the media landscape in profound ways. Over the last 2-3 years, the consumption of digital media has evolved significantly. As content is becoming digital and as devices are becoming more powerful and able, the mobile device is playing a central role in how digital media is consumed around the world. The digital rush has helped make mobile a $1.1 trillion (as of 2009) industry (Figure 1). As the demand for mobile content consumption is increasing exponentially, service providers are rushing to enhance their infrastructure to keep up.

7 The toxic terabyte, IBM Global Technology Services, 2006, http://www-

03.ibm.com/systems/resources/systems_storage_solutions_pdf_toxic_tb.pdf 8 By some measures we are probably already in the yottabyte space as the growth since the prediction was

originally made has been staggering 9 Mobile Advertising: Supercharge your brand in the exploding wireless market, Chetan Sharma, Joe Herzog, and

Victor Melfi, John Wiley, 2008. We wrote a section “Prepare for data everywhere – from gigabytes to yottabytes.” Also see a paper written by Balasubramanium Venkitachalam of WPP titled “Your Guide to Profits: Insights through Business Analytics available at http://www.wpp.com/WPP/marketing/reportsstudies/yourguidetoprofits.htm

http://www-03.ibm.com/systems/resources/systems_storage_solutions_pdf_toxic_tb.pdf

http://www-03.ibm.com/systems/resources/systems_storage_solutions_pdf_toxic_tb.pdf

http://www.wpp.com/WPP/marketing/reportsstudies/yourguidetoprofits.htm


5 Mobile Media Evolution | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Figure 1. Global Mobile Industry Revenue Distribution10 The percentage of revenues coming from data services are increasingly becoming a big part of the overall service revenues of the operators. Globally, over 26% of the revenue now comes from data services (Figure 1).11 The main drivers for increased activity on the mobile devices are four-fold:

better networks in the form of 3G/3.5G (and future upgrades to 4G+),

higher processing power devices being available at mass-market prices around the world,

consumers are not only consuming, but also producing content at an exponential pace,

the opening up of the mobile ecosystem has attracted thousands of developers who are building compelling applications and services for various mobile platforms.

Figure 2. Distribution of time spent on iPhone12

10

Source: Global Wireless Data Market Update 2009, Chetan Sharma Consulting, 2010 11

Source: Chetan Sharma Consulting, 2010 (as of Q4 2009) http://chetansharma.com/globalmarketupdate2009.htm 12

Source: Chetan Sharma Consulting, 2010

http://chetansharma.com/globalmarketupdate2009.htm


6 Mobile Technology Evolution | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

As such from the early days of ringtones and graphics, the mobile ecosystem has evolved to deliver more rich content experiences such as high-fidelity and multi-user mobile games, very high quality video in the form of multicast (though unicast is the one that is widespread), and social networking applications like Facebook and Twitter. The consumption patterns have also decidedly changed with consumers spending more time on the non-voice activities (Figure 2). Additionally, the smartphone/superphone13 boom that followed the iPhone introduction in 2007 changed the dynamics of the market and how consumers view their mobile devices. It is interesting to note that on devices such as the iPhone, consumers spend less than 32% of their time on voice; rest is on other applications and services. Such a shift is also changing the service provider business models, how they run their business and plan for future growth. Mobile media and data services are the only drivers for growth as voice revenues decline (worldwide). Significant mobile data usage is also putting strain on the operator’s network and as such they are forced to come up with data expansion (3.5/4G) and alternate (Wi-Fi/Femtocell) strategies so that they can profitably stay ahead of the curve. Also, the very definition of the mobile devices is changing. More and more consumer electronic devices are being launched with a wireless data connection (Wi-Fi or Cellular). Devices such as iPad, Kindle, Nook and others are changing the industry dynamics as well. Newer players are entering the marketplace and the competitive landscape is being impacted in many ways. Apple’s appstore changed the way applications found their way onto consumer’s handset. This made it easier and lucrative for a developer in a garage to launch new applications. The impact of open development and application platforms is rearranging the ecosystem in a profound way that is going to redistribute the revenues in the value chain. While the opportunities to exploit mobile media remain strong, the ecosystem needs to worry about meeting the expectations of the consumers. They have to invest in infrastructure, developer ecosystem, and continuous flow of new and improved handsets to keep up with the growing interest. It is clear that as digital media consumption grows; mobile will be the front and center of this evolution.

Mobile Technology Evolution By the end of 2010, the total mobile subscriptions are expected to cross the 5 billion mark.14 The Cell phone has clearly been the most adopted technology in the world. As it has moved from being a luxury to a utility to a necessity, both the cellular network technology and the devices have evolved to meet the growing demand and need for communications. From being used primarily for voice in the eighties and nineties, consumers are now spending more time on their mobile devices on non-voice activities. Mobile devices have become the gadget of choice for both increasing the productivity to save time and for entertainment to kill time. If we look at the technology evolution (Figure 3) over the last 30 years, the first generation or 1G was analog. With the arrival of the GSM Global Standard we moved into the digital or the 2G era but phones were still primarily voice communication devices with SMS based messaging starting to take off in various parts of the world by mid-nineties. Towards the tail-end of the last decade, NTT DoCoMo launched i-mode and defined mobile data. Soon, to meet capacity constraints and to take advantage of the promise of data services, 3G was launched with UMTS and EV-DO. Over time, most major carriers around the

13

Superphones are devices such as the iPhone, Droid, Pre - devices with full browsers, superior media capability, fast processor, etc. In the paper, the term is used primarily to segment the devices by traffic since they show distinct difference from traditional smartphones. In a few years time, the distinction between all these categories will diminish as it will be hard to find a “dumb” phone. 14

Source: Global Wireless Data Market update 2009, Chetan Sharma Consulting, 2010 http://chetansharma.com/globalmarketupdate2009.htm

http://chetansharma.com/globalmarketupdate2009.htm


7 Correlation of data traffic with data revenues | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

world15 have some form of 3G up and running by 2010. WiMAX deployments started a couple of years ago and now HSPA+ and LTE upgrades are in full swing in many regions.

Figure 3. Global Mobile Technology Evolution16 It should also be noted that each generation of the technology has a faster growth curve (compared to the previous generation). Put another way, the time it takes to amass the first 100M subscriptions has been shrinking with each evolution cycle. Another point that should be noted is that there is a significant overlap between technologies. Launch of 4G doesn’t mean that 3G is going away anytime soon. Each technology evolutionary step helps lower the cost of delivering the traffic. By using 4G, operators can lower their per MB costs and if they can aggressively provide coverage and move the heavy usage subscriptions to LTE, it will impact the bottom line.17

Correlation of data traffic with data revenues Up until 2007-08, mobile data was a relatively small contributor to both the costs and the revenues for majority of the operators. In fact the margins were fairly healthy and the overall traffic from data was a small fraction of the overall network traffic. However, with the advent of the iPhone and other superphones, the 3G/4G networks, and a flood of content; the mobile data traffic and the cost and revenue from data started to ride on the hockey stick curve that we are witnessing today. In the US, in 2009, for the first time, mobile data traffic exceeded voice traffic (Figure 4). Even globally, during the last couple of months of 2009, global data traffic went past the voice traffic.

15

China started their migration to 3G in Jan 2009 using TD-SCDMA, WCDMA and EV-DO paths while India will begin deployment in 2010 16

Source: Chetan Sharma Consulting, 2009 17

The move to LTE will mean faster connection which means more usage and hence more traffic but if this traffic is well managed as discussed later in the paper, the net of LTE deployment will be incremental cost savings. However, LTE might not be the most prudent strategy for every operator. The motivation and the need for 4G migration depends on a number of factors discussed on Page 22. LTE is also going to require substantial investment and will include a backhaul upgrade which is generally the biggest congestion point in the network.


8 Correlation of data traffic with data revenues | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Figure 4. US Mobile Traffic: Voice vs. Data (2004-10)18

20% 40% 60% 80% 100%

10%

20%

30%

40%

50%

Japan

Sweden

New Zealand

Philippines

Hong Kong

Australia

Germany

Switzerland

Netherlands

DenmarkPortugal

Finland

Italy

Spain

UK

Singapore

Korea

France

Russia

Canada

USIndonesia

Brazil

ArgentinaIndia

China

Mexico Malaysia

South Africa

Data as % of Overall Network Traffic

Dat

a as

% o

f Se

rvic

es

Re

ven

ue

s

Revenue-T

raffi

c Equili

brium

© C

het

an S

har

ma

Co

nsu

ltin

g, 2

01

0

Figure 5. Relationship of data traffic with data revenues in various countries19

18




9 The Mobile Data Tsunami | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Figure 6. Vodafone Europe: Share of Traffic and Revenue by Data Services

This is obviously having an impact on the margins as the cost to support such a surge is not in equilibrium with the revenues from such traffic and usage. We looked at the data consumption trends in over 45 countries (Figure 5) and, except for China, most nations (especially Western Europe and North America), the ratio of % data traffic to % revenue is out of equilibrium that has direct impact on the margins. For example, in the US, in 2007, data accounted for 17% of the revenues and 29% of the traffic. By 2011, the data will account for 34% of the revenues but 90% of the traffic.20 Similarly, for Vodafone Europe, the data traffic % grew from 48% in 2008 to 77% in Q1 2010 while the corresponding data revenue % grew from 24% to 28% (Figure 6).21 It is also clear that most of the operators will start falling in the 75%+ data traffic category and if they are not careful, it can quickly become a big problem for user experience, competitiveness, and ultimately the financial viability of data services. As a side note, the small Japanese operator Willcom who kicked off the phenomenon of the“flat rate” for mobile data filed for bankruptcy in Q1 2010.

The Mobile Data Tsunami It has become a cliché to say that the iPhone has changed everything. Apple’s iconic device has had a profound impact on the ecosystem on several fronts. First, Apple again taught the industry the power of simple user experience. Second, the appstore model has disrupted the traditional storefront model. Third, once exposed and addicted to the mobile way of life, consumers are not looking back. Fourth, some form of a flat data pricing is the norm for the new smartphones especially in the US market.22 Finally, all this

20


Source: Vodafone, 2010 22

Not all operators in all countries are a fan of the “all-you-can-eat” data pricing. While they are offering similar devices and services, the pricing of data is based bandwidth consumption which is prudent in the long run. In the US too, there are signs that the changes in the flat fee pricing are imminent



means significant jump in network traffic. As the percentage of smartphone subscribers grows, so does the traffic – by leaps and bound. Figure 7a shows data consumption in the leading markets.

Figure 7. a) Data consumption in some of the leading economies b) Data

consumption by device type23 The usage data from the iPhone, Android devices, and the iPad (Figure 7b) indicates that the data consumption increased 50-100 times compared to the previous generation of devices. As such the per user megabyte (MB) consumption is multiplying at an accelerated pace in almost every part of the world. For example, in Hong Kong, the total data consumption increased 379% (from 2008) to 638 Terabytes.24 It is expected that the demand for data will continue its exponential trajectory. Operator Zain reported 500% increase in data consumption in second half of 2009.25 Verizon, US’s largest operator reported over 3500 TB/month traffic in 2008 which could explode to 4,000,000 TB/month by 2014.26 T-Mobile USA has indicated that their average mobile data consumption for Android devices is around 400 MB/mo.27 In Norway, the mobile data traffic jumped 237% in 2009 with mobile broadband users consuming more than 1.2 GB/mo.28 In UK, the data traffic quadrupled in first half of 2009 compared to same time period in 2008 while the revenues declined.29

23

Source: a) Sweden, Hong Kong, and Denmark figures calculated from data published by the respective regulatory authorities, US estimates are from Chetan Sharma Consulting, 2010. b) Chetan Sharma Consulting, 2010 24

Source: Key Statistics for Telecommunications in Hong Kong, Office of the Telecommunications Authority, Hong Kong, April 2010 25

Source: http://bit.ly/952xEJ 26

Source: Dick Lynch, Verizon Wireless, 2009 27

Source: Robert Dotson, T-Mobile USA, 2010 28

The Norwegian Electronic Communications Services Market, Oct 2009 29

Ofcom, Mobile Revenue and Data Volumes, UK, 2010

http://bit.ly/952xEJ



Figure 8. Correlation of Speed/Latency with Consumption for Smartphones30 One of the reasons the consumption goes up with a newer generation of technology is that typically the bandwidth capability and spectrum efficiency increases. Additionally, the latency also drops. So, most of the times it is a 4x improvement in overall performance which leads to better user experience and if the pricing is right will have a 2x impact on consumption with the introduction of each new network technology (Figure 8). There are two key factors behind congestion:

The peak data usage o Main culprits: data cards and embedded laptops followed by smart/superphones

The signaling traffic o Main culprits: smart- and superphones

In fact, signaling traffic is growing faster than the raw data traffic due to smartphones/superphones as they are not very efficient with applications. The smartphone signaling is over 8x the data card signaling traffic.31 And since signaling consumes over 50% of the available network resources and smartphone while representing a small segment of the overall base disproportionately impact the network. For the raw data usage, it is the peak capacity that causes congestion and not the “total capacity” though it is a factor in CAPEX planning. Typically, in the urban areas, 15% of the cell sites manage 50% of the traffic and as such the load is not evenly distributed in the network.32 There are also two key congestion points in the network:

30


Source, Airvana, 2009 32

Source: NokiaSiemensNetwork, 2010


12 The March Towards the Yottabyte Era | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Air interface - Some of the relief can be obtained by getting more spectrum, deploying more cell-towers or upgrading to more spectrally efficient technology or offloading or using a set of solutions described later in the paper.

Backhaul traffic - Even if the air-interface is congestion free, the dilapidated T1/E1 infrastructure is just not cut out to handle modern day data traffic. As such, the backhaul needs to be upgraded to microwave, fiber, or light networks (or a combination of the three) to have enough capacity and data economics to make the data growth worthwhile for the operators.

The portfolio of solutions should address both peak data and signaling traffic to manage the networks efficiently. One should also appreciate the move towards providing wireless connectivity to all electronic devices from cameras to security alarms to energy meters. In fact, Wireless World Research Forum (WWRF) forecasts 7 trillion wireless devices serving 7 billion people by 201733 which translates into 1000 radios per person. Most will be in the form of sensors and tags that form the basis of ambient and context intelligence around us to truly make mobile device a remote control of our lives.34 While this Always On Real-Time Access (AORTA) environment provides for instant access to information and intelligence, it also creates a mountain of data traffic that needs to be understood and managed.

The March Towards the Yottabyte Era Consider average usage on a 3G iPhone: average session on NY Times - 0.5-1 MB, 10 min YouTube streaming - 10-15 MB, 15 min radio streaming: 12-16 MB, average application downloads: 2-10 MB, Average Facebook Sessions: 100-200 KB, etc. Eventually, they all add up to 0.5-1 GB of average data traffic/user/month (see figure 9 for broadband speeds by application type). Similar trends can be found on devices like the Droid, HD2, EVO, and the iPad. One can well imagine that with increasing population of such devices and better bandwidths and devices, how the multiplier effect starts to take hold and move from Tera- to Peta- to Exa- to Zetta-to Yottabytes.

Figure 9. Broadband speed requirements by application type

33

The WWRF Vision, Nigel Jefferies, WWRF, 2007 34

Mobile Advertising: Supercharge your brand in the exploding wireless market, Chetan Sharma, Joe Herzog, Victor Melfi, 2008, John Wiley & Sons


13 The March Towards the Yottabyte Era | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Methodology To get a better understanding of the mobile data traffic growth in the US market, we built a bottoms-up detailed model to estimate the current and future data traffic on the mobile networks in several countries and studied traffic patterns in more detail in the US market. One of the reasons we focused on the US market is that the US market has been ground zero for mobile broadband and data consumption trends with the highest penetration of the smartphones and data cards. With Verizon deploying LTE, AT&T and T-Mobile expanding the HSPA+ coverage, and Sprint and Clearwire launching the WiMAX handsets, the equipment and the edge and core networks are being tested to the limits in some instances which is leading to the rethinking of how mobile broadband is managed and paid for. While Japan and Korea have significantly higher 3G penetration, it is the smartphones, superphones, datacards, connected devices that can create high network congestion. AT&T’s network has become a real-time research laboratory for Ericsson and others in understanding the requirements for future generation network component design and architecture. For each of the major US carriers, we looked at the traffic generated by the data customers (excluding messaging traffic) across the four distinct segments of devices – feature phones, smart phones, super phones, and data cards (including embedded devices). Then we analyzed the traffic generated by messaging and added to the non-messaging data traffic to calculate the overall mobile data traffic growth in the US market. We realized that only by looking at a very granular level can one fully grasp the various factors that impact network traffic growth. Thanks largely to the iconic devices like the iPhone and the mobile data cards, the mobile data growth grew disproportionate to the revenues in 2008-9. In 2009, the mobile data revenues in the US market grew by 33% but the mobile data traffic grew by over 200%. Similar trends are being observed in other markets. Additionally, the types of applications and services that are becoming available on the smartphones have higher requirements for bandwidth compared to their earlier predecessors, for e.g. streaming audio might only consume .2 Mbps/stream but an HD streaming video can consume upto 10 Mbps of capacity. As such, the mobile data consumption per user/month is on the rise across the board. We expect the average mobile data consumption to increase from less than 50 MB/user in 2009 to almost 3 GB by 2014.

Figure 10. Mobile Data Traffic Growth in the US market35

35



14 The Distribution of the Mobile Data Traffic | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Looking at the picture on a more granular level, the situation is even more challenging. In HongKong, the data consumption grew 3.8 times in 2009.36 In TeliaSonera’s Nordic and Baltic region networks, traffic increased 500% in 2008. AT&T Wireless noted that they have experienced 6500% data growth in the last three years.37 The overall mobile data traffic in the US is expected to increase from over 500 Petabytes (PB) in 2009 to 40 Exabytes (EB) in 2014 or by over 70 times (Figure 10). The main driver for this growth will be the data cards and the increasing population of the smartphones in the market.

This traffic can be managed and reduced by techniques discussed later in the paper like the introduction of the Femtocells and the Wi-Fi hotspots to offload traffic in fixed locations or by active congestion management and optimization. While LTE (or for that matter WiMAX) will help in reducing the cost of the traffic, it will also increase MB consumed compared to its predecessor technologies like EV-DO and HSDPA/HSPA, sometimes by 50-100%, thus neutralizing some the benefits.

The Distribution of the Mobile Data Traffic As noted above, the data cards consume the most bandwidth on any major network in the western world. At the end of 2009, data cards accounted for over 61% of the data traffic in the US (Figure 11) while super phones were inching up with 27%. Though feature phones represented 75% of the device base, their contribution to the data traffic was only 2% with rest being consumed by other smartphones (Figure 12).38 Messaging - the biggest revenue generating category accounting for over 45% of the revenue had less than 1% of the traffic. What happens when the % share of the data cards increases? What if 25% of the subscriber base has a data card? What happens when phone-as-a-modem becomes more prominent or phone acts as a network conduit for projection screens? Network planning will need to take into account these scenarios.

Figure 11. Mobile Data Traffic Distribution by Device Type39

36

Source: Key Statistics for Telecom, Office of the Telecommunications Authority, Hong Kong, April 2010 37

Source: AT&T, 2010 38

This profile will obviously look different for developing countries which generally have lower smartphone and data card penetration as of mid 2009 39



15 Comparing Mobile Data Growth to Wireline | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Further, only a small percentage of the users are consuming a large share of data.

Figure 12. Mobile Data Traffic Distribution by Device Type40

Mobile Data Growth Will (Also) Come from New

Sources Most of the current discussion has been about traditional mobile devices, however, as we go into 2011 and beyond, devices such as the iPad, pico projectors (embedded or external), modules in automobiles, security devices, cameras and so on and so forth will start to create a category of their own. We estimate that in 5 years time, such devices will account for 5-10%41 of the overall mobile data traffic in the US.42 The level of consumption will also be raised by embedded pico-projectors and devices that work in concert with other screens like in an automobile. These devices will be all IP based and carrying just a data subscription. In fact, we are likely to see “family data plans” or “multi device data plans (multiple devices on a single data plan)” as early as later this year. Some of these new experiences will result in 5-10 times data consumption compared to today’s superphones.43

Comparing Mobile Data Growth to Wireline To better gauge how the mobile data growth is going to progress, it might be beneficial to take a look at how the data consumption grew in the Wireline world. If we study the time-period 1996-2014 in the US,44 one will notice that the mobile evolution has lagged Wireline growth by approximately 8-9 years though with each passing year, the gap is closing.

40


Traffic resulting from embedded picoprojectors is included in the device type traffic forecasts 42


The role of connected devices became apparent when in Q1 2010, AT&T and Verizon Wireless reported they added more connected devices than postpaid subscriptions for the quarter. 44

This profile will look different in advanced broadband nations such as Japan and Korea which are significantly ahead of the US in broadband penetration both in Wireline and wireless. Comparing broadband profiles of various nations is outside the scope of this paper.


16 Comparing Mobile Data Growth to Wireline | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

As indicated in figure 13, when the penetration of broadband grew in Wireline, so did the number of Petabytes consumed on a national basis for the US market. And the penetration grew with each enhancement of the technology from ISDN to FTTH. During the early years of mobile broadband in the US (2004-2009), the data consumption has followed similar patterns but with the advent of devices like the iPhone and with the increased penetration of the data-cards, the traffic is rising faster than it did in the Wireline world and we expect that by 2014, mobile data consumption will be less than 4 years behind the Wireline data consumption. While P2P (Peer-to-Peer) and video streaming have been the main reason Wireline consumption shot-up, in the mobile world it has been the browser usage, video streaming, and apps on smartphones, netbooks, smartbooks, laptops, and similar devices that is contributing to the bulk of data traffic followed by streaming and application downloads. Also, since the peak usage significantly increases the network expenditure, the key statistics that need to be understood are around the applications that drive traffic during peak utilization, not necessarily the applications that consume the most bandwidth across the entire traffic landscape. For example, P2P applications might consume a significant amount of traffic but their relative percentage during peak utilization might be smaller on a given network. As such, devoting too much time to P2P management might not be that effective (though it needs to be tackled nevertheless). So, as we try to understand mobile data consumption in the next decade, it will be worthwhile to keep an eye on how things are evolving in the Wireline space and what solutions are working in managing the traffic growth and can be adapted to the mobile environment.

Figure 13. Broadband penetration and traffic for Wireline and Mobile data networks in the US (1996-2014)45

45

Source: Chetan Sharma Consulting. Data Sources: Wireline Traffic Data from Minnesota Internet Traffic Studies (MINTS) -http://www.dtc.umn.edu/mints/igrowth.html. We took the mean of the yearly ranges (forecasted for year 2010). Wireline Broadband Household penetration data from Pew Internet - http://www.pewinternet.org (forecasted beyond 2009). Mobile Data traffic – Chetan Sharma Consulting analysis. Mobile Broadband Data Subscriber Penetration – Chetan Sharma Consulting analysis

http://www.pewinternet.org/

Managing Growth and Profits in the Yottabyte Era

The need for a holistic approach to managing data The growth in the network traffic can be quite injurious to the financial bottom line of the operator and to the industry’s ability to maintain pace with the demand. Unless a long-term plan is put in place that addresses and manages the traffic at a very granular level, the cost incurred due to an explosive demand will become unsustainable by 2013. At that point the revenue being generated could fall below the cost of sustaining such traffic. However, if the operators attack the problem using several different strategies (an illustration in Figure 14), the growth can be managed and brought in line with the technology evolution such that the industry can take advantage of the falling per MB costs.

2G

4G

3G

Core Network

Radio Access Network Core Network

Mobile Backhaul

Fiber

Microwave

Copper

LightNetworks

GGSNRNC SGSN

Offloading Traffic

WiFi/FemtoCell

Offloading Traffic

Inte

rnet

Ser

vice

s

Compression/Optimization of Data Traffic

Policy Management

Compression/Optimization

Optimization

© C

het

an S

har

ma

Co

nsu

ltin

g, 2

01

0

Figure 14. A representative framework for a holistic approach to managing data

growth46 Some of the strategies for managing network traffic growth are:

1. More Spectrum 2. Faster HSPA+/LTE deployment 3. Wi-Fi/Femto-Cell deployment 4. Upgrading Backhaul 5. Core network offload 6. Congestion Management 7. Network Optimization 8. Policy Management 9. Adopt Broadcast Mobile Video

46



19 Managing Growth and Profits in the Yottabyte Era | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Figure 15. Managing Network Traffic Costs (US)47 The eventual carrier strategy will depend on their particular situation which will differ on several fronts:

1. Spectrum position 2. 3G and 3.5G technology deployment lifecycle 3. Existing Broadband infrastructure investments 4. Network traffic and distribution of devices and traffic 5. Demographics and subscriber growth 6. Short-term and Long-term customer acquisition strategies in consumer and enterprise segments 7. Network Coverage and Pricing Plans 8. Competitive position 9. Financial resources 10. Multi-play position and strategy 11. Others

Depending on the carrier’s broadband deployment and their position in the market, different set of solutions might be considered. For example, carriers who have just started to deploy 3G might want to hold-off on 4G upgrades while others who are 5-7 years into the lifecycle might be more incented to move to 4G. Some operators might not have the additional spectrum needed to operate parallel networks. It is important to understand the importance of spectrum in the continued growth of mobile data services. In countries where sufficient spectrum is not allocated for 4G and the related services or the spectrum is not harmonized with the rest of the world or the spectrum caps are imposed, they will stand at a big disadvantage.48 In 2010, the mobile broadband penetration will surpass the fixed penetration globally.

47


It will be difficult to get economies of scale that can help lower the price of equipment and services at a faster pace. Developing countries are the ones who will get most impacted by non-harmonized allocation of frequencies.



Countries that are behind the curve in spectrum allocation will lag behind as lack of spectrum will delay the launch of broadband services.49 However, one shouldn’t just expect newer networks to take the load of the growing demand. One must consider a combination of strategies to lower the overall cost of managing the mobile data network traffic. Figure 15 shows the cumulative downward impact on the network data traffic costs by different strategies. To have meaningful savings, operators will have to develop holistic strategies that take advantage of multiple technologies and strategies over time. The ability to drive the cost/bit lower would become a significant competitive advantage.

Dealing with the Spectrum Crisis Spectrum is a finite resource and the demands being placed on the national spectrum are enormous if not unrealistic. One doesn’t need to be a spectrum expert to determine that the majority of the spectrum has been taken. What’s left is just reallocation of the spectrum. While the mobile data tsunami is occurring right now, even the identification of available spectrum, its reallocation and assignment is often a long, multi-year process. And while consumer data demands make a strong case for additional spectrum so do other critical services like public safety, smart grid, telemedicine, military, and the likes. Additionally, the cost of broadband deployment can limit the deployment in rural areas where the cost to support the subscribers is relatively high. It is apparent that to achieve next generation broadband speeds like 50-100 Mbps, new contiguous spectrum is needed. We clearly need more spectrum and the process of procurement and harmonization should start urgently. However, it will be a mistake if the dominant solution for the broadband capacity crisis is procurement of more spectrum for the following reasons:

1. There isn't enough spectrum, especially the right spectrum

2. It takes 7-10 years to procure the spectrum for wireless use despite the sense of urgency

3. By focusing on spectrum only, we will be just postponing the current crisis

4. By giving out spectrum too soon, industry won't have the opportunity to learn to thrive within its

means and let new technology and business innovation show the way to handle the increased data

consumption

As such the industry and national interests will be best served if we take a look at alternate solutions in parallel with the spectrum needs like shared used of spectrum and infrastructure or designing solutions and architectures that don’t require additional spectrum or invent new technologies that manage the rapid growth of data consumption without impacting consumer choice and user experience. As we go about refining the plan and executing on the recommendations, the industry will need to keep a few things in mind. First of all, the additional spectrum is not going to be the panacea for the mobile broadband consumption, especially in the short-term. By the time the first batch of spectrum becomes available, the consumption would have more than quadrupled, so the industry needs solutions for the immediate future. Next, the industry needs to figure out ways to build the infrastructure that is not utterly dependent on the additional spectrum. So, we must invest in R&D that provides breakthroughs in using existing spectrum more efficiently to provide more throughput as well as technologies and strategies that don’t require

Spectrum caps are also detrimental for mobile broadband growth as it limits the deployments. For example in South America, most nations have put spectrum caps (e.g. Argentina 50 MHz, Brazil 80 MHz, etc.). 49

Full treatment of the impact of broadband on economies is beyond the scope of this paper. This subject is dealt with in detail in “Wireless Broadband: Conflict and Convergence,” Vern Fotheringham and Chetan Sharma, John Wiley & IEEE Press, 2008 in chapter “Broadband and the Information Society.”



additional spectrum like using the optical wireless broadband networks.50 Also, the fiber penetration will matter a lot as it can be used to off load a good majority of the mobile data traffic on an as needed basis. Nations who have deeper fiber deployment will end up having a more robust mobile infrastructure in the long run. Wireline and wireless should be considered part of the same “one” network.

Role of Regulators The role of mobile broadband in a nations competitiveness is well established. Given that regulators are in charge of two key aspects of the broadband plan - spectrum and market competitiveness, the way regulators go about defining the priorities and policies for their country can have a profound impact on the evolution of mobile broadband in that country. Each country of course sets its own goals and priorities that suits its economy and players in the ecosystem. Regulators have to be careful to not “over-regulate” or “under-regulate” as it can set the market back by a few years. In most of the major markets, regulators have taken a proactive role in finding new spectrum for mobile broadband as well as introduce new players to keep the market competitive. However, regulators will need to be ahead of the demand curves as any regulatory action typically is very time consuming and by the time some ruling takes effect, it is already too late. Regulators and political entities will do well to draft 10-25 year policies to address the needs of the nation’s infrastructure. The debate around being “open” also needs some direction. What does it really mean to lay claim to being “open.” Should only the entities that are under the regulatory purview be scrutinized or should the industry heavyweights be under the same microscope. In a “for-profit” industry, who decides what should be opened up to preserve consumer choice and market competition? How far do the regulators go in asserting an “open Internet?”

The Case for Policy Management Consider these facts for a minute. In 2009, in AT&T’s network, approximately 3% of the smartphone user base accounted for over 40% of the smartphone traffic.51 Similarly, in other regions, approximately 3-5% of the users are consuming 60-75% of the traffic.52 If this consumption is taking place in off-peak hours when the capacity is lying un-utilized, it is one thing but if it is impacting other users during peak-hours, it is another. Operators who have managed the expectations, pricing, and policies well have had a better time managing the growth from “hyper-users.” The performance of the network and the pricing for the usage scales with demand. In countries where operators have a tiered pricing structure or have instituted stricter policy controls, the average consumption is less than 25% compared to operators who have had an unlimited pricing and policy structure on smartphones.53 There are multiple ways to institute a workable or “fair-use” policy

1. Pricing Tiers - User pays by the tier of consumption. For e.g. Telstra has 6 tiers for iPhone data consumption from $5/5MB to $119/9GB.54

2. Unlimited to a limit - Users have unlimited access but after a certain threshold (for a given month) the speed drops for rest of the month. For e.g. T-Mobile USA offered this to its user base in April 2010 (the soft unlimited limit is 5GB).55

50

For a detailed treatment of this subject, please see - Role of Optical Wireless Broadband in the Evolving Mobile Ecosystem, Chetan Sharma, 2010, http://chetansharma.com/opticalwirelessbroadband.htm 51

Source: Ralph De Vega, AT&T, Oct 2009 52


It should be noted that the data cards used to have unlimited usage plans as well but fairly quickly most operators around the world instituted some limits, typically 5GB monthly. 54

http://www.telstra.com.au/mobile/phones/iphone/pricing.html. There are overage charges or one can upgrade 55

http://www.fiercewireless.com/story/t-mobile-drops-data-cap-5gb-plan/2010-04-27

http://chetansharma.com/opticalwirelessbroadband.htm

http://www.telstra.com.au/mobile/phones/iphone/pricing.html

http://www.fiercewireless.com/story/t-mobile-drops-data-cap-5gb-plan/2010-04-27



3. Choose your speed and consumption limit - Users can chose the tiers by data speed and total consumption for a given month e.g. Leap introduced the plans where users can chose between $40 - 2.5GB/600kbps, $50 - 5GB/1.4Mbps, and $60 - 10GB/1.4Mbps.56

4. Max Ceiling - Users pay on a per bit/packet basis up to a maximum limit after which there is no charge for extra usage. For e.g. Softbank Mobile in Japan as the iPhone data pricing from $11 to $64 and there are grades of usage/pricing between the two pricing end-points but when the user hits the $64 limit, there is no additional charge for rest of the month. They also send an alert if the user exceeds a certain usage/pricing threshold.

5. By application usage - Pricing by applications for example, video streaming for certain channels is included but for additional channels, the price of consumption goes up. This is indirectly implemented in programs such as Vcast.

6. Buying additional bundles - Consumers can buy additional data bundles beyond the initial fixed limit. For e.g. Vodafone customers have a 500MB limit on smartphones and non-contract subscribers can purchase an additional 500MB bundles for £5 each or 50p/10MB.57

7. Quality of Service Tiers - This is an unchartered territory but with the advent of LTE, we are likely to see some operators consider segmenting the pricing by QoS. The challenge is going to be putting together the value proposition and a communication message that resonates well with the consumers.

Policy management is by far the quickest way to manage data traffic since the implementation doesn’t require hardware or software upgrades. It does require significant customer education in markets where unlimited usage has been the norm until now. For example, had AT&T instituted a 1 GB/month limit (which wouldn’t have impacted 90% of the smartphone users) on smartphones in 2009, its smartphone traffic would have likely reduced by at least 30% and the overall traffic by over 17%.58

Faster HSPA+/LTE deployment The primary driver for LTE is mobile data. As of mid-2010, several carriers have announced their LTE deployment plans. TeliaSonera became the first operator to launch an LTE network. In the US, the two large operators Verizon and AT&T have indicated rollouts by the end of 2010 with more aggressive deployments in 2011 and beyond. If past experience is any indicator, it always takes more time to deploy and perfect new technologies esp. by the first-movers.

56

http://www.fiercewireless.com/story/leap-experimenting-new-unlimited-broadband-plans/2010-04-28 57

http://gadgetmosster.com/?p=65 58

Source: Chetan Sharma Consulting, 2010. On June 2nd, 2010, AT&T announced new capped data pricing with $15/200MB and $25/2GB pricing for smartphones (including iPad) http://bit.ly/9meqNc

http://www.fiercewireless.com/story/leap-experimenting-new-unlimited-broadband-plans/2010-04-28

http://gadgetmosster.com/?p=65

http://bit.ly/9meqNc



Figure 16. LTE Subscriber growth forecast59 LTE is expected to lower the cost of per MB delivered by 50-60% (compared to preceding technologies if you keep other things constant like the frequency band and amount of spectrum).60 However, these cost savings can only be delivered if there is sufficient coverage so that the data traffic can be off-loaded to the LTE network. Unless there is 80-90% coverage of the major metropolitan areas by 2012 (Figure 16), we won’t see LTE making a big dent in the network costs as the traffic will continue to accelerate through 2013-14. In fact, in our conservative model, LTE might only provide 1-3% cost savings.61 If there is full deployment (80-90% POP coverage) and if a good majority of the data card users can be off-loaded to the LTE network, cost savings can be up to 30-40% by 2014.62 This requires faster time-table of LTE deployment, device rollouts, and move towards an all-IP infrastructure. Many operators like T-Mobile USA, Vodafone, Telefonica are really not rushing to deploy LTE. Instead they are looking to use the efficiencies of HSPA+ (which is a software upgrade vs. a new and expensive network layout for LTE) and backhaul fiber to get the net effective capacity benefits as a new deployment of LTE would. Additionally, these operators are waiting for the LTE ecosystem to mature so that the advantages of economies of scale can be had, some of the early technology kinks can be worked out and there are ample LTE devices in the market to make a difference. However, in countries where LTE networks are being introduced, HSPA only operators will be at a decidedly marketing disadvantage. HSPA+/LTE will also help in managing the signaling load with the introduction of Continuous Packet Connectivity which helps manage the resources more efficiently by reducing the “state” switches from idle to active and vice-versa. However, higher speeds are addictive. This has proven throughout history. Once users get used to faster speeds, there is no going back63 and it can be used as a competitive advantage until the competition catches up.

59





http://teliasonera.com/Global/Media/Cision/487480_EN_wkr0011.pdf

http://teliasonera.com/Global/Media/Cision/487480_EN_wkr0011.pdf



Upgrading Backhaul64 The backhaul investments consume up to 30% of OPEX of a typical operator. Depending on whether the operator is leasing the bandwidth vs. owning it; the CAPEX can be more than double the backhaul OPEX. To be both economically sound and technically prudent, the next generation of backhaul implementation needs to have the following key characteristics:

Adaptive and Scalable Traffic growth patterns can be uncertain and difficult to plan for completely so solutions need to be adaptive and can scale with demand on a short-notice without re-architecting or significant new investment. The backhaul architecture should also accommodate different physical infrastructure in various markets. On the business front, the solution should provide flexibility of either leasing or owning the connection to the aggregation backhaul network. This helps in both scalability as well ensure that the future enhancements can be easily accommodated.

Coexisting of solutions/Support for legacy Due to legacy issues, any new solution will need to interface and work with the existing infrastructure to enable easy transition into the next generation architecture. This allows for graceful upgrade and better control over CAPEX. Fiber and microwave are both viable, and will coexist with other solutions for a long time. Addresses CAPEX/OPEX restraints Operators are looking to work under restrained OPEX and CAPEX requirements. The emergence of 4G and long term evolution (LTE) networks only compounds the challenge of under-engineered backhaul networks. According to ABI Research, the global CAPEX for cellular backhaul is expected to reach $23 billion by 2012 from $9B in 2009 and OPEX expenditures reaching $6 billion in 2012 from just over $2B in 2009. The total cost of ownership (TCO) needs to be manageable and should be proportional to demand. Long-term solutions In 2009, to meet the demands of iPhone and other smartphones, AT&T deployed 100,000 additional T1/E1s.65 However, this clearly is not a good long-term solution. To be cost-effective and future-prudent, operators will need to get ahead of the problem and architect their backhaul solutions for the long-term. Postponing the problem till next quarter will get you only so far.

The backhaul bandwidth management is a multi-dimensional issue that involves legacy, time to market, capacity, anticipated growth in demand, time to revenue, power consumption, OPEX, CAPEX, market and competitive dynamics, spectrum, and the availability of technologies. Operators have to take all of these into consideration as they plan for their current and future backhaul bandwidth needs. For example, while fiber might look like a good option in some cases, there is a high upfront cost of laying the fiber which can be as high as $100-200K/mile. Additionally, permits are needed and it can take 9-15 months for the whole project. Microwave has strong penetration but it requires licensing of spectrum, right of way approvals, and despite the lower upfront cost (compared to fiber), it can still cost $20-$40K/link. The need for spectrum licensing, RF analysis and roof leases can take up to 6-9 months. The replacement cost of equipment is also quite high which can drive the overall TCO higher and thus needs to be built into the OPEX budgets. As we alluded to earlier in the paper, to be most cost-effective and future-prudent, operators will need to adopt a hybrid approach that gives them the most flexibility in any given situation. In a saturated market, operators compete at a market level, and hence require ability to scale and deploy bandwidth on demand. It can be only done by using a toolbox approach of using technologies that best help fulfill the short-term requirements while keeping the investment future-safe.

64

For a more in-depth treatment of the backhaul evolution, please see our paper “Role of Optical Wireless Broadband in the Evolving Mobile Ecosystem” http://chetansharma.com/opticalwirelessbroadband.htm 65

http://ow.ly/167WTx

http://chetansharma.com/opticalwirelessbroadband.htm

http://ow.ly/167WTx



We will also see operators cooperating on sharing the 4G infrastructure and follow the lead of Bharti and Sprint in outsourcing some of the operations to really focus on what they can compete well on.

Data Offload - Femtocell/Wi-Fi/Core Over 60-80% of mobile data is consumed within the confines of a building, and most of the times, it is within homes or workplaces. This provides an excellent opportunity to off-load data traffic to the Wireline broadband connection and is especially important for the congested residential areas and for the heavy users. The consumption patterns in such areas and for such users need to be well understood to devise effective offloading strategies.66 In fact, it can be argued that in light of the data demand, wireline and wireless ought to be thought of as one network and regulators and operators should be focused on getting as deep a fiber deployment as possible. By encouraging users to deploy femtocells – both Cellular and Wi-Fi (along with Wi-Fi hotspots and corporate67 Wi-Fi access points), carriers will benefit in multiple ways, namely, getting into the home media and communication management business and better indoor coverage (in addition to network cost savings). If a majority of the data users were to deploy femtocells over the course of next 4-5 years, we can expect lowering of the network traffic cost by at least 25-35% depending on the extent of the Femtocell and Wi-Fi deployments.68 In fact, it might make sense to off load certain types of Internet traffic (browser, video, audio) right away or at the core before the traffic hits the SGSN and GGSN. This reduces the cost of carrying the bit as well as reduces the latency. Operators will also have to reassess their enterprise offload strategy that helps them better manage their highly valued customers. Carriers who are laying down their off-loading strategy now will be better prepared to handle the traffic load that will start to accelerate further around 2011-12.

Congestion Management The significant cost of managing a network is in the planning for “peak traffic capacity.” If this capacity can be managed through policy, quality of service, and congestion management, the peak time traffic costs can be lowered by 10-20% across the network. Only a small % of the users are typically consuming majority of the data. Additionally, only a fraction of the users typically go over the monthly capped limits (for example, the 5GB/month limit). If their network usage can be managed, the cost savings will be meaningful. Also, now is right time to start educating consumers about the “Quality of Service” and associated network traffic performances. The earlier we start doing that, the better off we will be. Wireless spectrum is a finite resource and it needs to be managed as such. It is reasonable to correlate Quality of Service with tiered service plans especially for peak traffic hours. Additionally, congestion management should also fine-tune traffic by application for example; the streaming traffic is more sensitive to delays vs. an application download or a browsing session. A financial transaction is more sensitive than a twitter update. As such, the network should have the ability to handle traffic by priority and importance of the bits as assigned by the application provider, the carrier, and the consumer (for some twitter update might be more important than a bank transfer going through).

66

For more discussion on data offload strategies, please see Bridgewater Systems whitepaper “Sharing the Load” 67

There could be interesting business models that can be developed that allow for significant offloading of traffic within the corporate premises. 68


http://www.bridgewatersystems.com/Document-Library.aspx



The debate over Net-Neutrality looms large in many countries especially in the US. This also means that the regulators will have to play a more constructive role in assessing how network growth might occur and what are the best ways to move the industry forward while preserving consumer interests at large. Regulators will serve their nation better if they worry about the market competitiveness more than the network management. Market forces have generally proven to be better guardians.

Caching Like Wireline, users often consume similar content on wireless connection as well. Whether it is going to the same websites or the blogs or downloading the same applications or P2P sharing or streaming the same clips, the pattern remains the same. Wireless networks can benefit by using caching at various levels in the network to offload the core network from repetitious transfer of same data. There are differences in caching for Wireline vs. wireless for example, wireless sessions are usually shorter especially on smartphones and featurephones, the content freshness is more stringent requirement in mobile, caching capability of mobile devices is not as robust as Wireline, etc. As such, the caching technology will need to take into account the specific state of affairs in a given market but overall caching will help in managing the data traffic.

Incentives to fill the troughs As we noted earlier, congestion management is about managing peaks to keep the incremental capacity needs to the minimum. If consumer behavior can be changed through incentives and education to use non-peak hours for bandwidth hungry applications and services, then the burden during peak hours can be lowered, sometimes significantly. Several types of incentive schemes can be designed to help shape consumer consumption patterns. This puts the emphasis back on Mbps (capacity) rather than MB consumed.

Network Optimization Compression and transcoding have been around since the mobile web was envisioned some 10-12 years back. Whether it is WAP, full browser, applications, web services, video and communication sessions – all can benefit from optimization on both ends, the device and the network. With full-browsers starting to become quite popular, the data traffic per web page delivered has started to rise. Similarly, application-based traffic is increasing but there are several opportunities for optimization from compressing and/or transcoding individual objects within a frame, page, or stream to the use of device cache or network-end points or content cache servers, data streams can be optimized.69 Developers should be required to adhere to strict traffic requirements so as not to make their applications overly chatty (unless they absolutely must). Operators should enable network APIs that can inform applications of network state in real-time so that the traffic can be adapted in real-time.

69

Companies such as Bytemobile and Openwave provide solutions in this area



Figure 17. Increasing the throughput by optimizing mobile data at the transport

layer70 Over the last year, many new innovative optimization concepts (besides traditional caching and optimization which have shown clear cost savings) have come to the market like TCP optimization that helps in increasing the bandwidth and reducing the latency (see figure 17). This is especially beneficial during congestion periods as such a solution can allow capacity recovery of 20-25%.71

Adopt Broadcast Mobile Video Some 5 years ago, mobile video broke into the wireless landscape with a lot of promise but the uptake thus far has been less than stellar. There are two main reasons for the disappointing performance a) quality and b) pricing. While we have come a long ways from the 1-2fps video delivery services in the early days, there are still quality issues with mobile video on the cellular network especially if you are trying to do live video. As proven on Wireline, video is a huge driver for the traffic demand. Over time, it is natural for this trend to follow on to the mobile world. According to the Cisco VNI Global Mobile Data Traffic Forecast, video will be responsible for the majority of the traffic growth until 2014 accounting for 64% of the overall traffic.72 Already, there are a number of smartphones where 40-60% of traffic is video. As more live video and video conferencing become the norm, we will need to have strategies to optimize the ROI and monetize the content streams. However, doing live video entirely over cellular doesn’t make sense for cost and performance reasons. Operators will have to evaluate mobile broadcast technologies such as MediaFLO, DVB-H, iMB (Integrated Mobile Broadcast73, etc.). They take the load off from the otherwise burdened cellular networks. Obviously, one must build a compelling value proposition for broadcast mobile video for widespread adoption. Operators will still need to have an optimization strategy for video on demand.

70

Source: Decreasing Network Congestion by Optimizing Mobile Data at the Transport Layer, Mobidia, 2010 71

Source: Mobidia Transport Optimization Test Report, 2010 72

http://www.cisco.com/en/US/netsol/ns827/networking_solutions_sub_solution.html 73

iMB is a 3GPP Release 8 technique that incorporates the broadcast mobile TV into the mobile network infrastructure but uses separate TDD spectrum.

http://www.cisco.com/en/US/netsol/ns827/networking_solutions_sub_solution.html


28 Ecosystem needs to step up as well | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Ecosystem needs to step up as well Operators often get blamed for the network congestion issues, however, the content and developer ecosystem needs to step up to the plate as well. If the optimization is done at the source, benefits can be enormous. For their part, Operators must enable and empower developers with network APIs that help developers control their content flow to devices that benefits the user experience and the network. This will be a win-win for all parties involved. Such APIs should be standardized across operators so developer load is minimum. Operators should set up testing labs and provide simulation tools to developers so they are more informed about the traffic load generated by their applications and are better prepared to address data consumption issues. OEMs should consider providing caching framework and tools that help migrate content during off peak hours. Only by working together can the ecosystem address the network data issues in the long run.

New Business Models and Services It should be noted that the low-cost bandwidth services (like SMS) might end up being the biggest revenue generator. However, the network must be planned for traffic in aggregate and have enough flexibility in the business models to help discover new revenue generating services. It is quite apparent that the current business models74 and pricing schemes will be inadequate to maintain the levels of current profitability. If the revenue equation stays flat with price pressure and the cost equation is only going up (at accelerated rates), at a certain point in the time graph, the cost of delivering data services overtakes the revenue being generated from them. So, new business models are needed that take the bandwidth consumption into account to manage the traffic especially during peak times.

Figure 18. Reallocation of Consumer Spending (US) for Access services75

74

Some business models are only going to be new to certain countries. For example, the flat rate pricing phenomenon is not widespread. 75



29 New Business Models and Services | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Operators typically have great intelligence on voice usage but for data, the infrastructure and efforts are generally not on par. There is little understanding of what consumers are doing, which applications and services they are tuned to at any given instant, forecasting traffic spikes, etc. As cost of supporting data services exceeds the cost of managing voice services and as the revenues from data services become more prominent than those from voice services, operators will have to pay much more attention to the specifics at a very granular level and design business models and pricing plans per the trends and forecasts. If we take a look at the spending habits of the US consumers on “access and communication services” which includes the spending on Telephone, Cable, Internet, and Cell phones, the total “access” spending over the course of last decade has been consistently around 4% of the total personal income per capita (figure 18). However, the share of each of the services has been changing steadily. Telephone used to have 65% share of the spending but is going to be below 30% by end of 2010. Others have been climbing at the expense of telephone revenues, especially the cellphones which since 2007 command the highest share. So, the overall spending has stayed constant while there has been significant reallocation of spending. Similarly, within cell phone services, data has gone from being less than 1% of the overall revenues to over 35% in 2010 and is going to be more than 50% of the overall revenue mix by early 2013.76 Mobile operators need to figure out how to manage these reallocation undercurrents and maintain the overall life time value of the customer. It will come from re-architecting of the business and technology practices as well through the introduction of new services. There are practical limits to how much subscribers can talk. For example, in the US which is the most talkative country on the planet in terms of MOU (Minutes of Use),77 voice traffic grew only 3.5% YOY in 2009.78 Same can’t be said of data traffic however, which grew over 180% during the same time-period.79 The voice side of the equation can be set free with unlimited plans as the incremental cost of adding capacity is fairly low compared to the data side where the incremental cost to boost capacity is relatively quite high. As the mobile networks add not only the subscribers but also the data-enabled devices (including sensors and vertically focused devices) that can be connected to these networks, data side of the equation will dwarf voice traffic very shortly (please see figure 4). This paradigm shift towards data is altering the economics of the industry and the markets which are in tune with the shifts of time will be able to respond better to the growing consumer expectations. It should be noted that over the last 10 years there has been a gradual move from on-deck traffic to off-deck traffic with on-deck accounting for very little traffic in most developed markets. So, operators will have to rethink business models that are just based on selling bandwidth. They need to migrate to models that are more based on value to the end customer and the ecosystem. In fact, it will be wise to figure out the business models prior to the technology investments discussed in the previous section.

Network as a Platform As we have discussed in the paper, the migration to data services is creating tremors in the ecosystem. The business of access and service is fundamentally going through a transition, slower in some places than others. Some operators have resigned to the fate of the being the wholesaler for the network and operate a access network that others in the ecosystem can develop services and application on and yes, monetize accordingly. However, operators should take a longer term view of the “opportunity” and view “network as a platform” for innovation. At times, the innovative applications might come from the ecosystem but the bigger role will be that of an “enabler” that helps the ecosystem to keep marching ahead at a fierce

76

Source: US Wireless Market Update, Q1 2010, Chetan Sharma Consulting, 2010 77

The US has more than twice the MOUs (at 839) of the highest ranked European country and double the MOUs of ANY country with nearest being Canada at 435. Source: Merrill Lynch, “Global Wireless Matrix 3Q 2009.” 78




30 Conclusions | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

pace. By focusing on building the access network as the platform, operators can build a long-term value proposition and competitive advantage.

Conclusions It is clearly a very exciting time in the mobile industry. The growth has been spectacular; its resilience a model for other industries, and its promise is something that keeps entrepreneurs on their toes. Leading operators are already seeing data revenues overtaking voice revenues for the first time. Gradually, rest of the world will follow suit. However, this growth comes at with the cost of managing growth of data consumption from billions of devices and trillions of sensors around us. In the coming years, there will be two types of opportunities that will be created, one that take advantage of the data being generated in a way that enhances the user experience and provides value and the other in technologies that help manage the traffic data that will continue to grow exponentially. To be able to stay ahead of the demand, significant planning needs to go in to deal with the bits and bytes that are already exploding. New technical and business solutions will be needed to manage the growth and profit from the services. Relying on only one solution won’t be an effective strategy to manage rising data demand. A holistic approach to managing data traffic is needed and our analysis shows that the cost structure can be reduced by more than half if a suite of solutions are deployed vs. a single dimensional approach and thus bringing the hockey stick curves of data cost more in line with the revenues and thus preserving the margins. The decision making process within the operator organizations will need to be streamlined as well. Operators should also consider creating a senior post which focuses on both the cost side and the solution side so they can devise and institute a sustainable long-term policy and keep the margins healthy. By introducing new business models and technology solutions such as femtocells, congestion management, optimization, broadcast video, new types of devices and others, carriers can manage the growth in the yottabyte era without a negative impact on their profits.


31 Acknowledgements | © Copyright 2010, All Rights Reserved. Copying without permission is prohibited.

Acknowledgements The author acknowledges assistance from Sunil Jain, Jeff Giard, Joanne Steinberg, Dr. Mallik Tatipamula, Dr. Hugh Bradlow, and Sarla Sharma in reviewing the paper. Disclaimer: Some of the companies mentioned in this paper are clients of Chetan Sharma Consulting.

About Chetan Sharma Consulting Chetan Sharma Consulting is one of the most respected management consulting and strategic advisory firm in the mobile space. We are focused on evolving trends, emerging challenges and opportunities, new business models and technology advances that will take our mobile communications industry to the next level. Our expertise is in developing innovation-driven product and IP strategy for clients worldwide. Our clients range from small startups with disruptive ideas to multinational conglomerates looking for an edge. We assist major brands formulate winning, profitable, and sustainable strategies. Chetan Sharma Consulting has significant experience and expertise in understanding mobile data traffic and devising strategies to cope with the growth as well as benefit from it.

Please visit us at www.chetansharma.com

About the Author Chetan Sharma is President of Chetan Sharma Consulting and is one of the leading strategists in the mobile industry. Executives from wireless companies around the world seek his accurate predictions, independent insights, and actionable recommendations. He has served as an advisor to senior executive management of several Fortune 100 companies in the wireless space and is probably the only industry strategist who has advised each of the top 6 global mobile data operators. Some of his clients include NTT DoCoMo, Disney, KTF, China Mobile, Toyota, Comcast, Motorola, FedEx, Sony, Samsung, Alcatel Lucent, KDDI, Virgin Mobile, Sprint Nextel, Skype, AT&T Wireless, Reuters, Juniper, Qualcomm, Comverse, Motricity, Reliance Infocomm, SAP, Merrill Lynch, Openwave, Ricoh, American Express, and Hewlett-Packard.

Chetan is the author or co-author of five best-selling books on wireless including Mobile Advertising: Supercharge your brand in the exploding wireless market and Wireless Broadband: Conflict and Convergence. His books have been adopted in several corporate training programs and university courses at NYU, Stanford, and Tokyo University. His research work is widely quoted in the industry. Chetan is interviewed frequently by leading international media publications such as Time, New York Times, BBC, CNN, Wall Street Journal, Business Week, Japan Media Review, Mobile Communications International, and GigaOM, and has appeared on NPR, BBN, WBBN, and CNBC as a wireless data technology expert.

Chetan is an advisor to CEOs and CTOs of some of the leading wireless technology companies on product strategy and Intellectual Property (IP) development, and serves on the advisory board of several companies. He is also one of the most sought after IP strategist and expert witness in the wireless industry and has worked on and testified in some of the most important cases in the industry such as Qualcomm vs. Broadcom, Samsung vs. Ericsson, Sprint vs. Verizon, and Upaid vs. Satyam. Chetan is a senior member of IEEE, IEEE Communications Society, and IEEE Computers Society. Chetan has Master of Science degree in Electrical Engineering from Kansas State University and Bachelor of Science degree from the Indian Institute of Technology, Roorkee.

http://www.chetansharma.com/clients.htm

http://www.chetansharma.com/

Managing Growth and Profits in the Yottabyte Era Second Edition

Documents

Transcript of Managing Growth and Profits in the Yottabyte Era Second Edition