Mobile Internet Data & Implications for the Wireless Industry

Introduction:

Due to the success of smart phones, app stores, and open platforms, a strong
consensus now holds that mobile data is creating the next telecom revolution. 
The volume of data transmitted over mobile wireless networks is skyrocketing,
growing at rates over 100% per year. 

But is that growth rate sustainable? Bandwidth per user will continue to be
limited by overall 3G-network capacity.  Moreover, considerable CAPEX would
be required to upgrade mobile networks to 4G.  Also, most subscribers are
unwilling to pay rates in excess of $50 or even $70 per month, which most mobile
operators need to get a reasonable ROI .  Yet wireless data growth conjures
up "visions of sugar plums" for many telecom professionals and wireless industry
stakeholders.

To examine the hype vs.  reality of mobile Internet data growth, we checked
in with Professor Andrew Odlyzko – a former AT&T Researcher who exposed the myth
of exponential (wireline) Internet traffic growth back in 1998 – just prior to
the dot com bust and optical networking melt down.  Odlyzko has been
tracking Internet traffic trends for over 12 years.  During most of that
time period, the actual traffic growth did not live up to the over-hyped
projections.

Based on an IGI Group tele-briefing and a follow up series of interviews and
email exchanges, we are pleased to bring you Professor Odlyzko’s analysis and
opinions. 

In this article, Odlyzko states why mobile Internet traffic growth will slow
substantially, that the actual growth rate will be very difficult to predict,
and that a large increase in CAPEX would be required to upgrade wireless network
capacity so that it could support the explosive growth in mobile Internet
traffic forecasted.  We’ll explore the implications of this for 3G and 4G
technologies, in conjunction with subscription policy and pricing plans of the
cellular network operators.  In a maverick point of view, Odlyzko suggests
that cellcos are missing a great opportunity by not pursuing higher quality
voice.  Finally, we examine if "seamless mobility" between wireless and
wireline networks is realizable.

Background:

While a researcher at AT&T, Dr. Odlyzko presented a talk on "Internet traffic
trends" to a small audience in Rio di Janeiro, Brasil at IEEE Globecom- 1999. 
His main theme was the fallacy of the enormous Internet traffic growth
projections that were commonplace at that time.  Odlyzko’s iconoclastic
view seemed to contradict what his employer was doing – AT&T was busy building
out its optical network backbone from 1998-2001.  Why would AT&T need a
huge core network transport capacity if it didn’t believe that Internet traffic
growth was exploding off the charts?

Fast forward ten years later at Infocom 2009 in April (ironically at the same
Sheraton hotel in Rio de Janeiro).  Professor Odlyzko presented a very
similar talk on Internet traffic trends.  The key difference this year was
that now the hype had shifted from wireline to wireless mobile networks. 
According to Odlyzko, "The contents of the two presentations did not differ all
that much in their essentials! Most of the false dogmas that were identifiable
then still hold the industry in thrall today."

The Reality of Mobile Data:

- According to an Accenture study in March 2009, 54% in US "don’t want or need
mobile video." The number of people who watched video on their mobile phone grew
from only 12% in late 2007 to 14% in late 2008.  This implies that any
takeoff in mobile video will not likely be rapid, especially when mobile
carriers are blocking video sites (AT&T currently blocks Slingbox media on the
3G iPhone).

- The worldwide mobile industry revenues were over $1 Trillion in 2008.  Of
that total, 80% came from voice while most of rest was from text messaging (SMS). 
But while Internet data revenue is very small, it still generates one-half of
cellular network traffic¹ and this will only increase with smart
phone penetration.  This poses a real dilemma for mobile operators-
Internet data traffic is using much more of their network capacity, but they get
the majority of their revenues from voice.

- Wireless traffic is growing at over 100% per year, far faster than wireless
technology is progressing.  (We have written extensively about 3G topology
tricks and other network upgrades that are happening very slowly).

- Traffic from smart phones is over 40 times that of a RIM Blackberry, which is
used primarily for email.  The more smart phones sold, the more Internet
traffic and the more pressure on overall cellular network capacity.  An
executive from Ericsson recently claimed, "the iPhone has given us a new purpose
for living."

- There is a low willingness to pay for mobile data by most subscribers, but
especially those in developing countries.  The exception is for messaging
and a few other services like email (but that’s not the rich content that could
be offered by a mobile broadband Internet).

-Voice is the "killer app" of yesterday, today, and tomorrow.  It has been
severely underestimated by the cellcos who offer very poor voice quality over
their mobile networks.  The opportunity for enhanced voice services is
discussed later in this article.

- Mobile operators continue to be fixated on content and control, even though
they would have to pay content providers for videos they deliver to their
subscribers.  The Financial Times refers to mobile operators as" plumbers,"
that will be providing dumb wireless pipes that users will fill up with their
own content.²  Odlyzko says, "The fraction of wireless capacity
that voice takes will soon be quite small, so the rest of the pipe will be
filled with various types of data.  User-generated data may well be the
best way for mobile operators to utilize that extra capacity (note the
popularity of You Tube, and that the user generated content creation does not
cost Google or the carriers anything).

What’s a Mobile Operator to Do?

There’s an urgent need for mobile network operators to protect their cash cows –
voice and SMS.  In order to achieve this objective, Professor Odlyzko
believes that mobile carriers have incentives to use "kludgy" technology
solutions to accommodate the growth of mobile data on their 3G networks. 
Those solutions don’t necessitate the huge increase in CAPEX that would be
required for 4G mobile networks.  The solutions might include some
combination of: off loading traffic to WiFi hotspots, pico-cells, more cell
towers, assigning mobile Internet subscribers to cells based on their traffic
patterns (rather than location), self organizing networks (many definitions of
this), femto-cells (for home or small office), and other topology tricks. 
In addition, higher capacity backhaul will be needed from cell towers to the
wireline network point of presence (PSTN or Internet). 

But that’s not all.  Significantly, the mobile operator will be throttling
data traffic in one form or another, according to Odlyzko.  The metering of
Internet traffic results in a data cap charging model, rather than a flat rate
for "all you can eat" traffic.  He compares this metering of mobile
Internet traffic to traffic lights that are strategically placed at the on-
ramps of highways to slow down the flow of cars.  Giving mobile Internet
users unlimited downstream and upstream data transport would most likely result
in saturation of network capacity, which would jeopardize highly profitable
voice traffic on 2G or 3G cellular networks.  Clearly, that must be avoided
at all costs.  Hence, blocking high bandwidth video apps.

What about blocking high bandwidth apps (i.e.  video)? For example, AT&T
currently blocks Slingbox video on its 3G network, but not on its WiFi hotspots. 
Odlyzko believes there will be many a fight between "network neutrality"
advocates and cellular operators over which applications are blocked on their
mobile networks. 

Should operators allow or prohibit smart phone tethering (mobile phone acts as a
cellular modem for an attached notebook PC – a feature of the 3G iPhone) on a 3G
network? Currently, AT&T doesn’t allow iPhone tethering.  Odlyzko thinks
that mobile operators might allow smart phone tethering, but only for certain
subscription plans, e.g.  in conjunction with a higher priced monthly
subscription with a long- term contract that locks-up the subscriber for several
years. 

Professor Odlyzko maintains that 3G networks are not adequate to keep up with
the explosive growth in mobile data traffic.  "3G will have difficulty
handling lots of video and it’s a stop-gap measure," he says.  Nonetheless,
3G or 3G+ based mobile data networks will not necessarily hit a brick wall that
would sharply curtail mobile data traffic and subscriber growth.  While
growth data growth will continue, 3G networks should also be used for delivering
better quality voice (more about this later in the article). 

The Difficulty in Quantifying Mobile Internet Traffic:

According to Odlyzko, mobile Internet traffic is likely to slow from over 100%
per year to 30-40% per year.  An accurate prediction, however, is extremely
difficult since mobile operators don’t readily make available their traffic
statistics or growth rates.  But equally important is that the combination
of topology changes, metering, blocking and tiered pricing plans tend to create
multiple feedback loops that will effect traffic growth in unknown ways. 
"There are all kinds of knobs that the mobile operator can twirl to regulate
mobile Internet traffic on its cellular network", says Odlyzko.  For
example, if cellcos offered flat rate/uncapped mobile Internet access for $80
per month, how many road warriors or wealthy individuals would sign up? He
thinks that $30 per month is needed for mass adoption of mobile Internet service
(and so does this author).  Price per megabit of data transferred will
continue to be a key issue for mobile network operators as will the CAPEX for
upgrading mobile networks to support much higher volumes of data traffic.

Along the way there will be strong tussles between open network initiatives
(e.g.  Google Android) that might encourage innovation vs.  the
"walled gardens" of cellular network operators (e.g.  AT&T) that limit
subscriber choices of devices and applications. 

Using 3G Networks to Deliver Higher Quality Voice & Services:

Odlyzko strongly believes that cellular operators have missed a great
opportunity to offer enhanced, higher quality voice services – something that
was envisioned when 3G was conceived more than two decades ago.  Most don’t
appreciate the "orality" of human culture- one reason video telephony never
succeeded.  Odlyzko states, "There are definitely opportunities for higher
quality voice.  Wireless voice quality is terrible? it has been selected to
be the lowest quality one can have that people wouldn’t reject."

Odlyzko suggests that wireless network operators wanting to get more value from
voice telephony should consider creating higher-quality premium-priced services. 
The cellular industry is so focused on video services that seems to have
forgotten that there are opportunities to improve wireless voice services and
make money in the process.  Here are a few examples of enhanced telephony
services envisioned:

- Higher quality voice and audio (does anyone remember the 7KHz audio ISDN
bearer service for higher quality voice and music?)
- Toll free calling, e.g.  free 1 800 calls
- Voice messaging that’s easy to use and access
- More attractive pricing plans (cellular voice now cost 10X more than wireline
voice and many free voice services are coming, e.g.  Skype and Google
Voice)
- Combining voice with SMS based transaction processing (e.g.  paying
taxicab fares)

There’s also the important question about mobile VoIP.  Will it cannibalize
the cellcos highly profitable cellular voice minutes? Odlyzko wonders how
cellular operators will be able to incorporate mobile VoIP into their service
and pricing plans.  Currently, AT&T blocks Skype on its 3G network and
Apple has blocked Google Voice on the iPhone. 

Thoughts on 4G, Mobile WiMAX & LTE:

Odlyzko thinks that 4G might take six years or more to happen in a big way. 
And a lot of CAPEX will be required for the necessary "forklift upgrade" of the
Radio Access Network, Backhaul, and Network Management elements.  4G
concepts are now so hazy that it is difficult to predict what kind of
performance and availability users might expect from such networks.  Even
when 4G is deployed, "seamless broadband mobility" is a fantasy because there
are too many complicated issues that need to be resolved. 

Mobile WiMAX has potential.  (Clearwire has said its CLEAR network delivers
on average 4 times the data rates of 3G networks at a lower price). 
Odlyzko thinks that if the subscription and device costs are low enough, mobile
WiMAX is likely to find many customers.  In fact, much of wireline traffic
could shift to mobile WiMAX, as mobility is valuable to many people. 
However, customer willingness to pay for just data, without voice- even for
mobile data-is not high.³  It is hard to see how Mobile WiMAX
could guarantee high quality for VoIP services and if it can’t, will be stuck
with low prices and low revenues, making its economics questionable. 
Odlyzko asks us to recall the initial excitement and subsequent commercial
failure of previous data-only wireless services, such as Ricochet. 

Odlyzko believes LTE will support voice by creating a separate logic network for
it.  He speculates that a LTE network might actually be composed of two
logically segregated "flat IP" networks: one for VoIP with different codecs and
bandwidth used depending on quality desired, and one for
data/video/multimedia/audio where much higher bandwidth is available.  The
two networks would be strictly segregated in terms of overall bandwidth and QOS,
but they might use the same physical transport and network infrastructure.

Is Seamless Mobility Possible?

Despite the widespread claims, Odlyzko believes that "seamless mobility" between
mobile wireless and wireline networks is unrealistic.  "The problem with
seamless mobility is the huge mismatch in bandwidth and traffic between wireless
and photonic (optical network) connections", says Odlyzko.  Wireless data
constitutes only about 1% of wireline data.  "So if you wanted to move all
traffic to wireless radio channels, you would need 100x growth in wireless
network capacity and CAPEX, which I don’t see happening any time soon." (Odlyzko
points out that the growth in wireless network capacity would have to be even
larger, since the wireline network capacity is still growing.)

What will the future bring? Odlyzko thinks that we are likely to have a universe
of many hot spots, with nearby fiber offering very high wireline speeds. 
There will also be relatively high (although lower) short-range wireless speeds. 
In between (i.e., almost everywhere else, in terms of area) we’ll have much
slower speeds commensurate with 3G, 3G+, or 4G mobile networks.

Summary and Conclusions:

- A combination of topology changes, tiered pricing, and mobile operator
policies will scale back mobile data traffic growth to 30% or 40% from over 100%
per year now.

- Mobile network operators will have a tough time offering true high quality
data services without negatively impacting their very profitable voice services. 
(This is because 3G networks are bandwidth limited, resulting in high volumes of
data consuming the bandwidth that would otherwise be used for voice calls.) The
result will be data caps and blocking of high bandwidth traffic types by the
mobile operators.

- There are likely to be a lot of "net neutrality" debates as cellcos block high
bandwidth consuming traffic types, e.g.  Slingbox, Bit Torrent, Hulu, etc
and also from prohibiting smart phone tethering.

- Cellcos should seriously consider providing enhanced voice services to 3G
subscribers.

- Mobile WiMAX has potential if the subscription price is low enough and good
quality voice capability is added as a service (VoIP over WiMAX).

- 4G is a very fuzzy paradigm, which will take longer to realize than most
experts believe.  When LTE goes mainstream, it may actually be two
logically segregated flat-IP networks.

Andrew Odlyzko:

Footnotes:

1 – Cellular voice and mobile data traffic are each around 70 petabyte/month,
in contrast to wireline Internet traffic of over 7,000 petabyte/month.

3 – With the exception of the HTC device sold only in Russia, no WiMAX phones
are available at this time.

References:

1 –
Internet pricing and the history of communications, Computer Networks 36
(2001), pp.  493-517