CLWR Network Report
CLWR Network Report
Executive Summary
After a year of aggressive market launches across the U.S., Clearwire Corp. is
becoming fairly well recognized for its pioneering use of the wireless broadband
technology known as WiMAX, the "Wi-Fi on steroids" technology that allows providers
like Clearwire to build "hotspots the size of a city."
While WiMAX's ability to provide broadband Internet access with cellular-like mobility is
certainly the most recognizable attribute of Clearwire's deployments, there is a lesser-
known but just as important level of innovation taking place inside the company's
network, from the connections to the Internet's core out through the radio towers and
down to the end-user devices.
In this paper we'll take an in-depth look at how Clearwire's innovative implementations
of both backhaul and core network infrastructure allows the company to build out its
network more quickly and cheaply than its cellular predecessors, and how the
company's broad spectrum portfolio gives it the ability to scale quickly to address the
imminent explosion in demand for mobile Internet connectivity.
Of these two, the use of microwave for backhaul is undoubtedly a huge change of pace,
so much so that Clearwire has had trouble finding enough qualified technicians to build
out its networks in a timely fashion. On one hand, it's easy to understand why since
microwave-based wireless transmission is historically seen as a less-reliable choice for
connectivity than a wire or a strand of optical fiber. But Clearwire's conservative
approach (it uses a mesh-type deployment where each tower has at least two wireless
paths back to the core) seems to be working well, and it may be imitated by more
providers going forward since the use of microwave backhaul can significantly speed up
tower deployment simply by eliminating the need to dig trenches for cables and fiber.
On the network infrastructure side, the economic and innovative power of an open IP
architecture model is very familiar to anyone who knows the history of enterprise
networking. In that business sphere the advent of open networking protocols in the
1990s (chiefly the Internet protocol, TCP/IP) helped enterprises move from the
expensive, proprietary protocols and gear of the IBM/DEC past to the highly
competitive, much cheaper and more flexible world of the open network model,
championed by the likes of Cisco, Nortel, 3Com, Hewlett-Packard and others.
Instead of being able to "lock in" enterprises with their proprietary protocols (which
would only work with gear from a single manufacturer), network equipment vendors in
the era of TCP/IP were instead forced to compete on innovation and pricing, a change
that led in part to the tremendous era of networking technical innovation and reduced
equipment prices the networking industry has witnessed since TCP/IP became
mainstream.
When cellular-based data services emerged, the highly complex nature of the
infrastructure (which had to convert data packets into signals that could be transmitted
by the voice-centric cellular networks, and vice versa) led to a return to the lock-in
phenomenon, where one supplier (chiefly big providers like Ericsson, Alcatel-Lucent
and Nokia Siemens Networks) provided end-to-end gear deployments that promised
service providers dependibility and reliability in exchange for a premium price.
Topping off such innovative approaches is Clearwire's crown jewel, its massive
spectrum position at the 2.5 GHz band. Unlike the cellular providers, who are already
testing the limits of their spectrum positions with their 3G deployments, Clearwire's
average spectrum depth (the amount of bandwidth it has in any given market) is around
150 MHz per market, a number anywhere from three to seven times as large as the
holdings for AT&T and Verizon in major markets across the country. While not yet fully
exploited, Clearwire's spectrum position gives the company plenty of running
room for future bandwidth demands as well as the ability to scale quickly and add
capacity to existing networks without having to "split cells" or add additional towers to
relieve network stress.
Other future attributes that Clearwire is building into its network today include the ability
to support "off the shelf" WiMAX devices that might include WiMAX-enabled cameras,
automobiles, or machine-to-machine devices like "smart" power meters. Clearwire is
also already working with device and application developers who are interested in using
the company's open application programming interfaces to network attributes like
location, presence and preference, setting the stage for a future where devices and
applications might make use of the network's knowledge to offer services directly
connected to where a user is and what that user might want to do.
If and when such hyper-connected devices and applications emerge will depend in part
upon Clearwire's ability to survive as a business in the increasingly competitive world of
wireless broadband services. Let's take a closer look now at how Clearwire's currently
deployed innovative strategies might help give the nascent national WiMAX provider a
business edge in both cost savings and time-to-market deployment speed.
"When you have no money, and you're a small company, and you are desperate to
differentiate yourself, you'd be amazed at what you can come up with," said Dr. John
Saw, Clearwire's Chief Technical Officer who has been with the company since its
inception -- his bio notes that he was the company's second employee hired. "The nice
thing about Clearwire is that the first day on the job, I had no legacy network to worry
about," said Saw, a veteran of AT&T's wireless operations before joining Clearwire.
"Craig [McCaw] told me let's not make the same mistakes that were made before."
One of the places Saw and Clearwire started innovating right away -- and this was
starting when the company was launched in 2004 -- was to figure out a better way to do
"backhaul," the term associated with bringing bandwidth from the Internet to the radio
towers. "From the first day we built the company we started asking what we were going
to do with backhaul," Saw said. In the buildout of previous cellular infrastructures, most
carriers used a pair of T-1 lines -- about 1.5 Mbps of bandwidth in each -- to provide
connectivity to their towers.
"That's enough [backhaul] to carry narrowband voice traffic, but we know that a couple
of T-1s is insufficient when you have a lot of bandwidth needs," Saw said. "If your
iPhone is slow, it might be the fact that AT&T's backhaul is completely full." Indeed,
AT&T announced in January of 2010 that it had spent the past year putting in an
additional 13,500 T-1 lines into just San Francisco and New York -- among the most
congested of its 3G markets -- along with 238 new optical backhaul lines.
• Design Criteria:
• 99.995% availability per PTP link
• Ring topology provides 99.999% network availability
• Diverse path using PBT (PBB-TE) from every site
Clearwire Confidential
Clearwire's early calculations on user demand, Saw said, led the company to believe
that it would need conservatively to provide 30 Mbps to 50 Mbps bandwidth to each of
its towers -- "That's 20 T-1s, or else you are going to need to bring optical fiber to the
sites," Saw said. While optical fiber connections could certainly support such bandwidth
needs, Clearwire had two expensive problems in the way of using that approach: The
cost of trenching the physical fiber to each tower location (which typically involves
digging up streets) and the cost of metro fiber facilities and fiber-based services.
Instead, the newest wireless broadband provider looked to the air when it came to its
own backhaul needs -- using technologies based on microwaves, which have long been
used to transmit television programs, long-distance phone calls and other
communications traffic.
"We didn't do this because we wanted to be different, we did it because we had to," Saw
said. "Clearwire does not own any fiber facilities, so we put a strong and heavy
emphasis on microwave backhaul."
A big problem, especially in 2004 when Clearwire started its initial buildout, was that
there were no vendors in sight with the equipment Saw wanted -- a microwave radio
that spoke Ethernet, so that the company could keep its flat, IP-based architecture
intact.
"When we asked for an Ethernet-based microwave radio, 5 years ago nobody even
knew how to spell that," Saw said. After many frustrating meetings, Saw and Clearwire
finally found the Canadian firm Dragonwave, whose Ethernet microwave radios fit the
bill. Saw and Clearwire used those radios to build what he describes as a "flat Layer 2
mesh network," where cheap Ethernet switches at each
tower site help establish a network that doesn't have a
single point of failure.
If you follow the cables down from the antennas on a typical "3G" cellular tower, they
almost always end up leading into a building the size of a largish tool shed or a small
garage -- typically a 10-foot by 12-foot (or larger) enclosure that houses all the complex
gear necessary to process, separate, administer and manage the phone companies'
mix of voice calls and data traffic.
Since Clearwire is building a data-only network, its tower-site infrastructure needs are
much simpler than cellular. To house a typical Clearwire tower site's power needs,
microwave backhaul, WiMAX antenna gear and an Ethernet switch takes merely an
enclosure the size of a large refrigerator, sitting on a 3-foot-by-5-foot space.
With far less gear than a comparable cellular site, Clearwire's Saw said his network is
PSTN
HLR
BSS (GSM /
EDGE) MSC
BTS SGSN GGSN
BSC Packet Internet
Backbone
MGW Network
Clearwire Cellular Clearwire
GGSN
Base Station Shelter Cabinet
RNC
BTS ATM
VOIP GW
Clearwire Confidential 1
not only obviously cheaper to operate but also more flexible, allowing for quicker and
closer deployment toward its customer base. "Where cellular providers get in trouble is
trying to shoehorn a packet network into a network built for another purpose," said Saw.
"That's why you have to put in boxes like a GGSN [a Gateway GPRS Support Node,
which supports traffic conversion between the cellular network and the packet networks
of the Internet], which cost an arm and a leg -- and they also impair performance. Nokia
and Ericsson make a killing selling those boxes. Those are very expensive boxes."
According to both Saw and Barry West, Clearwire's president of International operations
(and the former head of WiMAX operations at Sprint Nextel before its merger with
Clearwire), a typical WiMAX network is about eight to 10 times cheaper to build than a
3G cellular network covering the same area.
"Our average cost number per cell site, when you add in all the backhaul access, zoning
costs and everything else, is less than $150,000 per site -- and the real number is
actually much lower than that," Saw said. "You'll never get a cell site at that cost for 3G.
What we have done on the
network side is the lowest
cost approach we could get
to." The small costs and
size of Clearwire's tower
deployments also help the
company in
putting its network closer to
where its users are.
Clearwire's lean tower site infrastructure, Saw said, allows the company to "piggyback"
on existing cellular installations -- "since we use fiber to the antennas, not coax, we
don't weigh down the towers," Saw said. "I can go to tower guys who say they have no
more ground space and show them that we can just squeeze our cabinet in the corner.
Same for rooftops. Our implementation allows us to site in a lot of creative places,
where you might not otherwise be able to get a cell site in."
Inside its network, Clearwire also embraces an openness that will allow it to reduce
costs by introducing a competitiveness not typically found in cellular networks. In the
latest implementation of the WiMAX standard, there is a model that calls for true
open interfaces between infrastructure gear, like base stations, radios, and
administrative equipment like the Access Service Network (ASN) gateway, the
workhorse box of a WiMAX network that aggregates and distributes a wide range of
subscriber-related data, from session management information, billing data, traffic and
mobility management, quality of service and other administrative functions.
Clearwire tower site equipment cabinet, front; Cellular provider equipment building, rear. (Credit: Sidecut Reports)
To win a contract for Clearwire's new IP-based network, however, means that vendors
must comply with the open interfaces requirements. "What that really means to
Clearwire is that we can buy base stations from Motorola, Samsung or Huawei, and we
don't have to be beholden to also buy their ASN gateway," Saw said. "I can go find a
really low-cost ASN gateway from someone who makes ASN gateways for a living and
get a really good price -- say from WiChorus, Starent or some Taiwanese company in
the future who wants to build one as cheaply as possible. What this means is that the
cost of my core network is going to drop considerably."
Bruce Brda, senior vice president and general manager of the wireless networks
business at Motorola, said that opening up such internal interfaces inside a cellular
network allows Clearwire to act as its own system integrator, which could produce cost
savings and spur product innovation.
"These interfaces have never been open before -- now the carriers like Clearwire have
the ability to mix and match," Motorola's Brda said. "It's great to have a clean slate, and
to have the flexibility to pick the best vendors for the best elements. The downside is
control, in how you manage and debug a disparate network. That's a challenge."
But Kittur Nagesh, director of service provider marketing at Cisco, said the positives
outweigh the negatives when it comes to combining gear from different vendors.
"In some sense you can say Clearwire is now able to pick the best of breed to
meet their specific needs," Nagesh said. "Right now they have transport and Layer 4
to 7 gear from Cisco, and ASN gateways from WiChorus. When it's done right you can
actually combine the best of breed from multiple companies."
"It's a big step for us to open up the interfaces, but I really believe we have to have an
open network, open in many ways," Clearwire's Saw said. "Vendors like Ericsson have
been very successful, charging carriers a lot of money for gear. But I think you will start
to see a lot more of what we're doing in the 4G world."
For wireless network providers, there may be no more important asset than broadcast
spectrum -- the airwaves used to transmit signals from tower to user. While Wi-Fi
networks in the U.S. are typically run over "unlicensed" spectrum in a sort of free-for-all,
cellular providers and WiMAX providers like Clearwire use licensed spectrum bands,
reserved parts of the ether gained either via government auctions or through licensing
agreements with spectrum holders like schools, universities or other entities.
Though there are many complex nuances regarding wireless spectrum and data
networks, a simple rule of thumb for carriers is that the more licensed spectrum depth
Clearwire map showing how spectrum depth affects coverage and capacity. (Credit: Clearwire)
you have, the more bandwidth you can provide. And when it comes to spectrum
holdings, Clearwire currently has almost twice the total spectrum depth of either AT&T
or Verizon, and has as much as five to seven times more depth than the larger carriers
in the spectrum bands that each will use to carry its so-called "4G' network traffic.
Some quick caveats and explanations: Comparing wireless spectrum assets is not
exactly apples to apples, since the spectrum AT&T and Verizon plan to use for their
Long-Term Evolution (LTE) networks is in the 700 MHz band, while Clearwire's entire
spectrum portfolio is in the 2.5 GHz band. An entire research paper discussion could
take place arguing the merits and flaws of each band -- the 700 MHz spectrum, for
example, is regarded as being able to support better in-building reception than the
2.5GHz band, but its power may also make it harder and more expensive to manage
when it comes to interference issues. For the purpose of this paper, we will assume an
even comparison of spectrum slices since the main concern here -- providing bandwidth
to end users -- is roughly the same for both technologies, LTE and WiMAX, and for the
700 MHz and 2.5 GHz bands.
Diagram showing the relative amount of spectrum depth for different frequencies. (Credit:Clearwire)
describe the different "bands" -- as in 700 MHz or 2.5 GHz -- doesn't fully explain that
each "band" contains many different frequency channels; the 2.5 GHz band, for
example, actually includes frequencies from 2496 MHz to 2690 MHz, while the recently
auctioned bandwidth in the 700 MHz band includes frequencies from 696 MHz to 806
MHz. Depending on how the spectrum was sliced for distribution, companies can
compile spectrum "depth" -- as in chunks of those frequencies -- for their delivery
portfolio.
Clearwire, which built its spectrum assets through direct license acquisitions and also
via its 2008 merger of WiMAX assets with Sprint Nextel (which had a large license
portfolio at 2.5 GHz), has approximately 150 MHz of spectrum "depth" in most U.S.
markets. AT&T and Verizon, the biggest spenders in the government's auction at 700
MHz, have between 24 MHz to 30 MHz nationwide depth each in the 700 MHz band, a
figure somewhat murky since neither company provides exact details of its spectrum
holdings. AT&T and Verizon also hold spectrum at other frequencies for cellular, but
both have said that 700 MHz will be their primary frequency for "4G" LTE services.
The spectrum depth matters most when it comes time to bring bandwidth to users --
simply put, the more spectrum you have, the bigger "pipes" you can provide to
potential customers. Called "channels" in the wireless industry, both WiMAX and LTE
can use channels of 5 MHz or 10 MHz in size, with the bigger number offering more
bandwidth and support for more users per tower. By having such a large spectrum
portfolio, Saw said that Clearwire can easily add capacity by just adding more frequency
"channels" in any given market -- something Verizon or AT&T won't be able to do given
their more-limited portfolio.
"Right now I assign a 10 MHz channel on each of three sectors per tower, which
provides a very good ratio for reducing interference," said Saw. "So that's 30 MHz being
used already. But I have up to 12 channels of 10 MHz each that I could deploy in any
given market -- so if I need to add capacity in a given area, I can just keep adding
channels. This is something Verizon will never be able to do."
Since neither Verizon nor AT&T has an operating LTE network yet, it remains to be
seen how the companies will manage performance versus spectrum availability -- one
way to conserve spectrum is to slow down the bandwidth speed per user. Another way
to manage spectrum is to "split" cell sites, which means adding more towers closer
together so that users aren't competing for the same antenna. However -- if you
remember the earlier discussion of the costs of building, siting and supporting towers --
splitting cells isn't for the weak of wallet.
"Clearwire has enough spectrum to let me stack new channels all day long on our
existing towers, without having to split my cells," Saw said. "When you start splitting
cells, capital expenditures go through the roof because you are bringing in new towers
and everything else that goes with them."
Motorola's Brda said that the relatively small slices of spectrum AT&T and Verizon have
at 700 MHz may force the big carriers to be more creative with their 4G deployments.
"I don’t think the traditional mobile carriers have enough spectrum to meet all their
needs," Brda said. "They're going to have to mix and match spectrum for solutions."
Both AT&T and Verizon have said publicly that they plan to use their currently dormant
AWS (Advanced Wireless Services) spectrum (at around 1700 MHz and 2100 MHz) to
"help" with LTE services; however, such solutions also introduce more radios and more
power consumption to user devices, for added costs and performance drain.
Clearwire, on the other hand, has been reportedly contemplating "renting" some of its
spectrum to another provider (the most-repeated rumor involves T-Mobile USA), and
has publicly said that it has the ability -- should the need or demand arise -- to build a
separate network for LTE side-by-side with its WiMAX network.
"Some people miss the fact that Clearwire has 150 MHz in most markets and they're
only using 20 or 30 MHz right now," Cisco's Nagesh said. "With their next generation
mobile architecture and core, they're not limited to just having WiMAX."
Some of the device drought has to do with the fact that Clearwire simply doesn't have
the millions-strong kind of market-coverage or user numbers to get consumer device
manufacturers extremely interested. Yet.
"The example everyone gives [of a potential WiMAX-ready device] is a camera," Saw
said. "But you don't build a camera specifically to get onto a WiMAX network." However,
should a camera manufacturer so desire, it could perhaps "strike a small deal" with
Clearwire to support instant uploads of pictures or videos that could be shared instantly.
Amazon's Kindle operates under such a fashion, with the cost for connectivity built into
the price of the content downloads. "It's a challenge for us to ensure such devices can
be authenticated and authorized," said Saw. "But in the name of trying to open up the
network to the mobile Internet, we think we need to do things like that."
Also in the yet-to-be-tested category are Clearwire's proposed open network application
programming interfaces, programming hooks into network features like user location,
user activity or quality of service -- which could theoretically allow third parties to offer
network-aware services and applications (say, the ability to order a pizza from a
restaurant that knows you are nearby) in concert with Clearwire's WiMAX network.
"The creativity of the mobile Internet has not yet been unleashed -- all we've really seen
is the tip of the iceberg," Nagesh said. "Once you have true open networks, with open
APIs into attributes like location and security, developers will be confident they are
writing for networks, and not just a single device. Clearwire has shown a lot of
leadership here, because they have thought well beyond a WiMAX-only device. Their
network will eventually support a lot more capabilities than just pure transport."
Any Sidecut Report ("Report") and the information therein are the property of or
licensed to Sidecut Reports, and permission to use the same is granted to site-wide or
single-category Subscribers ("Subscribers") under the terms of this Subscriber License
Agreement ("Agreement"), which may be amended from time to time without notice.
When requesting a Report, you acknowledge that you are bound by the terms and
conditions of this Agreement and any amendments thereto. Sidecut Reports therefore
recommends that you review this page for amendments to this Agreement prior to
requesting any additional Reports.
OWNERSHIP RIGHTS
All reports are owned by Sidecut Reports and protected by federal, state and
international copyright, trademark and intellectual property laws, under and by
applicable treaties and/or conventions. Subscriber agrees not to export any Report into
any country that does not have copyright, trademark and intellectual property laws that
will protect Sidecut Reports' rights therein.
In case of any distribution of Reports under this Agreement, Subscriber agrees to retain
all copyright, trademark and other proprietary notices on the Reports.