WiMAX chipset

frank_hu_zheng

1
TITLE: THE RACE FOR THE 802.16 CHIPSET MARKETS
The first 802.16-2004 Red d compliant chip is finally being shipped to BWA equipment
manufacturers by Montreal-based Wavesat Inc. The DM256 is a baseband IC
developed by Wavesat and manufactured in France by Atmel. It is being shipped only to
current Wavesat customers which issued POs varying from a few samples to several
thousands of units. The chip is said to meet all WiMAX and 802.16-2004 specifications
for both the base station and subscribers systems.
The Wavesat strategy from the time it decided to focus its energies exclusively on 802.16
baseband chips was to be the first to market, one way to differentiate itself from its rivals
and especially the giants Intel and Fujitsu. Wavesat has done just that.
Context
Wavesat has been working on OFDM technologies for more than 7 years to create its 6th
generation product, the DM256. . Founded in 1993, it started developing amplifiers for
the satellite and cellular industries . Much of its customer base was in the early stages of
the broadband wireless industry, albeit in the pre-standard days. However, in 2003, the
company changed directions and sold all of its primary technology in those areas to Mitec
Telecom. The portion that remained with Wavesat was its Broadband Wireless Access
business unit established in 1997.
“We lost the bulk of our revenue and employee base at that time,” Guay, CEO of
Wavesat said. The new version of Wavesat had 25 employees and revenue in 2003 of less
than $100,000. The number of employees has jumped to 52, with revenue increasing to
$1.5M by year-end 2004. He expects the head count to double in 2005 and revenue to
increase 10 times.
Selling the bulk of the company and refocusing on an emerging market put Wavesat back
into a startup mode, and it went to the venture market for funding to go with the income it
gained from the sale. Its latest round was for $10.5M and included Pac-Link
Management., Skypoint Capital, BDC Venture Capital, Monet Capital and Sunsino
Ventures Group.
Strategy
In the past year, Wavesat had been selling development kits to more than 30 companies
around the world with a special focus on Chinese companies. Development kits included:
digital cards, source code, FPGAs, and engineering support.
It is developing chipsets for both base station and customer-premise equipments. At the
same time, it is feverishly working to develop an 802.16e offering, the mobile version of
the WiMAX technology. This is the sector that looks to be the largest portion of the
2
WiMAX market, so it is even more important that the company has an established record
to show customers when that emerges. A company can miss in its first efforts in the fixed
broadband wireless market and still recover, but if the mobile version develops as
predicted, the margin for error will be much smaller.
The next logical step for Wavesat will be to sell reference kits with added features such
as RF cards and MACs from partners. Reference kits will go for $50K each. It has
already sold Development Kits to more than 30 customers.
What does it mean for the overall 802.16 chipset roadmap? A subscriber station or
CPE is composed of three main elements: The PHY which includes a base band, the
MAC (Medium Access Controller) and an analog RF front-end that serves as the means
to place signals into a specific frequency band.
Equipment vendors look to chip makers to provide complete reference designs, bill of
materials, components, software/firmware to manufacture WiMAX- certifiable
equipment.
Wavesat is shipping the base band or PHY layer for subscriber and base stations and
wants to remain a PHY only manufacturer. The price point of the current base band is
$35 for 10,000 units and lowers for larger volumes. Wavesat is partnering with Agere
Systems to integrate RF front-ends for the 5.8GHz band and with RF Magic for the
widely international 3.5 GHz band. The final development to an Integrated circuit will be
completed in the next 6 months with partner Atmel.
As far as the MAC, Wavesat is still developing partnerships with other companies due to
limited resources and time to develop, “ to develop a MAC requires 50 full time software
engineers” says Frank Draper, VP Marketing & Sales at Wavesat.
With a total of 40 engineers ( See Table 1) focusing on the development of the next
generation in-door chipset which will be called “Shark”, and later the mobile version
802.16 e, Wavesat believes it has enough on its plate for now and will thus integrate and
“distribute” MACs into its system on chip. The next generation chipset due at the end of
2005 will include subchanelisation on both the uplink and downlink to allow a gain of
12dB and thus better link budgets for true Plug & Play indoor installations.
To summarize, Wavesat is shipping basebands for subscriber and base stations to a
limited number of customers and claims general availability at the end of January 2005.
RF reference kit will include first the unlicensed 5.8GHz band.
So what about base stations?
3
Wavesat claims a low cost base station at $2,000 per sector can be produced by
integrating their base band DM256 along with a 1GHz processor and the RF front-end,
with API interfaces and drivers.
While silicon is available, equipment vendors have still to go through the WiMAX plug
fest and certification with at least 3 other vendors to be considered WiMAX certified.
The next generation chipsets for base stations will include four separate FPGAs that are
programmable with a basic MAC processor.
Overall Cost
Wavesat also announced a strategic partnership with EDOM Technology Co, for the
distribution of its chips in Taiwan and the rest of the region.
With basebands at $35 or lower at greater volumes, RF front-ends including filters, BPFs,
in the $35-$40 range and MAC processors at $20-25 with ARM 9, the total cost of
chipsets to manufacturers will initially vary around a base price point of $100. Therefore
the goal of providing subscriber stations at $250 or less is within reach, all depending on
when the mass volumes materialize. The market for fixed WiMAX will not necessarily
bring those volumes but it is a required step towards integration and lower equipment
cost for the future mobile standard which will require integrated ICs’ into laptops,
handsets and PDAs at much lower costs to compete against broadband mobile contenders
such as 802.20 or 3G.
Mobility
The mobile version of 802.16 (e) will be ratified somewhere in the middle or end of
2005. Meanwhile chipset makers are busy tuning up their development boards. While
OFDM is the modulation of choice for fixed applications, various modulations are being
pushed by various technology players for 802.16e with OFDMA being the leading
contender. This will not be good news for Wavesat who is pushing hard to promote
OFDM 256 FFT for mobile applications to guarantee backward compatibility.
“This will allow manufacturers and operators to easily evolve from fixed to full mobility
networks over the many generations of systems to come” adds Frank Draper.
How important is it to operators, to be able to use is whether the equipment purchased in
2005 will still be useable in future years. Will the carriers be able to amortize their
purchases over multiple years? Technologically this means: will base stations in the
future support the fixed applications and provide migration towards nomadic, portable
and even mobile applications?
4
“One box cannot do it all” says in substance Rupert Baines, VP Marketing with Picochip,
a UK-based chipset and software developer. The mobile version of WiMAX will be four
times more complex than the fixed version with different power and coverage
requirements. You need the capabilities of OFDMA to provide true mobility and
portability. Sai Subramanian, VP Marketing with Navini Networks agrees “successful
portable and mobile deployments will require higher link budgets and the use of smart
antennas. Scalable OFDMA will be the required modulation for mobility”.
Wavesat believes that it can use OFDM 256FFT as the PHYsical layer, in both fixed and
mobile systems to provide a natural evolution towards future systems. “Future base
stations will be able to recognize and operate both fixed and mobile applications, but to
do so the PHY must minimally have the same number of sub-carriers (256) or FFT (Fast
Fourrier Transforms)” says Frank Draper, VP Marketing with Wavesat.
“This has always been the premise on which the IEEE 802.16 committees have been
working, that is, to provide backward compatibility between mobile (802.16e) and fixed
(802.16-2004) systems” Frank added.
Competitive Landscape
So what about the other players? The table 1 below shows in more details the respective
product offering of the various companies involved in the development of 802.16
chipsets. In general most players will provide a system on chip to the equipment vendors
whether they decide to develop both MAC and PHY internally differ. All have to partner
with outside suppliers for the RF-front ends. On the mobility side, the battle will be
between OFDM and its cousin OFDMA modulation. The case of Motorola is unique as it
is the only equipment manufacturer developing its own chipset internally.
Table 1: Product Offering
5
One notable company that has silicon available is Pico Chip Designs, with its PC102 Pico
Array chip that has just started shipping. The company is providing reference designs for
customers seeking to build base stations around the technology. It announced a strategic
partnership with Airspan Network in October 2004 to deliver upgradeable WiMAX base
station reference designs called SoftMax. Airspan developed the MAC and higher layer
software while Pico chip is responsible for the PHY, the base band processor and
customer support.
Picochip also signed more recently a key agreement with the Chinese Institute of
Computing Technology (ICT) of the Chinese Academy of Sciences (CAS) to collaborate
on the acceleration of WiMAX in China. ICT will use picoChip’s industry- WiMAX
reference designs and PC102 processors.
In fact Picochip is believed to have a dozen licensing agreements already in place at the
time this article is going to press. Among them are believed to be at least 2 of the top
traditional FBWA market share holders. Picochip is clearly aiming at building a strong
position for the upcoming mobile version of WiMAX “ We have established strong
relationship with Korean partners for the upcoming WiBRO testing and deployments but
we cannot disclose who they are at this time” says Rupert Baines, Vice President of
Marketing with Picochip.
6
Sequans is another fabless semi-conductor company who recently reported major wins.
the availability of its FPGA platform and two strategic partnerships. The company has
signed agreements with Airspan Networks and Cambridge Broadband to deliver an
integrated WiMAX System-on-Chip (SoC), addressing both base station and subscriber
station with a comprehensive PHY and MAC offering. In reality Picochip and Sequans
are not completing each to serve Airspan needs :”The Sequans technology allows us to
complete a full range of WiMAX infrastructure products, including now a micro base
station offering with a similar architecture to successful WipLL product line”. said
Jonathan Paget, Chief Operating Officer of Airspan Networks.
Sequans is believed to charge substantial amount of money to partners they work with in
order to insure they are serious about their commitments. The company is expected to
announce further agreements in the weeks to come.
7
Table 2: Competitive Landscape & Roadmap
As shown in the table 2, different chipset vendors have different strategies and solutions.
While many are planning to provide a system on chip with PHY and MAC, some tend to
specialize on CPEs and less power devices like Sequans and Intel. Others suchas Wavesat
are building “one does it all” SOC for both CPEs and Base stations with little variants.
Some like Beecem and Runcom will leapfrog the fixed version of Wimax and
concentrate on 802.16e.
8
The Holy Grail is the mobile market of end-user devices. Surprisingly, start-ups entering
late in the game were able to secure significant funding from venture capital firms. I say
surprising since they will be competing against Intel and Fujitsu without the head start
that Wavesat enjoys.
In such an emerging market with leading figures such as Intel, to be the first to market
can be a powerful differentiation strategy but it will not be enough to survive. 2005 will
be critical to all chipset vendors as the deals they land now will decide their faith
tomorrow. Some may become attractive acquisition targets while others won’t even enjoy
this easy exit…
About the Author:
Adlane Fellah, MBA, is CEO and founder of Maravedis In a world-leader in market
research and analysis, specializing in BWA and VoIP markets . He is the author of the
landmark report "WiMAX, NLOS and Broadband Wireless Access (Sub-11Ghz)
Worldwide Market Analysis 2004-2008" which has been updated in February 2005. He is
a leading industry analyst on wireless broadband technologies and recently conducted an
extensive survey of regulators worldwide to build a unique BWA/WiMAX license
holders database.
Copyright 2004 Maravedis Inc. All rights reserved. This material may not be published, broadcast,
rewritten or redistributed.