Power Behind The Throne

Trivia question. Who built the first Internet firewall and when? The surprising
answer to the second part of the question is 1992. Firewalls today are ubiquitous.
Then, with the Net just coming out of its university-government incubator phase,
they didn’t exist because there was no perceived need for them.

The answer to the first part of the question may be lost in the mists of time.
However, we do know who built the second Internet firewall. It was present-day
fixed wireless network software developer KarlNet
of Dublin, OH…. Or so says KarlNet co-founder and director of business development
Doug Harman.

At almost the same time KarlNet developed one of the first fixed wireless base
stations for outdoor use and has been a key back-room player in the development
of the wireless industry ever since. With OEM (original equipment manufacturer)
customers such as Lucent ORiNOCO, Dell, Sony and Toshiba
using its software, including its high-performance TurboCell protocol for fixed
wireless networks, KarlNet’s influence and reputation are out of proportion
to its size.

In 1992, KarlNet founder and chief technology officer Doug Karl was on the
computer science faculty at Ohio
State University
(OSU) in Columbus and managing the school’s Internet-connected
network. It was a wild, wild west environment. The first generation of student
network hackers had figured out they could infiltrate any system on the network
— and print their resumes on somebody else’s printer, for example. There was
nothing to stop them.

So Karl built a firewall. "It was made with standard computer parts —
mother boards, network interface cards — and programmed in assembly code,"
Harman recalls. "Doug was told it would never work. But he ended up with
a very low-cost firewall." He then deployed it successfully in the OSU

At that point, Karl persuaded Harman, a friend with no technical background
but solid marketing chops, to help him launch KarlNet to commercialize the firewall.
At first it was just a way for both to make some extra cash. They ran the company
part time while Harman operated his retail distribution business and Karl continued
at OSU.

Also about this time, Karl built a fixed wireless network base station, again
using standard PC components along with circuit board radios and off-the-shelf
antennas. "That made a pretty big splash," Harman says. "We were
doing something unique and selling at a price point that was very hard to beat."
The fixed wireless systems very quickly became the company’s lead products.

Then big competitors like Lucent — soon to be customers as it turned out —
persuaded KarlNet it should give up making complete systems and sell its market-leading
fixed wireless software as an OEM firmware offering. KarlNet put the software
on a circuit board and sold it to companies like Lucent which integrated it
with their own radios and antennas to create the final product.

The company has never looked back. "We’re one of the few companies that
have been profitable right from month one," Harman says.

Although it remains small, with a staff of fewer than 30, KarlNet has seen
revenues double every year until quite recently when the economic slow-down
flattened sales growth. Even still, it has seen average year-over-year revenue
growth of better than 30 percent in the past few years, Harman says. He won’t
reveal current annual revenues.

Today, KarlNet’s core intellectual property is the TurboCell wireless network
protocol. The basic technology is a trade secret protected only by copyrights
and trademarks, though some recently added security functions are patent pending.

TurboCell is something Karl started developing with the help of graduate students
while he was still at OSU. Many of those graduate students now work at KarlNet.
"A bunch of very smart guys have spent virtually their entire adult lives
perfecting this," Harman says. "Doug [Karl] likes to say there’s about
50 man-years invested in it."

TurboCell has three innovative features that set it apart from standard 802.11
technology: adaptive dynamic polling, super packet aggregation and bandwidth

The adaptive dynamic polling algorithm ensures that base stations poll idle
remote stations with less frequency than remotes with traffic pending, thus
providing more bandwidth for the active remotes. Dynamic polling is something
most network systems developers have ignored until quite recently, Harman says.
Now a few others are offering it.

While remote stations wait to be polled, the super packet aggregation feature
bundles packets together in a super frame so they’re all sent together. This
eliminates some of the per-packet network overhead, resulting in significant
efficiencies. Small packets of less than 100 bytes — e-mails, notices of receipt,
page requests, etc. — transmit six times faster than on standard 802.11 networks,
KarlNet claims. And small packets represent as much as 65 percent of traffic
on an 802.11 network, Harman points out.

The bandwidth control feature allows wireless network operators to throttle
bandwidth to and from remote stations by reducing the frequency of polling of
lower-bandwidth stations. This allows wireless ISPs to offer different prices
and service levels and optimizes bandwidth use on the network by preventing
remote stations from transmitting as fast as they can.

Standard Wi-Fi client devices in networks that use the TurboCell technology
must use a proprietary software driver to communicate on the network. Some OEM
vendors produce dual-radio base stations with one radio working in TurboCell
mode and another working in standard 802.11 mode, Harman notes.

About 80 percent of the software the company writes for wireless systems is
applicable to indoor wireless LAN applications as well as outdoor fixed wireless
systems, he points out. The other 20 percent is TurboCell.

Part of KarlNet’s core business is licensing its software and firmware — including
TurboCell and its routing and firewall software — directly to major OEMs like
Lucent ORiNOCO, Sony, Toshiba, etc. They incorporate it in the indoor wireless
LAN and outdoor fixed wireless products they build.

Another part is selling to smaller OEMs, VARs and systems
integrators who custom build systems for Tier two and three wireless ISPs (WISPs)
and major enterprises and institutions.

There are other name-brand OEM customers, Harman notes, but most don’t want
competitors and customers to know they use KarlNet and don’t develop their own
software. The company has about 300 customers altogether. "That may not
sound like many," says Harman. "But it’s all relative. It’s actually
a pretty good size number when you’re talking about the fixed wireless market."

Although its market share or penetration is hard to gauge, Harman believes
the company’s software could be in as many as 45 percent of all fixed wireless
systems deployed in North America.

KarlNet also sells integrated hardware-software systems — base stations designed
for enterprise and WISPs, remote stations, customer premises equipment and repeaters
— that incorporate its software and other vendors’ radios and other electronics.

Harman maintains that KarlNet has few, if any, direct competitors. Most wireless
network systems developers that don’t use KarlNet software develop their own
or use software from "one-man shops." Cisco Systems with its Aeronet
technology has something comparable but doesn’t make it available to OEMs.

Harman sees much of the company’s future growth coming from the enterprise
market, especially the market for enterprise hotspot networks on corporate and
institutional campuses. KarlNet will have products that solve security and business
model problems in those networks, he says. It is also working on new technology
that will help lower the cost of customer premises equipment in fixed wireless

The company’s other major new initiative is opening up European and Asian markets.

KarlNet is that rare thing, a pioneer with no arrows. Its OEM strategy and
fiscal conservativism have helped make it both an unsung pillar of the industry
and a very successful company.

802.11 Planet Conference

News Around the Web