If there’s one constant in life, whether it’s fast horses, powerful cars, or
wireless networking, it’s that we all want more speed. Which is why, Texas
Instruments (TI) has been trying to put its 22Mbs version of IEEE 802.11b
(sometimes called 802.11b+) technology using Packet Binary Convolutional
Code (PBCC) since May 14, 2001. Now, over a year later, vendors like D-Link
and NDC are delivering Wireless Access
Points and network interface cards (NIC)s that can both work with existing 802.11b
equipment and double the throughput when used with compatible "b+"
equipment.
"Gosh," You might be asking if you’ve paid attention to wireless
standards, "wasn’t this what 802.11g was supposed to do?" Why, yes
it was, and eventually (say the third quarter of 2003) you may see standardized
IEEE 802.11g equipment delivering similar speeds.
The reason why we now have non-standard 802.11b+ delivering high-speed 2.4GHz
wireless networking while 802.11g is still wading its way through the IEEE certification
process is that Intersil, with its Complementary Code Keying-Orthogonal Frequency
Division Multiplexing (CCK-OFDM), and TI spent over a year fighting with each
other over whose fast 2.4GHz technology would be included in 802.11g.
This is no small matter. It’s a truism in the technology business that whoever
controls the standards wins. So, when Intersil, TI, and other concerned companies
sat down to work on building a better 802.11b, the stakes were high and the
resulting standard battle was messy.
The battle didn’t start with TI though. It seeds were planted when Alantro Communications
with Chris Heegard and Matthew Shoemake proposed PBCC as a way to increase 802.11b’s
speed to 22Mbps back in early 2000. Shoemake was appointed to head a new study
group for high throughput extensions to IEEE 802.11b. This group would later
become the IEEE 802.11g workgroup.
TI, recognizing PBCC’s potential, set about acquiring the privately owned Alantro
for about $300 million in stock on June 23 2000. By September 8 that year, TI
had completed the acquisition. The PBCC theory was in place, the committee was
at work, and TI had the resources to create the necessary chip sets.
So why weren’t we networking at 22Mbps in 2001? Part of the reason is technical.
CCK, which is what ordinary 802.11b uses, has a short blocklength for its eight
quadrature phase shift keying (QPSK) data symbols. PBCC, on the other hand,
uses 64-state symbols. In English, a PBCC symbol can carry more data, but it
also requires a more powerful digital signal processor at the access point and
the NIC to make that data available.
Another advantage PBCC has over CCK is that its ‘convolutional coding’ is a
method of forward error correcting that enables you to reduce the bit-rate error
without increasing your transmission power. In real life, this means you can
get a higher data transmission rate and expand your range, all while
not using any more power than a conventional 802.11b device.
Intersil, then, as now, an 802.11 powerhouse, saw that TI’s owned PBCC had
the potential to make its rival a major player in the 802.11 marketplace. With
that in mind, and their own combination of 802.11b’s CCK and 802.11a’s OFDM
technology at hand, Intersil brought out CCK-OFDM.
While CCK-OFDM lagged behind PBCC in development, it had the potential to deliver
up to 55Mpbs speeds at 2.4Ghz, whereas PBCC currently would top out at 33Mbps.
So the 802.11 committee had two major, and quite incompatible, ways of handling
high-speed 2.4GHz wireless traffic. The two companies then started marshalling
support from other vendors for their own standard.
Technically, the arguments from Intersil is that CCK-OFDM is more scaleable.
TI says that PBCC causes less interference with other 2.4GHz devices, is more
compatible with existing 802.11b, and has more range. Which side most vendors
fall on depends on their business partnerships and, to a lesser extent, whose
benchmarks you choose to believe in.
All the standard politics came to a head in May 2001 when the IEEE 802.11g
committee met in Orlando, FL, with Shoemake at the gavel. There, 58% voted for
Intersil’s while 42 percent voted for TI. PBCC was no longer in the running
for the 802.11g standard.
As the meeting progressed, though, Intersil was unable to come up with the
75% super-majority it needed for CCK-OFDM to become the 802.11g standard. The
politics then boiled over when Shoemake ruled that since CCK-OFDM hadn’t reached
the 75% mark, it would no longer be considered.
This did not go over well.
To make a long story short, CCK-OFDM was put back on the table, but there was
no time to have another vote on whether it should be made official.
Afterwards, with PBCC out of the running for official IEEE standardhood, TI
decided quickly to go ahead and introduce it in its chipsets anyway as a superset
of 802.11b with higher speed and more range. TI introduced the ACX100 wireless
chipset and vendors have taken to it like a duck to water.
That might seem like the end of the chapter on PBCC in 802.11g story, but
TI wasn’t done yet. In November 2001, the committee met and quickly deadlocked
again. After many attempts at compromise, Intersil and TI agreed to let both
incompatible standards into 802.11g. This odd deal was made because Intersil
couldn’t get a 75% supermajority. Pragmatically, had Intersil not agreed, 802.11b+
(even though not an official standard) along with the already approved and (then)
soon to be shipping 55Mbps 802.11a standard from Atheros would have left little
room for Intersil’s standard.
TI, on the other hand, wanted PBCC in the official IEEE 802.11g specification
because it would make their future products more attractive to industries that
demand adherence to standards.
The result? If 802.11g makes it from a draft to final in the third quarter
of 2003 without major changes, 802.11g will include two standard modes, the
old 802.11b and a version of 802.11a’s OFDM, and two optional modes, Intersil’s
CCK-OFDM and TI’s PBCC.
This hasn’t stopped chip vendors from making sample runs of chips with early
versions of 11g. While these experimental versions of 802.11g are only now emerging,
TI and its partners have been shipping PBCC b+ devices to customers for a few
months. In the end, while TI may have a hard time of it in the standards battles,
they may be on their way to winning the customer war.
