Manhasset, N.Y. - Magnolia Broadband Inc. says it has
achieved what it calls a breakthrough in CDMA handset design,
more than doubling network capacity for uplinks and downlinks
through the use of antenna diversity. By leveraging existing
hooks within CDMA standards and using patent-pending
algorithms and techniques, the three-year-old fabless company
claims to have achieved full, two-way antenna diversity
without the need for network upgrades or protocol changes.
The doubling of capacity is a direct result of Magnolia's
technology, and the increase can be traded off against
improved coverage, higher data rates and lower power to allow
an operator to dynamically optimize a user's connection,
according to the company. "Capacity, coverage and battery life
are the operators' chief concerns right now," said Haim Harel,
president and chief executive officer of Magnolia (Clinton,
N.J.).
Operators have been interested in antenna diversity for
some time, but "it has generally required costly network
infrastructure upgrades or protocol changes and has only been
performed on the downlink," Harel said. Handsets with antenna
diversity announced by Qualcomm and Nokia over the past 18
months, for example, have been limited to downlink
implementations, Harel said.
Separately, Tantivy Communications Inc. has proposed a
diversity-enabling proprietary protocol upgrade to CDMA called
I-CDMA. But "that's a small player pushing a proprietary
protocol that needs a network upgrade," said John Moon, senior
vice president of corporate development at Magnolia.
Harel said Magnolia's DiversityPlus technology requires no
infrastructure or protocol upgrades and operates on both the
uplink and downlink to achieve an improvement of between 4 and
5 dB in the signal-to-noise ratio (SNR). "It also reduces PA
[power amplifier] power consumption by up to 20 percent to get
a 30-minute increase in talk time," Harel said.
"Because CDMA is interference-limited, every decibel
increase in signal strength is an improvement in capacity,"
said Harel; hence the targeting of CDMA rather than GSM
handsets, despite the larger potential market for GSM. "With a
3-dB improvement," Harel said, "you can double network
capacity."
For downlink diversity, Magnolia's implementation is not
much different from current implementations in that it
leverages the well-known signal-quality indicator that's part
of the CDMA standard. It uses that per-symbol update to
perform equal-gain combining to achieve the downlink
diversity.
But differences are apparent in the uplink diversity
portion of the implementation, around which most of Magnolia's
intellectual property and six pending patents revolve. "While
the downlink has the quality indicator, not many have noticed
that the uplink has similar capability," said Harel. "We
change nothing; it's all there."
Magnolia's hook is the power control bit that follows
channel fading and tells a handset either to increase or
decrease output power depending on the channel impairments.
"While a small change takes you out of the null, this is a
brute-force technique," said Harel. "Instead we take that
[power control bit] signal and, rather than ramp the PA, we
change the phase to get a 4- to 5-dB improvement in signal
strength." Field tests show transmitter and receiver
improvements of 5 and 2 dB, respectively.
Exactly what Magnolia does with the control bit is what
Moon calls "our secret sauce"-and the chefs aren't talking.
The end result is implemented on two silicon germanium chips
that reside between the antennas and the
downconverter/upconverter sections of the CDMA transceiver.
The first chip contains the low-noise amplifier (LNA) with the
receive-side diversity circuitry; the second chip comprises
the power amplifier and the transmit-side diversity circuitry.
Both are linked to the host processor, which performs the
proprietary signal processing needed to form the complete
diversity solution.
"We don't add chips; we just replace already established
chips," said Harel, referring to the LNA and PA. "Also, while
the algorithms run on an ARM [processor], they're
processor-independent." The processing consumes less than 0.5
percent of an ARM7's or ARM9's processing ability, he
said.
Three physical implementations are planned. The first, the
MBC-1100 chip set, performs the baseband processing on a host
such as an ARM processor. The MBJ-1100 will perform the
diversity processing within the LNA and PA chips themselves,
while the MBS-1100 will include the upconverters and
downconverters to form a complete RFIC solution.
The company is using a 0.18-micron SiGe process from Jazz
Semiconductor, with packaging from Amkor Technology. Samples
are expected to be available early in the fourth quarter of
this year, with production by the third quarter of next
year.
The company said it has already completed field trials with
Samsung, Sprint PCS and SK Telecom. It expects to begin
second-phase trials by June or July.
See related chart
http://www.eet.com