Vendors set their sights on small cells, Wi-Fi, and LTE

Mobile World Congress (MWC) 2012 in Barcelona ended the first day of March and had record attendance of 67,000 from 205 countries, of which probably half the attendees could be counted as company executives, including some 3,500 CEOs.

Clearly, the hit of the show was Nokia’s 808 PureView camera/cellphone. With a 41 Megapixel (MP) sensor that can capture 4x zoom in full HD 1080p, the pixel count was mind-bending. The company also introduced the stunning Nokia 610, which provides a high-end Windows Mobile 7 smartphone experience for the mid-to-low-end market. It was probably #2 in attendee buzz, despite the introduction of Huawei’s expansive (and expensive) Ascend D smartphone with a quad application processor, billed as “the fastest smartphone on the planet” … even though it’s throttled at HSPA+ speeds but not LTE.

Small cells increase in popularity

At MWC, small cell wireless, which includes femtocells, picocells and anything short of a full macro base station, seemed to be a popular theme. Although market-dominant Mindspeed Technologies formalized its Picochip acquisition at MWC, a new chip entrant also emerged. Montreal-based Octasic Inc. demonstrated its OCT2224W baseband DSP in what is claimed to be the world’s smallest microcell/small cell platform, WhiteStar by Deltenna. Octasic’s OCT2224W SoC ships with a 3GPP PHY software library that offers standards-compliant GSM, EDGE, W-CDMA, HSPA+, and LTE protocol stacks. Deltenna adds a dual-core ARM Cortex-A9, a Wi-Fi access point, and GPS-based timing, and markets the device to OEMs, integrators, and operators. The company claims that up to 64 HSPA+ users in a 10 km cell can be supported.

Last month, the Femtocell Forum changed its name to the Small Cell Forum and expanded its scope. Good timing.

Intel claims Wi-Fi on dual-Atom die

Before getting too excited about Intel’s presentation at ISCC (which always seems to coincide too close to MWC dates), Intel does not yet have Wi-Fi on a production-ready x86 die. However, Intel did present a paper at the conference on their “Rose Point” 32 nm SoC that implements Wi-Fi on a dual-Atom die. The paper addresses the 2.4 GHz RF capability on the same x86 die that may also be clocked near 2.4 GHz, clearly an interference challenge for the RF circuitry. Note that this is not an 802.11n or other specific WLAN solution. Intel admits that the design probably won’t be ready for the market for “another year or two.”

Intel once designed a standalone 802.11g Wi-Fi chip and was probably the last major company to ship such a product (for laptops, of course). I remember my discussion with Pat Gelsinger who headed Intel’s R&D at the time. He was roundly critical of the unfair companies (like Broadcom) who dared to offer pre-standard 802.11g chips before the standard was finalized (and before Intel could ship product). If memory serves, Intel was working on that standalone Wi-Fi chip for at least three years before they finally shipped a product.

Texas Instruments was probably the first company to put significant digital RF on the same die with a processor, their GPRS DSP/ARM chip for cellular baseband at the time (operating at a much lower 850 MHz). Later, TI was able to produce single-die EDGE baseband/RF chips, but by then Broadcom and ST-Ericsson were also shipping single-die EDGE products. Now, the latter two companies “own” that segment of the market. Currently, TI, Broadcom, and Qualcomm Atheros are shipping hundreds of millions of “combo” cellphone peripheral chips that include not only 11n Wi-Fi, but also Bluetooth, FM, GPS, and more on a single die. Each combo chip has one or more microprocessors (OK, DSP baseband engines) along with RF circuitry. Of course, those companies don’t have an x86 processor – though they are all major producers of ARM-based (Cortex-A9) processors – but have chosen to keep RF on separate dies, and are doing very well with that approach to the market.

Since Intel’s SoC Wi-Fi solution won’t be ready for perhaps a couple of years, they had better engineer it for 802.11ac operation, which will probably be the norm for smartphones and tablets in 2014.

NVIDIA keeping the Icera name

Most companies that acquire others tend to dispense with the acquired company’s identity – witness the many acquisitions by, say, Intel, Broadcom, and Cisco. NVIDIA, though, is proud of its acquisition of baseband (and RF transceiver) house Icera Semiconductor and plans to keep the Icera identity with its cellular product line. Now NVIDIA can offer communications processors that include both baseband and application processors, the dominant market segment.

CogNovo impresses at MWC

At the previous MWC’11, UK-based CogNovo was demonstrating their LTE-A baseband in FPGA form. This year, at MWC’12, they were the only company demonstrating a real LTE-Advanced baseband chip. Their Samsung-fabbed chip was highlighted in Rhode & Schwarz’ booth as well as in their own suite. Moreover, they were demonstrating carrier aggregation (perhaps the main feature of LTE-A over LTE) at the R&S booth. Although CogNovo has real silicon now, it remains committed to licensing the technology. When asked if Samsung had licensed its IP, the company had no comment.

Cellphone and tablet chip trends

To provide product and market planners with calibration on the market, Forward Concepts is offering its new “Cellular Handset & Tablet Core Chip Trends’12” – an extensive market study that covers the core integrated circuits that enable cellphones and cellular-connected tablets. In this study, we don’t just track basebands and application processors, we also track and forecast RF transceivers, power amplifiers, and power management units in great detail. The 312-page PDF is available at for $3,850, including an enterprise-wide license. The report provides 2011 vendor market shares for each core cellphone chip type and forecasts units, ASPs, and revenues for all of them in detail through 2016.

As always, I invite your comments.

Will Strauss
President & Principal Analyst Forward Concepts