LONDON — For Qualcomm Inc. the cellular wireless future is bright and much of the color comes from the seeds of Snapdragon, the company's multi-mode development platform for wideband CDMA, CDMA2000 and higher data rate follow-on versions of the communications standards.

The platform integrates a 1-GHz control processor called Scorpion based on ARM's Cortex instruction set supported by an array of 600-MHz DSP and accelerator cores for baseband and video processing that offer 128-bit single instruction multiple data functionality.

Qualcomm expects the Snapdragon platform to form the basis of multiple phone and gaming units, with support for additional networking features such as Wi-Fi, Bluetooth and, depending on rates of adoption, different flavors of mobile TV and video streaming.

A first chip implementing Snapdragon is due to sample in the third quarter, according to Sanjay Jha, chief operating officer. “Getting this right is at the top of our priorities. We have been working on the concept and design for three years now. The micro-architecture of the Snapdragon was designed in-house; it is a 500 milliwatt part with huge capabilities for broadband, with great graphics functionality and possibilities for integrating different radios as and when the market demands.”

Jha, who was educated in the U.K. at the universities of Liverpool and Strathclyde, was one of the first chip designers recruited by Qualcomm. Impressive as the Scorpion and Snapdragon hardware looks set to be “it is the software that runs over it that is giving us huge advantages, and that is the case on most wireless silicon these days,” Jha stressed.

Announced in November 2005 and scheduled to be made on a 65-nm process the Scorpion processor issues up to two instructions per clock using speculative out-of-order techniques. The first OEM expected to take a Snapdragon product to market is Samsung, although the consumer giant declined to indicate what that product would be.

The wireless vision that drives his engineers is a future “with 1-Mbit per second broadband connectivity and perhaps 50 milliseconds latency. It mandates portable computing and almost all consumer electronic devices being wirelessly connected. It will happen, but only if we make it easy and make it happen seamlessly,” Jha told EE Times.

Scorpion has been developed under an architectural license on the ARMv7 instruction set and the resulting microprocessor core is also being used to drive Qualcomm’s Mobile Station Modem (MSM) family, giving eight times the performance of previous MSM devices.

Also high on Jha’s list of priorities is further development and exploitation of the company’s s OFDM (orthogonal frequency-division multiplexing) and OFDM-A (orthogonal frequency-division multiple access) intellectual property.

When asked whether OFDM, which is utilized in many streaming codecs, is the new CDMA for Qualcomm, Jha first stressed how fast the basic market for CDMA is growing, but noted that ODFM and OFDM-A will be “increasingly meaningful technologies for wireless broadband, and we have been working on and investing in these since 1997. OFDM will play a key technological role in technologies such as the Long Term Evolution of the 3G Partnership Project’s iteration of cellular, in mobile WiMax and 802.20 wireless broadband.”

Jha also stressed that Qualcomm’s own flavor of LTE, dubbed Ultra Mobile Broadband (UMB), being pushed by the 3GPP2 organization and seen as the route to faster data rates as will be based on OFDM and multiple-input multiple-output (MIMO) antenna technologies, but cautioned this does not mean compatibility between the two approaches or with mobile WiMAX. WiMAX, touted as potential rival to developing 3G mobile phone standards, is a bone of contention with Jha, and Qualcomm. “The bottom line is we don’t think it is exceptional technology and the business case is not convincing either.”

“As it stands, the standard is flawed; there are poor control channel structures, and the system will not do latency or hand off. Then there is the spectral efficiency. There are so many versions and radios that have to be supported – where is the scale, where is the economy, where is the ubiquity coming from? And if there is one thing that irks me more than anything about what its proponents and analysts say, it is that WiMax is and will be cheaper to deploy for an IP network than cellular. I have yet to understand why that should be so. From Release 7 on there is full support for IP and backhaul in HSDPA and HSPA, not to mention HSPAplus.”

Despite all this, Jha says Qualcomm cannot discount the technology and does have a team working on it.

As to other radio technologies Qualcomm could integrate into Snapdragon chips, Jha enthuses about Wi-Fi, especially after the acquisition late in 2006 of Airgo Networks Inc. (Palo Alto, Calif.) with its patents on MIMO technology and Wi-Fi expertise, though he cautions that for handsets, “the attach rate will have to increase significantly before we take the plunge.”

When asked by EE Times what plans he has for WiBree integration, Jha said: “We, of course, have to keep an eye on any evolution of Bluetooth networking, but I am not really convinced about how much this low-power technology brings to the party.”

Meanwhile Qualcomm is taking renewed interest in Europe with its venture capital arm embarking on what could be seen as a ‘charm offensive’ backed with €100 million and looking at companies working on mobile applications, platform software, handset components, network infrastructure and core technologies.