During the period of 1998-2002, the usual suspects (Motorola, Ericsson, Nokia) were all diligently examining SDR technology. The reason, as explained by senior designers and one CTO, was pragmatic: Each company had over two dozen "platforms" to maintain in each major market - US, Europe, etc. That was BACK THEN. Except for the spectrum in which each operator was authorized to use, the air interface protocols and networking stacks were "all up in the air" with evolving standards. As defined, SDR technology - although an "invisible technology" not seen by the consumer - was the only viable alternative, since "velcro radio" simply increased Size Weight Power (SWAP) and cost. Also, even in 2002 the continuation of Moore's Law looked to be me "good" for many more cycles, and by then, FPGA "process technology" was only 1 to 1.5 Moore's Law generations behind the ASICs... The only part of "SDR" not needed was continuously variable multi-band RF transceivers (demanded by the military) since even 2, 3 and 4 band RF chipsets need only function in rigidly harmonized spectrum bands.

The example of iPhone 3G/3Gs is an excellent one. In 2009, iPhone rapidly overtook the "other" premium SmartPhone, the BlackBerry Curve. By Christmas 2009, the iPhone is expected to sell over 21 million units worldwide, in comparison to projected sales of 6 million units worldwide (together, about 26% of the global cellphone market share!). This DESPITE the worst global recession since the Great Depression! Apple recently revealed a "deal" with China, the world's largest emerging cellphone market, likely using the Chinese WCDMA standard. With a re-negotiation of the AT&T exclusive contract coming up, the use of SDR technology for iPhone, BlackBerry and their other competition is crucial for these "Hot App" multimedia phones. I imagine the new "3G Netbooks" like the new Nokia will also be using SDR extensively for the same reason.

One troublesome issue, of course, is the potential damage done by "hacking" of widely deployed SDRs. This topic has been covered in past SDR Forum Technical Conferences, and will also be discussed in the coming SDR'09, and the good news is, there are countermeasures available, such as HAWCS(r).

Interesting discussion!

The ¨problem” with SDR is indeed the definition, or better the lack of a definition that everybody is happy with. In the broad sense one could indeed say that every smartphone today is using SDR because 'some' of the physical layer processing is done in software.

Let's take the example given by Lee, the Infineon PMB 8877 . Clearly this chip contains a DSP which is running (part) of the GSM/EDGE physical layer, so strictu sensu this can be considered 'SDR'.   But, do not try to map another waveform on this platform, it has been designed for this format only and if you would like to add for example WLAN to your product you will need to add additional components. In that sense it is a very unflexible architecture with a limited scope.  

Now, consider this platform from the same baseband supplier:  X-GOLD SDR 20. This chip also contains the ARM+Teaklite combo like the one above, but there's an additional array of SIMD cores that adds the horsepower to turn this device into a powerful programmable baseband processor that has the ability to switch between multiple standards by changing the software only. The chip comes with software packages for GSM, GPRS, EDGE. WCDMA and HSPA and is upgradeable by SW to LTE, Wifi and Mobile TV standards! I think we can all agree that this is a different level of SDR than the implementation above. Note that this is also the first platform that was explicitly labelled SDR by Infineon in their marketing material.

At the RF front-end side the story is somewhat different. There are quite some RF multi-band solutions popping up, supporting multiple standards in one monolithic IC. This bottom-up approach certainly helps to reduce the component count and cost of a design. It does however lack the flexibility of a software defined architecture which allows the center frequency and bandwidth of a RF transceiver to be reprogrammed over a wide range. Imec's Scaldio front-end is such an example, as well as Bitwave's RFIC and Lime's LMS-6002. I believe there is certainly a need for such flexibility. Think for example of the uncertainty that exists in various regions about which spectrum bands will be used for carrying LTE signals. With a fixed front-end IC in your design you cannot easily adapt to potential changes in assigned frequency bands. A flexible reconfigurable front-end on the other hand offers an elegant solution where the same hardware can be reused accros multiple products in various regions of the world. And of course there is the advantage of reducing the number of componenents by replacing several individual transceivers by a reconfigurable front-end IC. This results in a considerable reduction in size, time to market and bill of materials. Designing and taping out a chip in an advanced technology node like 45nm and beyond is also  becoming so expensive that the flexibility to use it in many applications and scenarios is very welcome!