This week at Mobile World Congress, NXP is demonstrating what it says is the world's first LTE/HSPA/UMTS/EDGE/GPRS/GSM multi-mode baseband platform. NXP achieves this feat thanks to the magic of software defined radio (SDR). At the heart of NXP's solution is the Embedded Vector Processor (EVP), a DSP that can "support multi-standard baseband processing with negligible area and power penalties – a significant breakthrough from previous generations of baseband architectures." (NXP's words, not mine.)
How is NXP able to pull this off? The trick is that the EVP is specifically optimized for OFDM. (For technical details on the EVP, check out this PDF.) Also, NXP isn't trying to handle the entire baseband processing load on its DSP. As NXP points out:
"There are still areas of baseband processing where software programmability may not be the answer and where hardwired, more dedicated submodules are more appropriate…
Software-defined radio is therefore unlikely to be a fully software-programmable solution. Instead, SDR will be a mix of programmability and software-controlled reconfigurability in the RF front end in which embedded microcontrollers, digital signal processors, vector processors and hardware accelerators all play their parts."
Smart thinking, if you ask me—and I predict that we'll see more talk like this in the coming years. As one example, look at startup 3Plus1's CoolEngine platform. Like the EVP, the CoolEngine is tuned specifically for OFDM baseband processing. Like the EVP, the CoolEngine, and combines a high degree of flexibility with a high degree of specialization.
It's also worth studying the history of the EVP. This architecture was developed by the NXP DSP Innovation center (also known as DSP-IC). This group began existence in 2001 as Philips spinout Adelante. The company struggled to find customers, and just two years later, Philips re-absorbed the group. In 2006, Philips spun out it semiconductor operations into NXP. In case you didn't follow that, the history is:
Philips → Adelante → Philips → NXP
I point this out because this group has been around a long time without a lot of commercial success. In other words, Philips (and its various spinouts) had to invest a lot of money to get to this point. (Happily, it is finding success this year.) While the story of the EVP is an extreme case, it illustrates how vendors will have to invest increasingly large amounts of money to succeed in our brave new world of DSP.