Bluetooth needs advanced subsystem IP

First-generation applications used three ICs-an ASIC with a host CPU running the Bluetooth application, another contai ning the Bluetooth baseband and the last the Bluetooth radio. While rather high in cost and power, this initial solution enabled some products to quickly add Bluetooth functionality as a luxury feature.

Second-generation Bluetooth solutions reduced the IC count to two, moving the baseband into either the radio IC or the application ASIC, sharing its CPU. Both are an improvement over the first generation, but not optimal for most applications. Combining the baseband with the radio compromises one or the other, since neither is in the optimal process technology. Moving the baseband into the digital ASIC as an IP block is a move in the right direction, but sharing the application CPU is an inefficient use of a 32-bit processor for real-time Bluetooth activity. It also makes for very complex software integration.

Third-generation Bluetooth solutions add a dedicated 8-bit CPU to handle the Bluetooth lower stack entirely within the baseband core, thus offloading the host CPU of all real-time Blue tooth processing. This architecture makes for modulelike integration into existing designs, while offering a partitioning more suited to the problem than previous generations. More complex applications use this baseband IP as a drop-in solution, while smaller applications (for example, keyboards and headsets) can be implemented entirely within the baseband's 8-bit CPU, eliminating the need for a separate host CPU.

Phil Dworsky is director of marketing in the DesignWare Star IP Program at Synopsys Inc. (Mountain View, Calif.).