Bluetooth’s embrace of Wibree presents an architectural dilemma

According to last week’s news, the Bluetooth special interest group has decided to embrace Nokia’s Wibree—a knock-off of Zigbee that does low-data-rate, low-power communications over a Bluetooth-like air interface. This makes a lot of sense from a market standpoint, but it creates a rather interesting challenge to the SoC developers who will have to support Wibree—now to be known as Bluetooth ultra-low-power, in dual-mode chips.

The issue is the dual-mode part. Nokia envisioned two different use profiles. Wibree could be used in a stand-alone mode in extremely low-power, low-data-rate devices like watches, heart monitors and, one supposes, hand tools. But Wibree could also be supported as a second mode in conventional Bluetooth devices, even interleaving Bluetooth and Wibree packets.

This raises an interesting architectural question. Often, very-low-power radios use entirely different architectures than even low-power short-range radios like the ones used in Bluetooth. And yet the whole point of having Wibree instead of Zigbee, from Nokia’s point of view, appears to have been to utilize existing Bluetooth hardware blocks as much as possible. So does the hapless architect simply reuse the existing Bluetooth radio, RF and baseband, and gain little or nothing in energy consumption? Or is it better to redesign the radio so that it can operate at two very different points—one for Bluetooth, and another, with lower receiver linearity and noise margin, lower transmitter power, and reduced clocks and voltages in the ADC and baseband for Wibree? Or at this point is it simpler to just design two radios side by side, an existing design for Bluetooth and an all-stops-pulled low-power design for Wibree? If you do that, can you switch both of them on and off quickly enough to actually save power in a scenario where there are both Bluetooth and Wibree packets running around, or is this just a lost cause?

I’ll be fascinated to listen to this debate as it unfolds.