They plan to power a dogtag sized Bluetooth device by harvesting typical WiFi signal power.
The antenna will be at most about half an inch on each side (0.0125 meters, for a total area of 0.00015625 square meters.
Near a wireless network station used in homes and offices, the field intensity is typically below 0.5 mW per square meter.
Further upthread, someone posted the power required by a modern, low power Bluetooth chip, in its lowest powered sleep mode it requires a current of 900nA and 1mA when idle. Let's say it operates on 1 volt (it's probably 1.2v) so we will actually underestimate its power requirements by saying in its lowest powered sleep mode it requires only 900nW (that's 0.0000009 watts).
At 0.5mW (or 0.005 watts) per sq. m, the maximum energy a dogtag sized device, assuming 100% efficiency (not possible) will be able to harvest is 0.005 * 0.00015625 watts of power. This will work out to about 780nW.
780nW is less than 900nW. QED.
And that's giving them the benefit of the doubt (100% power efficiency, and a lower working voltage for the Bluetooth chip that it probably uses, and only considering the chip being in sleep mode, its lowest possible power mode, and not considering at all any other circuits the device will need - it will need more than just a bluetooth module, it'll need at least a minimal microcontroller of some sort).