Probably taking this whole scam way too seriously... but anyway
I do research in RFID and backscatter radio technologies, and this stuff makes me absolutely cringe. I've already seen papers from "asia" let's say that have looked at the
math of "mesh" IoT devices that can "transmit power" to neighbours and perform communication. I see those papers as fantasy; the English was probably up to standard (or not checked) and math looked fancy: they absolutely went crazy with optimization problems on this 'application' . But never taken into account some realistic RF channel properties or real world applications... Sadly, for some conferences this is what "peer reviewed" means.
Anyway, back to EE. RF-DC conversion is indeed absolutely inefficient. Just take a look at page 32 of this survey paper on the state of RFID, energy harvesting and IoT:
https://ieeexplore.ieee.org/document/7404333Most RFID passive tags (which are RF powered) have power sensitivities in the order of -10dBm to -15dBm incident power. The best RF energy harvesters top out at an (impressive) 30-40% efficiency with incident powers of -15 to -20dBm. So principally, if this technology is to become ubiquitous (e.g. you can use it anywhere without obvious deadspots), then atleast that amount of incident power must be present for *ANY* slow charging (microWatt's!) to occur. Oh.. and then start the problems:
Most modern RF chipsets halt operation if an incident power of that magnitude is present. WiFi coordinators have collision avoidance that waits until the occupied channel(s) to become free again. The thresholds for this behaviour should be standardized, but turns it's not (some routers crap out at e.g. -80dBm blocker power, while others don't care). WiFi clients suffer from reduced throughput due to blocked subcarriers in the OFDM modulation.
Then there is also the problem that RF transceivers are not designed for in-band blockers, obviously, but also have limits for moderately strong out-of-band blockers. E.g. a Bluetooth transceiver may have a sensitivity of -95dBm, but also have an adjacent channel rejection ratio of only 50dB at 3 or 6MHz. That means it's sensitivity will go down proportional to an incident power that is stronger than -45dBm [assuming that ACRR remains a fixed value for any stronger incident power, which does not have to be case when e.g. clipping occurs].
And then not to say you can potentially damage RF frontends if a strong signal of e.g. > -10dBm is present. The aforementioned RFID tags or RF energy harvesters typically use detector-diode based circuits (e.g. Avago HSMS-28xx or Skyworks SMS7630/7621); which are discrete components that can handle powers up to 10dBm or so. But integrated circuits often cannot. Due to square law the variability of standing 1 or 2meters from a 30dB source is very large, and therefore, how is a "smart" or mesh network going to change anything to that? It's a fundamental problem of physics.
Ofcourse they will argue that transmission of wideband noise is going to solve this problem; but in my view, that will only problems worse. It will likely not only break WiFi, but a wideband Tx in 2.4GHz also kills Bluetooth FHSS interference mitigation. Bluetooth is typically quite resilient to work in noisy environments. IMHO it's also classic anecdote that impossible challenges steer to wideband solutions. RF-based localization technologies is also being researched for many years, and although I'm not an absolute expert in it, IIRC most accurate localization/ranging solutions also only work with ultra wideband signals. And a regulatory body is not going to allow that, because of the aforementioned issues.
So summarizing:
- To have sufficient power to charge anything, you will blow up any other surrounding electronic device (and bake your brain in the process)
.
- To work with low power transmitters, you're fighting low-end power limits of RF energy harvesters caused by silicon junction thresholds.
- Using many many wireless transmitters to illuminate 1 receiver is absolutely ridiculous.
At that point buy a cable. Or if it's a really wire-constrained application, use solar panels.