Motion Sensing with Wi−Fi technology

[senmeis]

Hi,

I‘m reading the article „Wi−Fi Sensing: Revolutionizing Motion Sensing with Wi−Fi technology“ at https://www.onsemi.com/pub/Collateral/TND6334-D.PDF and interested in this technique very much. I just want to issue a related question which can hopefully be answered. As far as I understand the mentioned movement sensing is different for different scenarios so we have to recalibrate if changed to another place. Is it correct?

 

[D.A.(Tony)Stewart]

Recalibration is needed even if you move 1mm, or 1 degree of rotation. This is due to the scattering reflections for amplitude called Ricean Fading is even more sensitive for phase shifts.

 

[FvM]

The article doesn't claim speed or distance measurement, but detection of moving persons and to some degree, localization of WiFi peers. In the former case, some kind of sensitivity adjustment might be necessary, similar as we know it for ultrasonic or thermal IR motion sensors. In the latter case, we need to teach different regions.

Notice that WiFi doesn't provide TOF measurements, just magnitude and its variation patterns.

 

[D.A.(Tony)Stewart]

Don't underestimate the complexity of the software required or sensitivity of the motion of the transponder to errors.

By the same analogy, 3 kHz 56kbps modems had to compensate for noise, distance, and phase error by training with pseudo-white noise sequences to null the echo by group delay error vs f and flatten that response to make the SNR high again by echo cancellation. This process worked great if there was only 1 or 2 echos in the analog conversion, but beyond that DSP's were not designed to handle this due to the smallest echo of an extra delay beyond the design limitation. So often in the past, I had discovered when the audio was clear and loud, that a fax modem or a data modem would fail to connect because it was software based and did not have the DSP with 1 extra echo cancellation capability.

In Wi-Fi, if you have ever used an application that monitors the Wi-Fi RSSI signal in dBm vs bit-rate, as I have done many times, and you understand from Shannon's Law the relationship between SNR and Bits/Baud then you may appreciate the following. When Wi-Fi from multi-path distortion or is a weak signal <-80dBm or <-70 dBm (depending on bit rate), and a packet error or retry occurs there is always a CSI routine to retrain the path in order to equalize it for the optimum SNR in each sub-band of the channel. When this fails, it automatically negotiates to reduce the bit-rate until an adequate error-free channel is acquired. This was very common for 54 Mbps to down-shift to 11Mbps and lower on very distorted multi-path channels. Many times I had such a temporary problem, and I used this RSSI radar-like monitoring tool (forget name) to check my SNR and signal levels and understood that Ricean Fading (phase and amplitude cancellation) was degrading the signal with the very slightest of change in position of a Wi-Fi dongle or Laptop with respect to the router. Then I discovered by slow panning of the laptop in fractions of a mm when SNR was in the -80 to -70 dBm range, I could get a high speed error free path by bouncing off stationary tree leaves or the wall or whatever.

Thus, it is easy to imagine that h/w is capable of detecting path motion or demodulation of the error signals using the training sequences for getting a good channel. In some laptops, there were dual Wi-Fi antenna (diversity) such that if one antenna could not get a reliable signal, the other might be better.

So you can imagine or recall incidents where you thought you should have had a fair but adequate reception, but could not connect and then it changed suddenly from a slight repositioning.

Using Wi-Fi as a motion sensing is a very interesting topic new to me as I have designed similar sensors as other companies have also. I used a UHF -3dB splitter or Return Loss bridge for echo modulation in full duplex. In this ONSEMI article, they and the alliance members must have already acquired patents for future possibilities which are not available yet. As such they are safe to publish these opportunities not yet ready.

There is some history. https://en.wikipedia.org/wiki/WiFi_Sensing#Commercialization

https://wballiance.com/resource/wi-fi-sensing/