Wi-Vi: See Through Walls with Wi-Fi Signals
Fadel Adib      Dina Katabi
HOME PROJECT DESIGN EVALUATION PRESS

Tracking Multiple Humans

The figures below show the Wi-Vi screen when there are one, two, and three persons moving in a room respectively. Each of the graphs always has a line at "zero" which corresponds to the residual DC (static objects). In addition to the zero line, the left-most window tracks one moving human indicated by the bright curved line other than the zero line. In contrast, the middle window tracks two moving humans; and the third tracks three moving humans indicated by the disconnected bright curves.

One Moving Person Two Moving Persons Three Moving Persons
One Person Two Persons Three Persons

Automatically Detecting the Number of Moving Objects

Wi-Vi uses the fact that a person cannot be in more than one location at any given point in time to automatically detect the number of moving objects. This is the underlying intution behind a metric it uses called spatial variance to determine this number. The table below shows that Wi-Vi can detect with 100% accuracy the presence or absence of a moving human. The accuracy of the current version of Wi-Vi drops to 85% and 90% in detecting two and three persons respectively.
Detected
Actual 0 1 2 3
0 100% 0% 0% 0%
1 0% 100% 0% 0%
2 0% 0% 85% 15%
3 0% 0% 10% 90%

Gesture Detection Accuracy

For a human to transmit a message to a computer wirelessly, she typically has to carry a wireless device. In contrast, Wi-Vi can enable a human who does not carry any wireless device to communicate commands or short messages to a receiver using simple gestures. Wi-Vi designates a pair of gestures as a '0' bit and a '1' bit. A human can compose these gestures to create messages that have different interpretations.

The left and right figures below show the accuracy of Wi-Vi in decoding these gestures (bits) at different distances and through different building materials.
Gesture Decoding with Distance Gesture Decoding in Different Building Materials
Accuracy vs Distance Accuracy vs Building Material

© Massachusetts Institute of Technology