01
The pixels
TrojPix changes pixel values in a controlled way. To the eye, the picture remains almost identical.
pixel → sampleUSENIX Security ’26 / TrojPix
The picture is the transmitter.
The screen shows nothing suspicious. Inside the video cable, tiny pixel changes can produce a controlled electromagnetic emission. TrojPix shows how a physically isolated network can still have an exit through its monitor.
Run the simulation
A visual explanation of a mechanism presented at USENIX Security ’26.
screen
IMAGE / 01
video cable
receiver
VISIBLE
≠
EMITTED
≠
EMITTED
8.1Mbps
reported peak throughput
208m
reported maximum range
9monitors
manufacturers tested
15cables
video cables tested
01 / the object
Look at the screen.
Follow the cable.
This is a visual reconstruction, not a radio device. The picture stays almost unchanged. The line beneath it shows how the signal moves in the explanatory model.
What the eye sees01
What the receiver can get02
No file recovered.
The simulation emits no radio signal and does not implement the TrojPix decoder.
The eye and the cable do not observe the same thing. In the paper, the change is calculated to remain imperceptible on screen while being detectable on the video cable.
02 / the signal path
From pixel to emission.
The paper overturns a simple assumption: if a network has no physical connection to the outside, its data cannot leave. The monitor adds a surface that this assumption does not cover.
02
The cable
The values become samples carried along the video cable. Their transitions can produce controllable electromagnetic emissions.
sample → EM03
The receiver
A receiver at a distance reads the variation and reconstructs the data. The link does not pass through the network.
EM → data03 / what the authors measured
A lab result that changes the room.
The authors evaluated the mechanism across nine commercial monitor manufacturers and fifteen commercial digital video cables. They reported two test modes, called fake screen-off and foreground embedding in the paper.
across9monitor manufacturers
across15video cables
in2test modes
04 / the important limit
A measured maximum is not a promise.
8.1 Mbps and 208 metres are the maximum reported results under the authors' tested conditions. They do not describe every monitor, cable, or room.
The page keeps the mechanism as an explanatory experience. It generates no electromagnetic emission, accepts no files, and does not reproduce the paper's decoder.
Read the primary source at USENIX