A new method has been developed to steal data from offline machines using electromagnetic waves emitted by their power supplies.

So-called "isolated" PCs, those isolated from the public Internet, could have their data stolen from distances of more than six feet, and even through walls, by someone with a smartphone or laptop equipped with a special receiver. the experts warned. .

The method was developed by Mordechai Guri, a researcher at Ben Gurion University in Beersheba, Israel, who called it COVID-bit, possibly in reference to common social distancing rules preventing people from being in close proximity. los unos y los otros.

Fill the void (with air)

Sandboxes are most often deployed in institutions where highly sensitive data and tasks are handled, such as those related to energy, government, and military weaponry, making this new method a worrisome prospect.

First of all, the target system must have some malware pre-installed, which can only be done through physical access to the machine. This malware controls the load of the processor and the frequencies of its cores so that the power supply produces electromagnetic waves between 0 and 48 kHz.

Guri explained that the switching components within these systems create a square wave of electromagnetic radiation at specific frequencies when turned on and off during AC/DC conversion.

This wave can carry raw data, which can be decoded by those far from the machine with an antenna that can be easily connected to a mobile device's 3,5mm audio jack. A program on the device can then decode the raw data by applying a noise filter.

(Image credit: Shutterstock)

Guri tested his method on desktop computers, a laptop, and a Raspberry Pi 3, finding that the laptops were the hardest to hack because their power-saving credentials meant they didn't produce a strong enough electromagnetic signal.

Desktop computers, on the other hand, could transmit 500 bits per second (bps) with an error rate of between 0,01% and 0,8%, and 1000 bps with an error rate of up to 1,78%, which is still accurate enough for efficient data collection.

At this speed, a 10KB file could be transmitted in less than 90 seconds, and the raw data from an hour of activity on the destination machine could be sent in just 20 seconds. Such keylogging could also be broadcast live in real time.

As for the Pi 3, its low power supply meant that receiver distances were limited for successful data transmission.

ben gurion university

(Image credit: Opachevsky Irina / Shutterstock.com)

Guri recommends that isolated systems be kept secure by monitoring CPU loads and frequencies for suspicious or unusual activity. However, this can lead to many false positives, as these settings can vary widely during normal usage scenarios.

In addition, such monitoring increases the cost of processing, which means the possibility of reduced performance and increased power consumption.

A workaround is to lock the CPU at certain core frequencies, to prevent data from being decoded by its associated electromagnetic radiation. However, the downside here is that, as mentioned above, natural fluctuations in base frequencies are to be expected, so locking them will lead to reduced performance at some times and excessive usage at others.

Share This