The researchers at the Ben-Gurion University of the Negev in Israel are at it again. Previously, university researchers came up with ways to get air-gapped computers -- the ones that have no direct connection with the Internet -- to give up data through means such as noise emitted by hard drives and fans, heat emissions and other physical means.
This time, the data is leaking out the electrical line that is powering the machine.
There is the obligatory snappy moniker for the method "PowerHammer" -- that sounds like a 1970s comic book hero.
The researchers describe the two variants of the method they propose as line level power-hammering and phase level power-hammering. Both change the power consumption of the machine, which is dependent on the CPU workload.
In the line level approach, the computer's power line is tapped to read the output data. But in the phase level attack, the data comes from measurements at the main electrical service panel. Taps can be non-invasive and the information be converted into digital form.
The attack establishes two frequencies to represent a "1" bit and a "0" bit.
The researchers obtained obtain transfer rates of 1,000 bits per second using the line level variant and 10 bits per second with the phase method. The rates were best on a PC, followed by servers and Internet of Things (IoT) devices.
Those kinds of data rates are obviously best for small amounts of data such as passwords and the like. But the researchers also came up with a 44-bit data frames that, included a preamble which would signal the start of the transmission, as well as 8 bits of CRC code at the end of the frame for error detection.
Mitigation can include monitoring power lines as well as use of power line filters.
These EMI filters are primarily designed for safety purposes, since noise generated by a device in the power network can affect other devices and cause them to malfunction. It also should be remembered that these filters work best at the 450kHz-30MHz frequency band.
The PowerHammer exploit works at frequencies lower than 24kHz, which may mean that such filters will be ineffectual.
The fundamentals of network security are being redefined -- don't get left in the dark by a DDoS attack! Join us in Austin from May 14-16 at the fifth annual Big Communications Event. There's still time to register and communications service providers get in free!
A software process that executes random workloads was also thought to serve as prevention. The random signals interfere with the transmissions of the malicious process. The main limit of this approach is that the random workloads weaken system performance and may be infeasible in some environments like real-time systems.
Along with all the previous ways of mechanically getting data out of an air-gapped computer, this is yet another one to monitor as well as to try and counter.
— Larry Loeb has written for many of the last century's major "dead tree" computer magazines, having been, among other things, a consulting editor for BYTE magazine and senior editor for the launch of WebWeek.