RAMBO arXiv 2409.02292: Air-Gap Data Leak via RAM Radio Signals
RAMBO is a proof-of-concept air-gap attack that leaks data from isolated computers by making their RAM and memory buses emit radio signals. The paper arXiv:2409.02292, by Mordechai Guri, matters because it shows that a machine with no network hardware can still act as a low-bandwidth radio transmitter once malware is running on it.
What RAMBO demonstrates
RAMBO demonstrates that an air-gapped computer can become a low-bandwidth radio transmitter without any added network hardware. The paper states that malware on a compromised computer can generate radio signals from memory buses, encoding sensitive information such as files, images, keylogging data, biometric information, and encryption keys. An attacker with software-defined radio (SDR) hardware and a simple off-the-shelf antenna can intercept the raw radio signals from a distance, then decode and translate them back into binary information.
The work was accepted to the Nordic Conference on Secure IT Systems in 2023. The arXiv v1 was submitted on September 3, 2024. The abstract frames the contribution as both an attack demonstration and a defensive warning: it presents the design, implementation, related work, evaluation results, and countermeasures to mitigate what it calls the out-of-band air-gap threat.
How the RAMBO attack works
The RAMBO attack has two halves: malware running on the air-gapped computer and an external SDR receiver. The malware generates radio signals from the computer's memory buses, using software-generated emissions rather than a dedicated radio transmitter. The receiver captures those emissions with an antenna and SDR hardware, and the attacker decodes the signal back into binary data.
The abstract does not specify the modulation scheme or the exact software routine used to shape memory-bus activity. It describes the effect and the exfiltration path: compromised software causes RAM and memory-bus emissions that can be received and decoded nearby. The attack therefore does not create a network connection in the conventional sense. It creates a covert physical channel that exists because digital electronics emit radio-frequency energy during normal operation.
Data rate and range
RAMBO's headline data rate is 1,000 bits per second, but range and reliability depend on how fast the signal is transmitted. The abstract states that the paper presents fast modification methods to leak data from air-gapped computers at 1000 bits per second. A Hacker News commenter summarized the paper's range and rate trade-off as follows:
With a slow transmission rate (t = 10ms), A transmission is maintained at a distance of 700 cm away. With medium transmission rate (t = 5ms), A transmitted is maintained at a distance of 450 cm away and BER of 3%-4%. With a fast transmission rate (t = 1ms), a transmission is maintained at a distance of 300 cm away and BER of 2%-4%.
At 10 milliseconds per bit, the raw rate is 100 bits per second; at 1 millisecond per bit, it is 1,000 bits per second. The longest-range mode is therefore much slower than the headline rate, and error correction reduces effective throughput further. Another commenter characterized the practical rate as 100 bits per second at 7 meters with clear line of sight, which is consistent with the slowest mode in the quoted evaluation.
Threat model and practical limits
RAMBO assumes that the attacker already has malware execution on the air-gapped computer and can place an SDR antenna within a few meters with clear line of sight. That threat model is the main source of skepticism in the Hacker News discussion. One commenter wrote:
Cool paper even if the attack isn't practical, requiring getting the malware on the computer and a SDR antenna within 7 meters with clear line of sight of the affected computer for 100 bits per second (less after applying error correction). Hard to imagine a scenario where you can pull off both of those, but you couldn't plug in a USB, grab the PC, or record the screen directly.
Another commenter argued that the bit rate is too low to matter and that human and procedural risks dominate:
These attacks are all incredibly impractical and pointless, the bit rate under the most charitable assumptions is pathetically low. But even more, it ignores the most significant risk, which is people.
A third commenter made the same point more bluntly:
With malware on the computer you can change the heat signature of the room of you're bored enough.
The paper's contribution is not a turnkey exfiltration tool. It is a demonstration that the physical channel exists and can carry encoded data under controlled conditions. The attack requires prior compromise, proximity, and patience. Its value is in showing that an air gap blocks network paths but does not eliminate all electromagnetic paths.
Defenses and mitigations
The paper states that it proposes countermeasures to mitigate the out-of-band air-gap threat, but the abstract does not enumerate them. The Hacker News discussion focused on physical shielding. One commenter wrote:
Air gaps need to include a faraday cage.
Another commenter argued that modern hardware makes radiated emissions easier to achieve because cases are less effective at containment:
Computers used to come in proper full metal cases. Now they don't. Even when you try to find a metal case for a personal build most cases have giant holes in the side (case windows) acting as slot antennas. And plastic case laptop designs are even worse.
A Faraday cage or shielded enclosure would attenuate the radio signal, and physical inspection plus emission testing are the corresponding controls. The discussion also raised power-line leakage as a related question: whether signals from the computer's power supply or wiring can carry information out of the air-gapped environment. The provided comments do not establish whether RAMBO signals survive the power supply, but the question shows that shielding only the radio path may not address every out-of-band channel.
Why the discussion matters
The Hacker News thread treats RAMBO less as a practical attack and more as a reminder that air gaps are not absolute. One commenter quoted Ed Skoudis of SANS fame:
At best, an air gap is a high-latency connection.
The historical context in the thread reinforces that point. One commenter described a 2016 PennApps demonstration that won the hackathon and was based on earlier work by Mordechai Guri:
We demonstrated this in 2016 at PennApps and won the hackathon. ... Our work was based on the same author Mordechai Guri's earlier work 5 months prior, referenced on GitHub (USENIX Security 15, pp. 849-864). ... Also, apparently the capability ceiling: 1Kb/s and a greater distance is demonstrated here compared to 11 years ago.
Other commenters pointed to older emission attacks against CRT screens, where video output could be captured without malware, and to Stuxnet as the famous air-gap case. The common lesson is that isolation is a property of an entire system, including its physical emissions, supply chain, and people, not just its network interfaces.
Context: a long line of air-gap exfiltration research
RAMBO is one entry in a broader class of out-of-band air-gap attacks. The thread names Van Eck phreaking against video output, power-line monitoring, heat signatures, and Stuxnet as related examples. RAMBO's specific contribution is the use of RAM and memory-bus emissions as a software-controlled radio source, with an SDR receiver decoding the signal at distances measured in meters.
The paper's own framing is defensive: it presents the attack and proposes countermeasures to mitigate the out-of-band air-gap threat. The practical takeaway is that air-gapped systems should be evaluated for unintended emissions, not only for network connectivity. The headline 1,000 bits per second is real in the paper's fast mode, but it comes with shorter range and higher error rates; the longest-range mode is much slower. That trade-off is why the discussion treats RAMBO as a demonstration of a physical channel rather than an easy exfiltration method.