DARPA is looking for a platform that can tell whether Internet of Things devices have been hijacked based on fluctuations in the heat, electromagnetic waves and sound they put out as well as the power they use.
The agency wants technology that can decipher these analog waves and reveal what IoT devices are up to in their digital realms, according to a DARPA announcement seeking research proposals under the name “Leveraging the Analog Domain for Security (LADS)”.
The LADS program would separate security monitoring from the device itself so if it is compromised, the monitoring platform can’t be affected.
The ambitious goal of the four-year LADS program is to come up with a monitor that can read specifically what instructions and functions are being executed or what part of memory is being accessed simply by deciphering the analog byproduct of the digital activity.
This is similar to side-channel attacks that are used to help break encryption devices but turned around to help detect attacks instead.
This type of security device has already been developed in a rudimentary form, but DARPA is hoping for a quantum leap in its effectiveness.
Under the guidelines, the monitor itself could have its own power supply or could be an ASIC or FPGA tied to the power bus on the IoT motherboard or be connected via a USB port that draws power only.
Since IoT devices typically have limited storage, memory and processor speed, they are too puny to support endpoint security the way servers and laptops can, DARPA says.
The agency wants researchers to figure out what analog emanations are useful to detect specific details about the state of the device, such as whether the firmware is uncompromised, if unauthorized code is executing and whether configuration settings have been modified. It seeks a tool that can detect anomalous behavior that might indicate attacks.
Researchers would also be asked to figure out how such monitoring devices would be affected by their distance from the IoT device being monitored and whether they can monitor specific devices in an environment that includes ambient electromagnetic noise from nearby similar devices. DARPA is hoping for a monitoring device that can be deployed and then keep an eye on multiple IoT devices, which would be more cost-effective than needing a dedicated monitor for each IoT device.
DARPA wants the researchers to consider software to run on the devices being monitored that would shape the emanations to make them easier to read or make them readable from greater distances. Similarly, it wants them to explore whether such software changes could minimize the amount of data attackers could glean from monitoring the emanations.
The program calls for working prototypes that can act as the basis for shifting security off IoT devices themselves, which is a broad goal of vendors who are concerned about IoT security. They recognize that these devices are generally inexpensive, which means they don’t have a lot of headroom for adding security.
It also means the type of functions they perform and data they transfer is limited and predictable, making it more likely that patterns can be identified that indicate a normal baseline. Deviations from the baseline could warrant investigation, and the more specifically monitors can label activity, the greater the accuracy of the security analysis.