It’s possible to transmit life-threatening signals to implanted medical devices with no prior knowledge of how the devices work, researchers in Belgium and the U.K. have demonstrated.
By intercepting and reverse-engineering the signals exchanged between a heart pacemaker-defibrillator and its programmer, the researchers found they could steal patient information, flatten the device’s battery, or send malicious messages to the pacemaker. The attacks they developed can be performed from up to five meters away using standard equipment — but more sophisticated antennas could increase this distance by tens or hundreds of times, they said.
“The consequences of these attacks can be fatal for patients as these messages can contain commands to deliver a shock or to disable a therapy,” the researchers wrote in a new paper examining the security of implantable cardioverter defibrillators (ICDs), which monitor heart rhythm and can deliver either low-power electrical signals to the heart, like a pacemaker, or stronger ones, like a defibrillator, to shock the heart back to a normal rhythm. They will present their findings at the Annual Computer Security Applications Conference (ACSAC) in Los Angeles next week.
At least 10 different types of pacemaker are vulnerable, according to the team, who work at the University of Leuven and University Hospital Gasthuisberg Leuven in Belgium, and the University of Birmingham in England. Their findings add to the evidence of severe security failings in programmable and connected medical devices such as ICDs.
They were able to reverse-engineer the protocol used by one of the pacemakers without access to any documentation, and this despite discovering that the manufacturer had made rudimentary attempts to obfuscate the data transmitted. Previous studies of such devices had found all communications were made in the clear.
“Reverse-engineering was possible by only using a black-box approach. Our results demonstrated that security by obscurity is a dangerous design approach that often conceals negligent designs,” they wrote, urging the medical devices industry to ditch weak proprietary systems for protecting communications in favor of more open and well-scrutinized security systems.
Among the attacks they demonstrated in their lab were breaches of privacy, in which they extracted medical records bearing the patient’s name from the device. In developing this attack, they discovered that data transmissions were obfuscated using a simple linear feedback shift register to XOR the data. At least 10 models of ICD use the same technique, they found.
They also showed how repeatedly sending a message to the ICD can prevent it from entering sleep mode. By maintaining the device in standby mode, they could prematurely drain its battery and lengthen the time during which it would accept messages that could lead to a more dangerous attack.
One saving grace for the ICDs tested is that, before they will accept any radio commands, they need to be activated by a magnetic programming head held within a few centimeters of the patient’s skin. For up to two hours after a communications session is opened in that way, though, the ICDs remained receptive to instructions not just from legitimate programming or diagnostic devices but also the researchers’ software-defined radio, making it possible to initiate an attack on a patient after they left a doctor’s office.
Until devices can be made that secure their communications better, the only short-term defense against such hijacking attacks is to carry a signal jammer, the researchers said. A longer-term approach would be to modify systems so that programmers can send a signal to ICDs putting them immediately into sleep mode at the end of a programming session, they said.
Previous reports of hackable medical devices have been dismissed by their manufacturers.
The researchers in Leuven and Birmingham said they had notified the manufacturer of the device they tested, and discussed their findings before publication.