Engineers Evaluate Consumer IoT Devices for Security Risks

If you’re in the market for an internet-connected garage door opener, doorbell, thermostat, security camera, yard irrigation system, slow cooker—or even a box of connected light bulbs — a new website can help you understand the security issues these shiny new devices might bring into your home.

Consumer-grade Internet of Things (IoT) devices aren’t exactly known for having tight security practices. To save purchasers from finding that out the hard way, researchers from the Georgia Institute of Technology and the University of North Carolina at Chapel Hill have done security assessments of representative devices, awarding scores ranging from 28 (an F) up to 100.

Their site, https://yourthings.info, shows rankings for 45 devices, though a total of 74 have been evaluated. That’s hardly a complete roundup of the tens of thousands of devices available, but the idea behind the project is to help consumers understand serious issues before connecting a new IoT device to their home networks.

“A lot of people who purchase these devices don’t fully understand the risks associated with installing them in their homes,” says Omar Alrawi, a graduate research assistant at Georgia Tech. “We want to provide insight by providing security ratings for devices we tested.”

Voice-activated personal digital assistants are among the most common home IoT devices, but if not properly installed, they provide unwanted access to the home networks to which they are connected, warns Manos Antonakakis, a cybersecurity researcher and associate professor in Georgia Tech’s School of Electrical and Computer Engineering.

“If you have an IoT app that is vulnerable, whoever has access to that app not only has access to your personal information but could also get into your home and eavesdrop on your conversations,” he says. “Anything connected in the home near the personal assistant could also interact with it. If there is vulnerable software running on the device, it could be exploited within the home network.”

One problem is that most home networks were set up for simple tasks such as sharing printers, so they lack the kind of security controls found on enterprise systems at businesses, notes Chaz Lever, a research engineer in Georgia Tech’s School of Electrical and Computer Engineering.

“The home network is beginning to look a lot like enterprise networks with a range of services that have to be protected,” Lever says. “But the average consumer is not equipped to do that. They don’t have an IT staff doing audits and securing devices. If these devices are not secure out of the box and there aren’t easy ways to secure them, they can open the home up to a new type of attack.”

To give consumers helpful advice, the researchers developed a framework for analyzing devices’ security components. In what is believed the first effort to objectively assess risks of IoT equipment, they examined devices themselves, along with how they communicate with cloud servers, applications running on the devices, and cloud-based endpoints.

“The more services running on the device, the higher the probability some of them are vulnerable to attack,” Antonakakis says. “Providing many services may be attractive from a marketing perspective, but if you have several services, the risk increases.”

In their study of IoT devices, researchers found wide variations in security depending on the manufacturer. In some cases, equipment made by small and lesser-known companies performed better than devices made by larger companies.

“There are some devices that do security really well, and other manufacturers should learn from those exemplary devices,” Alrawi says. “We saw the full spectrum of good and bad, and sometimes we were surprised at the results of our evaluations.”

Because they are designed to be installed by consumers, these IoT devices must be easy to use. But ease of use can be the enemy of security. For example, a service known as UPnP makes devices known to the network during installation so communications can be established. But a device announcing itself on the network can attract attackers, Lever notes. “It’s helpful for devices to communicate what they do, but that opens up vulnerabilities. The choice of protocols affects not only the device, but also the security of the network on which it is running.”

Internet-connected light bulbs are unlikely to have a long service life, but that’s not the case with expensive appliances such as internet-connected refrigerators. Antonakakis worries that these devices could become security risks without regular updates.

“Ideally, consumers shouldn’t have to be aware that their refrigerator needs updates that have to be downloaded to the device,” he says. “We want that to happen automatically and securely. Why should anyone have to know how to update their refrigerator?”

While the notion of hacking slow cookers might seem amusing, they do have heating elements which could cause fires if malicious actors turned up the temperature. Attacks can also affect more than a homeowner. In 2016, the Mirai botnet took advantage of unsecured internet-connected cameras, many of them baby monitors, to create a distributed denial-of-service attack that left much of the internet unavailable.

Beyond educating consumers, researchers hope to encourage better security by device manufacturers by tracking security trends over time.

“We hope to inspire both technical and policy progress,” says Antonakakis. “There is a need for establishing policy and standards. We want to raise the security level of all these devices. There is a lot more that could be done.”