Energy Market Manipulation with High-Wattage IoT Botnets

Most concerns about vulnerable Internet of Things (IoT) devices have centered on adversaries targeting them to disrupt operations or to assemble large botnets for launching denial-of-service attacks or distributing malware and spam.

In a new spin, researchers at a Black Hat virtual event this week described how certain high-wattage Internet-connected devices such as smart air-conditioners and electric-vehicle (EV) chargers could be used to manipulate energy markets.

According to the researchers, from the Georgia Institute of Technology, their studies show that attackers can alter power demand on a grid if they are able to compromise and control a sufficient enough number of high-wattage IP-connected products. By simultaneously switching on or shutting down tens of thousands of EV chargers, for instance, or smart heaters, ACs, and ovens, attackers can change real-time demand for power enough to affect electricity prices.

"If we can somehow control the total power consumption of the power grid and change it slightly, we should be able to affect real-time systems for electricity market prices," said Tohid Shekari, a doctoral candidate at the Georgia Institute of Technology. If the attacks are carried out in a sophisticated enough manner, the manipulation would be almost unnoticeable, he said.

Shekari provided an example of an attacker with a botnet made up of 100,000 IoT devices, each consuming 3 kilowatts per hour (3,000 watts). If the attacker were to turn the bots on or off for about three hours a day for about eight days a month, it would be enough to trigger a barely noticeable but sufficient change in demand so as to affect prices.

Shekari said his research showed that such attacks can be carried out without the owners of the high-wattage devices noticing anything is amiss. As examples of stealth tactics, he pointed to early morning attacks on smart water heaters or EV chargers, or midday attacks on a smart fridge or garage door opener when no one is likely to be at home. If the attacks are smart enough, device owners are unlikely to notice the small changes in their electric bill that will likely result from such manipulation, Shekari said.

A market player could profit from such demand manipulation to the tune of millions of dollars per day. Or a nation-state player could deliberately manipulate power demand to cause substantial financial damage to targeted segments of the electricity market, Shekari said.

Feasible and Possible
According to Shekari, there are several reasons why such attacks are feasible and well within the capabilities of sophisticated adversaries. For one thing, data shows that millions of high-wattage devices are being connected to the Internet on a daily basis. Data shows there are more than 30 million smart thermostats installed in North America alone in 2020, he said. If an adversary has access to even a small fraction of the installed base of high-wattage IoT devices, they should have enough bots to manipulate market prices, he added.

Numerous underground markets readily offer IoT botnets for hire. It's conceivable that these markets will soon start offering botnets made up of high-wattage devices. Even if the prices for these botnets end up being a hundred times more expensive than regular botnets, there is still enough profit in energy market manipulation for adversaries, Shekari said. The researchers have not been able to test their model in a real-life situation. But simulations with real-world data from load-sensitive markets such as New York and California confirm that such attacks are possible, he said.

One potential measure against such manipulation is for energy industry stakeholders — including major manufacturers of high-wattage devices — to develop a database for monitoring the real-time behavior of devices so it is easier to spot when they are being manipulated. Limiting price sensitivity in real-time energy markets can also make it harder for an adversary to manipulate demand without being noticed, Shekari said.

Related Content: