Architecture and Hardware

Protecting Privacy from Wireless Spycams

Even items as simple as a clothes hook may contain a spycam (as this one does).
In response to the global problem of spycams, two research groups say they have developed effective wireless spycam detection protocols.

The features of microchips have shrunk to incredible levels, with tens of billions of transistors now capable of being shoehorned onto a single microprocessor. The flipside of this runaway success, however, is that image-sensing chips have shrunk so much that digital cameras can now be all but invisible—and that is fuelling some distressing privacy breaches.

For instance, in March it emerged that a gang that planted diminutive spy cameras in 42 hotels in 10 South Korean cities had livestreamed the private bedroom and bathroom activities of more than 1,600 hotel guests to the subscribers of a sleazy voyeur's porn website. The four men behind the privacy attack have been arrested, and their network shut down.

Seoul police said the suspects had managed to hide tiny wireless cameras in the most innocent-looking of products: digital TV receivers, power sockets, and hairdryer holders in the hotel rooms. With lenses only 1mm in diameter, they were extremely tough for guests to spot.

This kind of crime is particularly common in South Korea, where women have taken to the streets to protest the many thousands of cases per year in which they have been covertly filmed in public spaces, restrooms, and changing rooms. Marching under the slogan "my life is not your porn," victims—who are often left feeling suicidal or housebound in case they are recognized—are pressing police to prosecute many more offenders. One outcome of the protests, so far, is that police have begun regularly searching Seoul's 20,000 public restrooms for cameras.

An unexpected CES reveal

Spycams are a global problem.

In 2016, it emerged that seven people attending the Consumer Electronics Show in Las Vegas, NV, were covertly filmed by spycams hidden inside six smoke detectors in an Airbnb apartment. An Airbnb host is said to have confessed that he traded secret video of his guests with like-minded hosts.

In December 2018, the owner of a women's hostel in Chennai, India, was arrested after raising suspicions when he repeatedly visited the hostel to change spycam viewing angles. Two months earlier, a Florida couple even found a spycam in their cruise ship cabin pointed at their bed.

At the root of all this is the fact that tiny spy cameras that can be hidden in everyday objects are widely available online and can feed video via wireless or wired links, or save footage for later retrieval on memory cards. Some spycam-equipped household goods are even sold "ready made" online. For instance, Omejo Technology of Hong Kong sells pens, alarm clocks, fans and heaters with spycams built into them—ostensibly for acquiring "the most authentic evidence for a variety of illegal behavior." 

Omejo also sells spycams built into electric toothbrush bases, toilet paper holders, and even toilet brushes. Though the company warns buyers to "conform to the local laws when you use this product," the company had not responded at press time to a request for comment on where filming people in bathrooms is deemed legal.

Wanted: dependable spycam detectors

So what can people do to screen rooms for such spycams right now? Not much, at least reliably. Many websites offer what are claimed to be spycam detectors, either in the form of smartphone apps or handheld devices. The apps are said to seek out telltale reflections from a spycam lens, or use a phone's varied radio receivers to sense weak electromagnetic signals leaking from a working camera.

However, after evaluating a clutch of such apps, a team led by Xiayou Ji, a professor of security engineering at Zhejiang University in Hangzhou, eastern China, found them to be extremely unreliable.

The reason? "Both methods assume users are aware of the approximate location of cameras, but that is not always the case," Ji told the June 2018 ACM Asia Computer and Communications Security conference in Incheon, South Korea.

On top of that, the Zhejiang team analyzed handheld devices sold online that purported to detect hidden wireless cameras by sensing their 2.4GHz or 5GHz Wi-Fi signals. "But we found them unreliable as well," says Ji, thanks to the profusion of Wi-Fi signals in the environment. 

So to develop a reliable spycam detector, Ji and colleagues are focusing on what they regard as the most popular type: wireless spycams (since 80% of the top surveillance cameras on are Wi-Fi devices and the sector is growing strongly, at 21% per annum).

Since last year, the Zhejiang team have been focusing on perfecting an answer of their own: an Android smartphone-based system they call DeWiCam (a mnemonic for "Detecting Wireless Cameras").

To develop DeWiCam, the researchers have focused on precisely what is needed: first, the ability to detect a strong, local video stream in an 802.11 Wi-Fi format, and second, a way of proving that the room the user is in is indeed the space being filmed by that covert camera. 

AI to the rescue

To perform the first function, a supervised machine learning classifier in the DeWiCam app is trained to recognize wireless camera data flows using only the features of the Wi-Fi networking stack—in the physical and access control layers—that are relevant to certain key elements of video frame transmission. All other Internet transport data is filtered out.

So if video is being transmitted over the air in Wi-Fi format, this leaves the DeWiCam software with one or more candidate video channels. But are any of them being shot in the room the user is in? In other words, is there a spycam in the room? To find that out requires audience participation: the user has to walk around the room quickly, or wave their arms, so that any video signal changes some of the features in the data stream in lock-step with their body motion. 

"If a wireless camera is directly filming the user, the bitrate should show a significant rise when they begin moving, and a fall when the user stops," says Ji. Together, these tricks work: DeWiCam, the team claims, can detect that a covert camera is in a room with 99% accuracy within 2.7 seconds. This technique works even if the spy is encrypting the video stream.

In addition to the Zhejiang University team, others are also trying to defeat the spycam threat. In March, at the IEEE Pervasive Computing and Communications conference in Kyoto, Japan, Brent Lagesse and Kevin Wu of the University of Washington in Bothell, WA, revealed that they have also developed an effective wireless spycam detection protocol.

Flash dance

After sensing video streams with network-sniffing software running on a laptop, the Washington researchers use a different method to confirm whether the spycam is actually in the room of interest. Instead of DeWiCam's method of a user moving in the room to disturb the video signal, the Washington researchers use the camera flash LED on a smartphone to produce a sequence of flashes, and then seek corresponding signal variations in the video channels they are monitoring.

"We simultaneously measure the wireless network traffic that we can hear and then calculate how similar the network traffic patterns are to the randomly generated flashing patterns. When we identify a network stream that is similar to the flash pattern, we flag it as a spying webcam," says Lagesse, a professor of computing and software systems.

"Practically, this can be done all on a phone, all on a laptop, or with both a phone and a laptop," he says, and with speed and accuracy comparable to that of the DeWiCam solution.

Meanwhile, the Zhejiang team are building DeWiCam into a spycam-detecting add-on for a smartphone (which will work on both iOS and Android) that will get around a problem with their prototype: to analyse multiple Wi-Fi video streams, they need root access to the phone's hardware, and such jailbreaking of phones tends to void warranties and increase the risk of malware attacks.

"We are working with several well-known smartphone manufacturers now and hope to integrate the hidden camera detection function in smartphones with no rooting required, with a hardware product the size of a USB drive. We expect DeWiCam to be available soon," Ji says.

Both the Washington and Zhejiang teams were unfazed by learning about the others' seemingly competitive work. "We are glad to see others working in this area," says Ji. "Hidden cameras are now everywhere, and we need to be careful about defending against such threats to privacy."

Paul Marks is a technology journalist, writer, and editor based in London, U.K.

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More