2022-05-01: Summary of "Wearable Microphone Jamming"

In "Wearable Microphone Jamming" (CHI, 2020), Chen et al. explored and built a wearable device to disable surrounding microphones. Voice assistants like Alexa, Siri, and Google Assistant are widely used nowadays. Due to the increase in these voice assistants, a new security problem was also created. Since they can listen, record, and also save sensitive user data not only by using them. However, when they are used by surrounding users, is there any solution for these microphones' misuse?

Chen et al. have come up with the idea of a bracelet that can disable its surrounding microphone by emitting ultrasonic noise. This ultrasonic noise is emitted randomly in the form of white noise in the range of 24 kHz to 26 kHz.

This bracelet can disable or block hidden microphones in the wearer's surroundings, and it's based on the fact that when the microphone is exposed to ultrasonic noise then noise can leak into the audible range. Moreover, due to the natural hand movements of the wearer, this device can block blind spots thereby increasing jamming coverage. Additionally, this device is made in the form of a ring structure that can block in multiple directions. So, this device protects the wearer from multiple directions by blocking blind spots by taking advantage of natural movements.

Figure 1 in Chen et al. Jamming effect

Ultrasonic transducers can disable a microphone's capability to record human voice. Humans cannot hear this ultrasonic signal, but they leak into the microphone's audible range and create a jamming signal inside the microphone device hardware then the microphone records this jamming signal thereby making the microphone lose the actual recording of the human voice. In Figure 1, you can see that ultrasonic signals are able to leak into the microphone's audible range due to microphone hardware properties.  

At present, there are several microphone jammers available in the market which can disable microphones, but these microphones are unidirectional which means they can block only in one direction. However, we never know where the microphone is kept hidden, it can be kept hidden anywhere so if the jammer is unidirectional, we need to point the jammer exactly in the direction of the hidden microphone, which is not possible. Moreover, some jammers use multiple transducers to produce multiple ultrasonic signals, but the signals produced by transducers interact with each other and cancel out like how water waves interact with each other and cancel out when any object is dropped in it.

 So, when this occurs all the signals get canceled at specific locations around the jammer thereby creating a blind spot and canceling its jamming ability at that point. To destroy blind spots, we must create movement in the bracelet. In this case, it is created naturally since the user wears it, and when the user makes small movements, these blind spots are destroyed, and almost the entire surrounding's hidden microphones are blocked. Users need not worry about where the microphones are kept.

Figure 3 in Chen et al. Wearable Components

In Figure 3, you can see that the jammer is made in the form of a ring and since this jammer is made in the form of a ring, it's convenient for the user to wear it instead of holding it or carrying it in a bag. Additionally, this device won't disturb the user by producing sound because it produces signals that the user cannot listen to, and also it uses the user's natural hand gesture movements to increase its jamming coverage and to destroy blind spots.

Here is the demo video:

To find the efficiency of the jammer, word error rate (WER) is used. WER means how many word errors it is creating in the recording of the microphone. To disrupt the recording and disable microphones, we need to increase WER. 

It can be found that when the microphone is not at the effective blocking angle of microphone, the Word Error Rate (WER) is 26%, and when the jammer is worn by the user and kept static position then WER is 40-60% but when jammers are moved naturally during hand gestures, then WER is above 87%.

From Figure 6, we can see how WER varies based on the microphone angle. Three jammers are used for this planar jammer which has 9 transducers, an i4 jammer is available in the market which has 5 transducers attached to its side perpendicular to which two transducers attached on its top, and a bracelet-shaped jammer that has transducers attached to all around the ring. Both the i4 and planar jammer is able to produce good WER between 30 to 60 degrees, WER for i4 drops drastically falls after 90 degrees angle but the wearable bracelet-shaped jammer was able to produce higher WER at any angle when compared to the other two.

Figure 6 in Chen et al. WER at different microphone angles

Figure 5 in Chen et al. Relative power at various angles

Figure 8 in Chen et al. WER for different objects

From Figure 8, we can observe that WER is decreased when the microphone is hidden under a plastic box or cardboard or plastic box or fur screen. For all other remaining cases if the microphone is hidden under paper tissue or cloth or foam screen or paper sheet or when this device was switched off, then WER is increased which means this device is performing well. 

This work has demonstrated that (1) This wearable bracelet-shaped jammer is performing well when compared to static jamming in their jamming coverage areas, (2) This device is able to effectively jam even if the microphone is hidden under various material only except for a few materials like plastic box, cardboard, plastic bag, fur screen jamming, and (3) In a study of 12 users, the author found that users are able to believe that this device was able to give them privacy avoiding anyone to listen or record their sensitive information.

-- Mohan Krishna Sunkara (@mk344567)

Chen, Y., Li, H., Teng, S. Y., Nagels, S., Li, Z., Lopes, P., & Zheng, H. (2020, April). Wearable microphone jamming. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (pp. 1-12).

https://doi.org/10.1145/3313831.3376304