Noise is everywhere, from people talking in offices to the never-ending traffic jams and airport announcements. Current noise canceling technology comes in the form of headphones and earbuds, which emit an anti-noise signal to counteract the external sounds. The time available for the headphones to compute this anti-noise is extremely short. This results in some noise getting through, which is why all these devices must cover the entire ear with noise-absorbing material. However, wearing such ear-blocking devices for long periods of time is not comfortable, and can even be harmful to health. This proposal aims to design a new noise-canceling architecture that does not block the ear canal. The vision is a behind-the-ear device that achieves noise cancellation as good as the best headphones or earbuds available today. The main idea behind this research involves combining wireless Internet-of-Things (IoT) networks with noise cancellation. Briefly, a microphone is placed in the environment that senses sounds and forwards them over wireless signals to an earpiece. Since wireless signals travel a million times faster than sound, the earpiece can receive the sound information in advance of the actual sound itself. Similar to lightning and thunder -- where the lightning arrives much before the thunder, hence allowing people to prepare for the loud rumble -- the proposed ear device gets the sound in advance and has much more time to produce a better anti-noise signal for human ears. Such technology has the potential to change the future of "earable" devices so that people can wear them continuously and enjoy much better control on what they hear. The implications on health are also crucial since noise pollution is becoming one of the major health concerns, especially in urban areas in developing regions.
The core research contributions emerge from re-visiting classical ideas in noise cancellation with a networking and IoT lens. The proposal intends to demonstrate that new architectures are possible that not only allow additional time for acoustic signal processing, but also allow for non-causal filtering, crucial to better noise cancellation. Finally, the proposal also shows how lookahead in the sound enables filter caching and pre-loading, necessary to cancel fast changing sounds (like human conversations). Building on these foundations, the proposal is also keen on extending the ideas to device-free noise cancellation -- that is, the idea that noise can be (partially) cancelled at the ear without the need of a wearable ear-device.