Arrays are used in a range of applications, from voice recognitions systems to sonars and medical ultrasound. A subset of array applications rely on battery powering – portable voice assistants being one such example. This puts pressure on product designers to minimize power consumption, and effectively restricts the kinds of performance one could desire.
Wireless charging of sensors therefore opens up a new range of conceivable array solutions. In some cases, it makes it possible to reimagine where and when arrays can be used and define entirely new types of applications. SINTEF MiNaLab is developing a system for wireless charging using MEMS micromirrors to steer a beam of light towards a photovoltaic cell. The mirrors have some specific and desirable properties that differentiate them from the state-of-the-art of micromirrors, such as high deflection angles, high accuracy, low power consumption and low cost in high volumes.
In a master thesis, we seek to utilize this new power source, combining a high number of sensors or sensors array to create new “super-arrays” – configurations that would otherwise need cabling or large batteries and therefore be unfeasible in real-life situations. For an illustration, see Figure 1.
In this specific example, the small round dots on the walls are miniature voice recognition array sensors (they are often larger in real life, but small arrays can be realized with very good microphones, such as those commercialized by sensiBel, A SINTEF spin-out company). By combining many array sensors, far more accuracy in separating the voices of persons A and B can be expected than what is possible using state-of-the art arrays like those found in Amazon Alexa or Apple HomePods. The disc in the top of the ceiling is the power source, which beams light towards each of the small arrays. Each small array has a photovoltaic cell that can receive and store power. The signals recorded at the arrays can be transferred to a stationary PC (not shown) via WiFi 6.0 - a low power wifi technology. This PC can run sophisticated and CPU-hungry algorithms as needed.
Another way to utilize wireless multi-array sensing is in room mapping.In this situation we are not interested in voice signals, but instead to create a map of the room: where is the furniture? Are the people in the room? Are they moving around and where are they headed? This information could be used for many different purposes, such as turning on or off light or heating, or in hifi/audio applications, where the best cinematic audio experience depends on knowing where the listener is situated, and “beaming” sound from speaker elements towards the listener.
To realize this application, imaging each of the small wall-mounted units containing an ultrasound transmitter and several ultrasound receivers. At regular intervals each transmitter emits a burst of sound, which is reflected from the objects in the room and returned to the receivers. There, traditional ultrasonic array processing is used to detect directions and positions of strong reflectors. In a next step, this information is synthesized from all the various units placed around the room to create a high-resolution map.
A third way to utilize the concept is to create sound showers. Imagine now that each little disc around the room is a tiny loudspeaker array. When the location of a person in the room is known, we can use directional audio (beamforming) to steer the signal towards a single person, from each different speaker. Each audio beam can in itself be relatively low energy, but when the beams from multiple direction sum up at a centre location, the signal will be heard only locally. Few if any other person can hear the directed message. In effect, we are creating private audio pockets.
The three approaches above could be combines into a super-combination: we can locate where people are, we can separate their voices from each other, and we can beam private messages to each of them.
This thesis can be attacked from the signal (array) processing angle or from a systems/hardware building angle, where emphasis is more on the processing side or on the hardware building side. To demonstrate performance of the end application, we don’t need to implement the actual wireless charging part. Instead we can ‘fake’ the experience by using cabled or battery powered solution. Completing a thesis therefore doesn’t depend on building or utilizing a full wireless charging system.
A group of students could work together on different aspects of the solution.
Literature:
- Wireless charging of devices (A SINTEF patent)
- Sound showers
- Loudspeaker arrays
- How Amazon Alexa works (no longer available)
