Post by Romain Fleury, commissioned by David Abergel.
The original paper in Nature Physics can be read here.
Imagine you are playing the popular Nintendo game Mario Kart, and as you try to win the race one of the other players suddenly drives into the worst possible item box you can imagine: it covers the road in front of you with a very, very large number of banana peels, making it extremely unlikely for you to avoid these obstacles. If waves could have feelings, this is probably what they would think when a scientist tries to transmit them through a strongly localized disordered medium.
Yet, imagine now that you have the possibility to install some sort of magic boosters, or conveyor belts, that auto-pilot your kart seamlessly through these obstacles, while maintaining your precious velocity. This is certainly not possible in the game, but for our team of physicists and engineers, it made perfect sense to try this for waves in disordered media.
In our recent Nature Physics article, we have used acoustic boosters, or relays, to guide sound through a very nasty series of obstacles, and turned an Anderson-localized opaque medium into a perfectly transparent one by doping it with gain and loss. Interestingly, these acoustic boosters were made possible by recent advances in active noise control devices, similar to the ones you may use in your noise cancellation headphones during your next flight. Here is the story of how this idea came to life.
The project involved a team of researchers from three universities: the Swiss Federal Institute of Technology in Lausanne (EPFL Wave Engineering Lab — my lab — and the Signal Processing Lab 2, with Etienne Rivet and Hervé Lissek), the Technical University of Vienna (Stefan Rotter’s group) and the University of Crete (Kostas Makris’ group). When I first met Stefan Rotter at the META16 conference in Malaga, he described to me a new concept that Kostas was working on while being a postdoc in his group, about using amplification and absorption to guide light through disordered media. He also mentioned to me the difficulty he had to find opticians crazy enough to try and build such a challenging experiment. At that time, although my expertise let me think that it was certainly possible to observe such scattering states in acoustics instead of optics, I had no idea how to translate Stefan’s continuous optical approach into a discrete acoustic system.
It was only after starting at EPFL as assistant professor that I understood that the feedback control techniques built for active noise cancellation were flexible enough to meet our challenging needs. A few months after I arrived, I discovered the latest works of my office neighbor Hervé Lissek on active noise control and explained the idea to him. As his student, Etienne Rivet, had just defended his thesis, it was the perfect time for both of them to climb on board with Stefan, Kostas and I on our ambitious project.
It took quite a lot of effort to translate the theoretical concept, which worked with ideal continuous optical gain and loss distributions, into a real-world solution implementable with active acoustic feedback methods. Etienne and I spent many hours in my office trying to figure out a discrete version of the theory, which turned out to be different from the initial theory (continuous and discrete) proposed by Stefan’s group for optical waves. Figuring out the theory and validating it with simulations took us several months of intense work. Putting together the experiment: about a year.
The main experimental challenge we faced was that it is not trivial to obtain stable acoustic gain, even using active feedback methods. We had to resort to signal processing tricks, such as demodulation/modulation of the feedback loop signal, in order to control the gain bandwidth. As in many feedback techniques, one needs a good model of the system for the method to be efficient. After intense efforts, we finally got a working prototype and demonstrated 100% sound power transmission through an Anderson-localized acoustic system doped with gain and loss, which was an amazing result considering that the undoped medium was essentially as opaque as a wall.
It sometimes takes more than one area of expertise to tackle a challenging theoretical and technological problem, and our project is the perfect example of the power of team work. Through our theory–experiment collaboration and the digital audio control strategy we developed, we can effortlessly guide waves through complex environments — essentially by just a simple line of code in our FPGA (Field Gate Programmable Arrays) modules. Seeing our magic boosters work in practice is really amazing — implementing them also in the Mario Kart game may not be such a crazy idea after all.
Professor of Electrical Engineering
Laboratory of Wave Engineering
Swiss Federal Institute of Technology in Lausanne (EPFL)