Tweaking the ‘disorder’ in your room can optimize your wireless link

Go to the profile of Philipp del Hougne
Jan 16, 2019

As you are reading this blog, you are most likely sitting amidst a crazy chaos of microwaves. Unlike light and sound, which we can see and hear, microwaves go largely unnoticed and yet they have made their way into our everyday life where they are now irreplaceable. It is difficult to picture our modern life without mobile phone access to internet and the telephone network – enabled by wireless microwave links. And with the advent of context-aware environments like smart homes, the demand for wireless connectivity will only increase.

You probably wonder what ‘disorder’ the title of this blog refers to. While your room may well be tidied up, the microwaves still perceive it as disordered due to the room’s irregular geometry – distorted, for instance, by furniture. In fact, the microwaves are trapped inside the room and bounce around, which gives rise to the crazy chaos mentioned at the outset. At first sight possibly surprising, it turned out that this chaos can in fact enhance a wireless link.

To transfer more information than possible with a single communication channel, the concept of simultaneously using multiple channels became popular since the turn of the millennium. For instance, using more than one antenna on the transmit and receive sides constitutes a multi-channel approach. The underlying hope is obvious: if I use 10 channels, hopefully I can transfer 10 times as much information. Yet, multi-channel communication is not that easy. The crucial question is, are the different channels distinguishable? In fact, only the number of independent channels counts. In the case of identical channels, channel cross-talk makes the multi-channel concept pointless. But the disordered nature of most environments helps to make different channels to some extent distinguishable. Countless schemes to reduce cross-talk in such channels have been proposed, ranging from hardware approaches (i.e. designing the antennas themselves) to software approaches (i.e. designing the transmitted signals).

But why accept the propagation environment as given? In our recent Nature Electronics paper, we propose to tweak the randomness to impose optimally distinguishable channels – without any software or hardware efforts neither on the transmit nor the receive side. As tuning mechanism we use a simple reconfigurable metasurface that was first developed by my co-authors Mathias Fink (Professor at ESPCI Paris) and Geoffroy Lerosey (Chief Scientist at Greenerwave) a couple of years ago. In simple terms, the metasurface is an array of elements whose interactions with the microwaves can be individually altered via a simple electronic mechanism. Placed on the walls of a room, they constitute reconfigurable ‘mirrors’ for the waves.

Some first tests in indoor environments had not yielded any meaningful results. That was not particularly encouraging, but of course it didn’t mean much since there are too many uncontrolled variables under such conditions. Thus, I decided to first test the idea in a large chaotic metallic cavity. Such a stable and shielded environment enabled a systematic study of relevant parameters, like the number of antennas in the wireless link. Identifying an appropriate figure of merit to evaluate the channel diversity also took some thought.

The experiments were set up to be fully automated, everything being controlled by a computer (except for not-so-rare occasions when the Arduinos stopped communicating), such that experiments could run over night, weekends and holidays. The latter was indeed necessary, since the experiments were rather lengthy. Measuring a single n×n channel matrix with our setup required n² measurements; a single experiment required about 300 such channel-matrix measurements; each experiment was repeated for 30 different realizations; all of this was repeated for different numbers of antennas, metasurface pixels, cavity quality factors etc. Statistically thorough studies require averaging over different realizations of disorder, and are thus never fast, but this was the longest series of experiments I have set up to date.

After months of experiments, the systematic study had shown that it is possible to achieve optimal channel diversity by tweaking the disorder – which outperforms the maximum naturally attainable diversity in a perfectly disordered system. Now it was time to transpose the experiment (back) to real-life. In practice, this meant that I simply moved everything from the metallic cavity to an unoccupied office room in our lab (or at least thought to be unoccupied until I received emails saying otherwise). And now, everything worked fine also under realistic indoor-environment conditions. To further visualize the improved channel diversity, I decided to consider the wireless transfer of an image. Thanks to the linearity of wave propagation, it was possible to emulate this based on the measured channel matrix and thus using our available lab equipment (a network analyzer) without having to set up actual WiFi infrastructure.

Our results will play an important role for the commercial activities of my co-authors’ company Greenerwave which seeks to revolutionize wireless communication with ‘smart walls’ (based on the metasurface ‘mirrors’). From a more fundamental point of view, our results call for the development of new random-matrix tools to further our understanding of the underlying Physics. But I think the concept of ensuring optimal channel diversity will also prove useful in surprisingly unrelated contexts outside wireless communication.

In the future, if our technology manages to establish itself in the communication market, you may be sitting amidst a carefully tailored chaos of microwaves inside your room, ensuring your wireless link’s properties are optimal.



Philipp del Hougne, Mathias Fink, Geoffroy Lerosey, "Optimally diverse communication channels in disordered environments with tuned randomness," Nat. Electron. 2, 36–41, (2019)

Go to the profile of Philipp del Hougne

Philipp del Hougne

Scientist, CNRS

No comments yet.