The idea sounds great. But when you think realistically, you will know there is always a tradeoff. If you look at the paper and ideas by ignoring all the marketing messages, you can see there are flaws and it cannot apply to MIMO solutions.
Basically, the idea is to place the 1 Rx antenna between two Tx antennas so that the signals from two Tx antennas are out of phases at the Rx antenna position. Yes, it works for the purpose that you want the Rx antenna receives no signal from Tx antennas. However, this also changes the antenna beam pattern which will reduce the coverage for the reception of the signal by half. The placement of 2 Tx antennas is the same as have a Uniform Linear Array (ULA) with 2 antennas with some inter-element distances. For example, if you place the antennas with lambda/2 away (where lambda is the wavelength), you will receive very weak signal at the broadside angle. In the other word, you trade the reception quality at broadside angle for the null-out effect at your Rx antenna.
Just look at the image here and you can choose the null wherever you like
Actually, you have two degrees of freedom here to choose any position/angles you want to have the weakest signals (null-out effect). 1) by changing the phase difference between two Tx antennas. For example, making them 180 degree out of phase, the null will happen at the middle between two antennas. In this case, you trade your reception along the direction perpendicular to your antenna array with the so-call single duplex. 2) by changing the inter-distance between two antennas. For example, you can make the distance with lambda*3/4 away to create the 180 degree out of phase. The drawback is the same. You trade the reception at certain area with this null-out at certain position.
People can argue, you can increase the inter-distance more, so that more grating lobes can be observed. Because there are multiple paths, eventually, the whole space can be covered. Yes, you are right. But how far away two antennas should be separated? It definitely depends on the environment. You can make it adaptive. So a dedicated engineer needs stand at the access point to change the antenna separations per request. And if you place the antennas so far away, is that really practical to have such systems?
So, they are just using one mode of the ULA to achieve the signal cancellation at the receiver point. But since they are twisting the antenna patterns, if you add more antennas trying to achieve 3x3 or 4x4, you are doomed. Because the degree of freedom has been used to cancel the signal at receivers, you have no more degree of freedoms for more spatial streams. So, yes, they can increase the capacity by two fold (in some area of coverage), but not beyond as in MIMO solution!
I believe there is an use case. But we should know the tradeoff and whether it is suitable for real deployment.