Time Reversal of Broadband Signals in a Strongly Fluctuating MIMO Channel: Stability and Resolution

Abstract

We analyze the time reversal of a multiple-input-multiple-output (MIMO) system in a space-frequency-selective multi-path fading channel described by the stochastic Schr\"odinger equation with a random potential in the strong-fluctuation regime. We prove that in a broadband limit the conditions for stable super-resolution are the {\em packing} condition that the spacing among the $N$ transmitters and $M$ receivers be more than the coherence length $\ell_c$ and the consecutive symbols in the data-streams are separated by more than the inverse of the bandwidth $B^{-1}$ and the {\em multiplexing} condition that the number of the degrees of freedom per unit time at the transmitters ($\sim NB$) be much larger than the number of the degrees of freedom ($\sim MC$) per unit time in the ensemble of intended messages. Here $C$ is the number of symbols per unit time in the data-streams intended for each receiver. When the two conditions are met, all receivers receive simultaneously streams of statistically stable, sharply focused signals intended for them, free of fading and interference. This indicates the rough multiplexing gain of $NB$ in channel capacity, with the maximal gain per unit {\em angular} cross section given by $BL^d \ell_c^{-d}$ where $L$ is the distance from the transmitters to the receivers. We show that under the ideal packing condition time reversal can result in a high signal-to-interference ratio and low probability of intercept, and hence is an effective means for achieving the information capacity of multi-path channels in the presence of multiple users (receivers).