Channel Hardening of IRS-Aided Multi-Antenna Systems: How Should IRSs Scale?
Unlike active array antennas, intelligent reflecting surfaces (IRSs) are efficiently implemented at large dimensions. This allows for traceable realizations of large-scale IRS-aided MIMO systems in which not necessarily the array antennas, but the passive IRSs are large. It is widely believed that large IRS-aided MIMO settings maintain the fundamental features of massive MIMO systems, and hence they are the implementationally feasible technology for establishing the performance of large-scale MIMO settings. This work gives a rigorous proof to this belief. We show that using a large passive IRS, the end-to-end MIMO channel between the transmitter and the receiver always hardens, even if the IRS elements are strongly correlated. For the fading direct and reflection links between the transmitter and the receiver, our derivations demonstrate that as the number of IRS elements grows large, the capacity of end-to-end channel converges in distribution to a real-valued Gaussian random variable whose variance goes to zero. The order of this drop depends on how the physical dimensions of the IRS grow. We derive this order explicitly. Numerical experiments depict that the analytical asymptotic distribution almost perfectly matches the histogram of the capacity, even in practical scenarios. As a sample application of the results, we use the asymptotic characterization to study the dimensional trade-off between the transmitter and the IRS. The result is intuitive: For a given target performance, the larger the IRS is, the less transmit antennas are required to achieve the target. For an arbitrary ergodic and outage performance, we characterize this trade-off analytically. Our investigations demonstrate that using a practical IRS size, the target performance can be achieved with significantly small end-to-end MIMO dimensions.