IRS Aided Communication Model for Compact MIMO Systems

Traditionally intelligent reflecting surface (IRS) has been viewed as an ideal passive phase shifter with no mutually coupled elements. For this assumption to hold, IRS inter element spacing needs to be at least Nyquist spacing, thereby increasing overall aperture size with number of IRS elements. In this paper, we consider reverse problem wherein the aperture size is fixed. Here, packing more elements in a given space to enhance system performance results in mutual coupling and correlated fading. We consider an IRS aided communication network for a point-to-point (P2P) MIMO system with closely spaced transmit, receive and IRS antenna arrays. We derive channel models for various propagation scenarios and validate the theoretical expressions with actual antenna simulations. We allow for an arbitrary array configuration made from a material with finite conductivity. We show that such a communication network can be expressed as an equivalent MIMO system whose overall channel matrix is a non-linear function of IRS loading. The capacity expression is also a non-convex function of loads attached to IRS. We propose a proximal distance based algorithm to optimize capacity and analyze the effect of coupling aware optimization. Our results indicate capacity benefits on orders of 2-2.5 times for tightly coupled arrays when well designed optimization is performed using more accurate communication model that accounts for mutual coupling as opposed to naive methods that ignore coupling.