The capacity of massive multi-input multi-output (MIMO) networks is shown to be theoretically unlimited as the number of antennas grows, provided that the spatial correlation inherent in the propagation channel is exploited. This proposal employs this observation using the assistance of novel “intelligent reflecting surfaces†(IRS) that can adaptively modify the spatial characteristics of the MIMO channel. The overall project goal is to improve spectrum efficiency and co-existence. To achieve this goal, the proposed device employs a metasurface with cells whose reflection properties can be programmed to take an impinging electromagnetic field and reshape the reflected field. For coexistence, a narrow beam of energy impinging on the IRS could be redirected towards a receiver that might otherwise be blocked from the transmission. Furthermore, the IRS can provide very complex reflection patterns to re-equalize the wireless channel links. The novelty of the proposal is the fact that the phase shifts are governed not by individual controls that would require numerous bulky electrical connections, but by electromagnetic waves launched onto thin waveguides on the reflecting surface. Given the boundary conditions of these waveguides, the controlling wave is designed as a linear superposition of periodic modes whose coefficients determine the phase shifts. A prototype of this device will be built and algorithms for its control and the control of the massive MIMO network will be developed.
There will be four research activities. In the first, a novel IRS architecture with electromagnetic wave control will be designed. In the second activity, reduced-dimension algorithms for single-cell control of the IRS will be developed, together with corresponding methods for estimating the channel state information. The third activity will focus on the use of machine learning algorithms for incorporating multiple such IRS into a multi-cell setting with many users, in order to optimize the resulting huge number of parameters required for control of the phase shifts as well as channel state information acquisition and control functions such as base station handover. The fourth research activity will focus on testing our novel IRS prototype in a point-to-point MIMO network to demonstrate its ability to improve the link performance by means of the reduced dimension control variables. The development of the proposed IRS and its control will achieve high spectral efficiency while allowing for coexistence of multiple different co-channel networks. At the same time, the proposed IRS will have lower cost, higher energy efficiency, and simpler synchronization properties than alternatives based on relays. Control of the IRS will involve a multi-layer approach for scheduling and resource allocation in order to address issues such as multi-user interference and fairness.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
