Cascaded Channel Estimation for Distributed IRS Aided mmWave Massive MIMO Systems

Intelligent reflecting surfaces (IRSs) are envisioned as a key enabler for next generation wireless communications due to their capability to boost system performance without requiring additional bandwidth or transmit power. However, estimating the cascaded channels between the user equipment (UE) through the IRS to the base station (BS) is a major bottleneck in IRS-assisted systems. In this work, we present a novel method to estimate all the cascaded IRS channels in a multiple-IRS (called as distributed IRS in this paper) aided mmWave massive MIMO system. We exploit the inherent structure in mmWave channels to reformulate the channel estimation problem as one of direction of arrival (DoA) and departure (DoD) estimation in the cascaded channel. In turn, this allows us to use subspace based methods from array processing to develop a joint ESPRIT-MUSIC algorithm for estimating the DoA at the BS and the DoD from the UE. An attractive feature of the scheme is its low pilot overhead requirement: unlike existing methods, the number of pilot symbols does not scale with number of IRS elements or the number of antennas at the BS; it only depends on the number of IRSs deployed and the number of antennas at the UE. We compare our method against state-of-the-art methods, and numerically illustrate its superior performance and robustness to the number of IRSs in the system.