Optimal Pilot Power Based Channel Estimation Improves the Throughput of Intelligent Reflective Surface Assisted Systems

Abstract

Intelligent reflecting surfaces (IRS) have emerged as a promising technology of managing the radio signal propagation by relying on a large number of low-cost passive reflecting elements. In this letter, the optimal pilot power allocation required for accurate channel estimation of IRS-assisted communication systems is investigated. In contrast to conventional channel estimators, where pilot signals are usually designed to be constant-enveloped, we reconsider the pilot design to improve the passive beamforming performance thus resulting in an improved achievable rate. At first sight the result of our analysis appears counter-intuitive, suggesting that at a given total power, more power should be allocated to estimate low-gain channels, since the channel phase impairments are more severe than those of highgain channels. Our simulation results show that when the number of IRS elements is 4, the rate improvement of our proposed channel estimation scheme over the conventional counterpart may be as high as 25%.

Reconfigurable Intelligent Surfaces for Localization: Position and Orientation Error Bounds

“…Note also that the previous analysis considers τ km and τ bm to be a-priori known to provide the fundamental attainable limits. In practice, sub-optimal phases can be obtained by substituting the channel parameters, e.g., the TOAs and bearing angles, in ( 43), (44), or (60) with their estimates rather than their true values [76], [77]. For instance, a localization algorithm can start with a random RIS phase design; then, obtain an estimate for the TOAs, update the phase design, and repeat these steps till convergence [78].…”

Low-Complexity Channel Estimation and Passive Beamforming for RIS-Assisted MIMO Systems Relying on Discrete Phase Shifts

“…To be more explicit, the conflict probability of arbitrary pairs of (θ q , θ q ) from a Q-entry training set generated by (18), namely, the probability that at least two of the elements in the training set are the same, is given by…”

“…denotes the number of training sets having Q distinct elements, while B M Q represents the number of possible training sets given Q. For example, if we consider a RIS with M = 50 reflecting elements, each with B = 2 possible phase shifts, the conflict probability of arbitrary pairs of (θ q , θ q ) from the training set generated by (18) is P c ≈ 4 × 10 −14 for Q = 10. It is clear that further increasing the number of reflecting elements M will lead to a diminishing conflict probability, which is difficult to calculate even by numerical methods.…”

“…Hence, the channels of the access point (AP)-RIS link and the RIS-user equipment (UE) link can no longer be estimated separately by conventional pilot-based approaches. To solve this problem, several novel methods have been proposed for estimating the cascaded AP-RIS-UE channels instead [16]- [18]. For example, in [16], the simple ON/OFF method was proposed, where the direct and reflected channels are estimated by switching on/off the RIS elements in sequence.…”

“…Furthermore, a discrete Fourier transformation (DFT) matrixbased solution was employed in [17] for optimizing the RIS reflection coefficients during the channel estimation phase, which is capable of minimizing the mean square error (MSE) of the channel estimates. More recently, an optimal pilot power allocation strategy was proposed in [18] for improving the throughput attained by RIS-assisted communication systems having imperfect CSI. However, the aforementioned methods require the same number of pilot symbols as that of the reflecting elements at the RIS, which becomes excessive for a massive number of reflecting elements.…”